1 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
6 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
7 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
8 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
9 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
10 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
11 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
12 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
13 * config/i386/i386.c (ix86_push_rounding): ...this new function.
14 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
16 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
17 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
18 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
19 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
20 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
21 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
22 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
23 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
24 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
25 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
27 * expr.c (emit_move_resolve_push): Treat the input and result
28 of PUSH_ROUNDING as a poly_int64.
29 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
30 (emit_push_insn): Likewise.
31 * lra-eliminations.c (mark_not_eliminable): Likewise.
32 * recog.c (push_operand): Likewise.
33 * reload1.c (elimination_effects): Likewise.
34 * rtlanal.c (nonzero_bits1): Likewise.
35 * calls.c (store_one_arg): Likewise. Require the padding to be
36 known at compile time.
38 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
39 Alan Hayward <alan.hayward@arm.com>
40 David Sherwood <david.sherwood@arm.com>
42 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
43 Use plus_constant instead of gen_rtx_PLUS.
45 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
46 Alan Hayward <alan.hayward@arm.com>
47 David Sherwood <david.sherwood@arm.com>
49 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
52 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
53 Alan Hayward <alan.hayward@arm.com>
54 David Sherwood <david.sherwood@arm.com>
56 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
57 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
58 via stack temporaries. Treat the mode size as polynomial too.
60 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
61 Alan Hayward <alan.hayward@arm.com>
62 David Sherwood <david.sherwood@arm.com>
64 * expr.c (expand_expr_real_2): When handling conversions involving
65 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
66 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
69 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
70 Alan Hayward <alan.hayward@arm.com>
71 David Sherwood <david.sherwood@arm.com>
73 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
75 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
76 Alan Hayward <alan.hayward@arm.com>
77 David Sherwood <david.sherwood@arm.com>
79 * combine.c (can_change_dest_mode): Handle polynomial
81 * expmed.c (store_bit_field_1): Likewise.
82 * expr.c (store_constructor): Likewise.
83 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
84 and polynomial REGMODE_NATURAL_SIZE.
85 (gen_lowpart_common): Likewise.
86 * reginfo.c (record_subregs_of_mode): Likewise.
87 * rtlanal.c (read_modify_subreg_p): Likewise.
89 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
90 Alan Hayward <alan.hayward@arm.com>
91 David Sherwood <david.sherwood@arm.com>
93 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
96 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
97 Alan Hayward <alan.hayward@arm.com>
98 David Sherwood <david.sherwood@arm.com>
100 * match.pd: Cope with polynomial numbers of vector elements.
102 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
103 Alan Hayward <alan.hayward@arm.com>
104 David Sherwood <david.sherwood@arm.com>
106 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
107 in a POINTER_PLUS_EXPR.
109 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
110 Alan Hayward <alan.hayward@arm.com>
111 David Sherwood <david.sherwood@arm.com>
113 * omp-simd-clone.c (simd_clone_subparts): New function.
114 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
115 (ipa_simd_modify_function_body): Likewise.
117 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
118 Alan Hayward <alan.hayward@arm.com>
119 David Sherwood <david.sherwood@arm.com>
121 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
122 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
123 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
124 (expand_vector_condition, vector_element): Likewise.
125 (subparts_gt): New function.
126 (get_compute_type): Use subparts_gt.
127 (count_type_subparts): Delete.
128 (expand_vector_operations_1): Use subparts_gt instead of
131 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
132 Alan Hayward <alan.hayward@arm.com>
133 David Sherwood <david.sherwood@arm.com>
135 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
136 (vect_compile_time_alias): ...this new function. Do the calculation
137 on poly_ints rather than trees.
138 (vect_prune_runtime_alias_test_list): Update call accordingly.
140 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
141 Alan Hayward <alan.hayward@arm.com>
142 David Sherwood <david.sherwood@arm.com>
144 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
146 (vect_schedule_slp_instance): Likewise.
148 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
149 Alan Hayward <alan.hayward@arm.com>
150 David Sherwood <david.sherwood@arm.com>
152 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
153 constant and extern definitions for variable-length vectors.
154 (vect_get_constant_vectors): Note that the number of units
155 is known to be constant.
157 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
158 Alan Hayward <alan.hayward@arm.com>
159 David Sherwood <david.sherwood@arm.com>
161 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
162 of units as polynomial. Choose between WIDE and NARROW based
165 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
166 Alan Hayward <alan.hayward@arm.com>
167 David Sherwood <david.sherwood@arm.com>
169 * tree-vect-stmts.c (simd_clone_subparts): New function.
170 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
172 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
173 Alan Hayward <alan.hayward@arm.com>
174 David Sherwood <david.sherwood@arm.com>
176 * tree-vect-stmts.c (vectorizable_call): Treat the number of
177 vectors as polynomial. Use build_index_vector for
180 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
181 Alan Hayward <alan.hayward@arm.com>
182 David Sherwood <david.sherwood@arm.com>
184 * tree-vect-stmts.c (get_load_store_type): Treat the number of
185 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
186 for variable-length vectors.
187 (vectorizable_mask_load_store): Treat the number of units as
188 polynomial, asserting that it is constant if the condition has
189 already been enforced.
190 (vectorizable_store, vectorizable_load): Likewise.
192 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
193 Alan Hayward <alan.hayward@arm.com>
194 David Sherwood <david.sherwood@arm.com>
196 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
197 of units as polynomial. Punt if we can't tell at compile time
198 which vector contains the final result.
200 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
201 Alan Hayward <alan.hayward@arm.com>
202 David Sherwood <david.sherwood@arm.com>
204 * tree-vect-loop.c (vectorizable_induction): Treat the number
205 of units as polynomial. Punt on SLP inductions. Use an integer
206 VEC_SERIES_EXPR for variable-length integer reductions. Use a
207 cast of such a series for variable-length floating-point
210 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
211 Alan Hayward <alan.hayward@arm.com>
212 David Sherwood <david.sherwood@arm.com>
214 * tree.h (build_index_vector): Declare.
215 * tree.c (build_index_vector): New function.
216 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
217 of units as polynomial, forcibly converting it to a constant if
218 vectorizable_reduction has already enforced the condition.
219 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
220 to create a {1,2,3,...} vector.
221 (vectorizable_reduction): Treat the number of units as polynomial.
222 Choose vectype_in based on the largest scalar element size rather
223 than the smallest number of units. Enforce the restrictions
226 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
227 Alan Hayward <alan.hayward@arm.com>
228 David Sherwood <david.sherwood@arm.com>
230 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
231 number of units as polynomial.
233 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
234 Alan Hayward <alan.hayward@arm.com>
235 David Sherwood <david.sherwood@arm.com>
237 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
238 * target.def (autovectorize_vector_sizes): Return the vector sizes
239 by pointer, using vector_sizes rather than a bitmask.
240 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
241 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
242 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
244 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
245 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
246 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
247 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
248 * omp-general.c (omp_max_vf): Likewise.
249 * omp-low.c (omp_clause_aligned_alignment): Likewise.
250 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
251 * tree-vect-loop.c (vect_analyze_loop): Likewise.
252 * tree-vect-slp.c (vect_slp_bb): Likewise.
253 * doc/tm.texi: Regenerate.
254 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
256 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
257 the vector size as a poly_uint64 rather than an unsigned int.
258 (current_vector_size): Change from an unsigned int to a poly_uint64.
259 (get_vectype_for_scalar_type): Update accordingly.
260 * tree.h (build_truth_vector_type): Take the size and number of
261 units as a poly_uint64 rather than an unsigned int.
262 (build_vector_type): Add a temporary overload that takes
263 the number of units as a poly_uint64 rather than an unsigned int.
264 * tree.c (make_vector_type): Likewise.
265 (build_truth_vector_type): Take the number of units as a poly_uint64
266 rather than an unsigned int.
268 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
269 Alan Hayward <alan.hayward@arm.com>
270 David Sherwood <david.sherwood@arm.com>
272 * target.def (get_mask_mode): Take the number of units and length
273 as poly_uint64s rather than unsigned ints.
274 * targhooks.h (default_get_mask_mode): Update accordingly.
275 * targhooks.c (default_get_mask_mode): Likewise.
276 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
277 * doc/tm.texi: Regenerate.
279 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
280 Alan Hayward <alan.hayward@arm.com>
281 David Sherwood <david.sherwood@arm.com>
283 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
284 * omp-general.c (omp_max_vf): Likewise.
285 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
286 (expand_omp_simd): Handle polynomial safelen.
287 * omp-low.c (omplow_simd_context): Add a default constructor.
288 (omplow_simd_context::max_vf): Change from int to poly_uint64.
289 (lower_rec_simd_input_clauses): Update accordingly.
290 (lower_rec_input_clauses): Likewise.
292 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
293 Alan Hayward <alan.hayward@arm.com>
294 David Sherwood <david.sherwood@arm.com>
296 * tree-vectorizer.h (vect_nunits_for_cost): New function.
297 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
298 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
299 (vect_analyze_slp_cost): Likewise.
300 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
301 (vect_model_load_cost): Likewise.
303 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
304 Alan Hayward <alan.hayward@arm.com>
305 David Sherwood <david.sherwood@arm.com>
307 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
308 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
309 from an unsigned int * to a poly_uint64_pod *.
310 (calculate_unrolling_factor): New function.
311 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
313 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
314 Alan Hayward <alan.hayward@arm.com>
315 David Sherwood <david.sherwood@arm.com>
317 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
318 from an unsigned int to a poly_uint64.
319 (_loop_vec_info::slp_unrolling_factor): Likewise.
320 (_loop_vec_info::vectorization_factor): Change from an int
322 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
323 (vect_get_num_vectors): New function.
324 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
325 (vect_get_num_copies): Use vect_get_num_vectors.
326 (vect_analyze_data_ref_dependences): Change max_vf from an int *
327 to an unsigned int *.
328 (vect_analyze_data_refs): Change min_vf from an int * to a
330 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
331 than an unsigned HOST_WIDE_INT.
332 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
333 (vect_analyze_data_ref_dependence): Change max_vf from an int *
334 to an unsigned int *.
335 (vect_analyze_data_ref_dependences): Likewise.
336 (vect_compute_data_ref_alignment): Handle polynomial vf.
337 (vect_enhance_data_refs_alignment): Likewise.
338 (vect_prune_runtime_alias_test_list): Likewise.
339 (vect_shift_permute_load_chain): Likewise.
340 (vect_supportable_dr_alignment): Likewise.
341 (dependence_distance_ge_vf): Take the vectorization factor as a
342 poly_uint64 rather than an unsigned HOST_WIDE_INT.
343 (vect_analyze_data_refs): Change min_vf from an int * to a
345 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
346 vfm1 as a poly_uint64 rather than an int. Make the same change
347 for the returned bound_scalar.
348 (vect_gen_vector_loop_niters): Handle polynomial vf.
349 (vect_do_peeling): Likewise. Update call to
350 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
351 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
353 * tree-vect-loop.c (vect_determine_vectorization_factor)
354 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
355 (vect_get_known_peeling_cost): Likewise.
356 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
357 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
358 (vect_transform_loop): Likewise. Use the lowest possible VF when
359 updating the upper bounds of the loop.
360 (vect_min_worthwhile_factor): Make static. Return an unsigned int
362 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
363 polynomial unroll factors.
364 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
365 (vect_make_slp_decision): Likewise.
366 (vect_supported_load_permutation_p): Likewise, and polynomial
368 (vect_analyze_slp_cost): Handle polynomial vf.
369 (vect_slp_analyze_node_operations): Likewise.
370 (vect_slp_analyze_bb_1): Likewise.
371 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
372 than an unsigned HOST_WIDE_INT.
373 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
374 (vectorizable_load): Handle polynomial vf.
375 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
377 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
379 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
380 Alan Hayward <alan.hayward@arm.com>
381 David Sherwood <david.sherwood@arm.com>
383 * match.pd: Handle bit operations involving three constants
384 and try to fold one pair.
386 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
388 * tree-vect-loop-manip.c: Include gimple-fold.h.
389 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
390 niters_maybe_zero parameters. Handle other cases besides a step of 1.
391 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
392 Add a path that uses a step of VF instead of 1, but disable it
394 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
395 and niters_no_overflow parameters. Update calls to
396 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
397 Create a new SSA name if the latter choses to use a ste other
398 than zero, and return it via niters_vector_mult_vf_var.
399 * tree-vect-loop.c (vect_transform_loop): Update calls to
400 vect_do_peeling, vect_gen_vector_loop_niters and
401 slpeel_make_loop_iterate_ntimes.
402 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
403 (vect_gen_vector_loop_niters): Update declarations after above changes.
405 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
407 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
408 128-bit round to integer instructions.
409 (ceil<mode>2): Likewise.
410 (btrunc<mode>2): Likewise.
411 (round<mode>2): Likewise.
413 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
415 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
416 unaligned VSX load/store on P8/P9.
417 (expand_block_clear): Allow the use of unaligned VSX
420 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
422 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
424 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
425 swap associated with both a load and a store.
427 2018-01-02 Andrew Waterman <andrew@sifive.com>
429 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
430 * config/riscv/riscv.md (clear_cache): Use it.
432 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
434 * web.c: Remove out-of-date comment.
436 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
438 * expr.c (fixup_args_size_notes): Check that any existing
439 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
440 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
441 (emit_single_push_insn): ...here.
443 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
445 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
446 (const_vector_encoded_nelts): New function.
447 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
448 (const_vector_int_elt, const_vector_elt): Declare.
449 * emit-rtl.c (const_vector_int_elt_1): New function.
450 (const_vector_elt): Likewise.
451 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
454 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
456 * expr.c: Include rtx-vector-builder.h.
457 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
458 directly on the tree encoding.
459 (const_vector_from_tree): Likewise.
460 * optabs.c: Include rtx-vector-builder.h.
461 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
462 sequence of "u" values.
463 * vec-perm-indices.c: Include rtx-vector-builder.h.
464 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
465 directly on the vec_perm_indices encoding.
467 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
469 * doc/rtl.texi (const_vector): Describe new encoding scheme.
470 * Makefile.in (OBJS): Add rtx-vector-builder.o.
471 * rtx-vector-builder.h: New file.
472 * rtx-vector-builder.c: Likewise.
473 * rtl.h (rtx_def::u2): Add a const_vector field.
474 (CONST_VECTOR_NPATTERNS): New macro.
475 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
476 (CONST_VECTOR_DUPLICATE_P): Likewise.
477 (CONST_VECTOR_STEPPED_P): Likewise.
478 (CONST_VECTOR_ENCODED_ELT): Likewise.
479 (const_vec_duplicate_p): Check for a duplicated vector encoding.
480 (unwrap_const_vec_duplicate): Likewise.
481 (const_vec_series_p): Check for a non-duplicated vector encoding.
482 Say that the function only returns true for integer vectors.
483 * emit-rtl.c: Include rtx-vector-builder.h.
484 (gen_const_vec_duplicate_1): Delete.
485 (gen_const_vector): Call gen_const_vec_duplicate instead of
486 gen_const_vec_duplicate_1.
487 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
488 (gen_const_vec_duplicate): Use rtx_vector_builder.
489 (gen_const_vec_series): Likewise.
490 (gen_rtx_CONST_VECTOR): Likewise.
491 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
492 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
493 Build a new vector rather than modifying a CONST_VECTOR in-place.
494 (handle_special_swappables): Update call accordingly.
495 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
496 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
497 Build a new vector rather than modifying a CONST_VECTOR in-place.
498 (handle_special_swappables): Update call accordingly.
500 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
502 * simplify-rtx.c (simplify_const_binary_operation): Use
503 CONST_VECTOR_ELT instead of XVECEXP.
505 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
507 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
508 the selector elements to be different from the data elements
509 if the selector is a VECTOR_CST.
510 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
511 ssizetype for the selector.
513 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
515 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
516 before testing each element individually.
517 * tree-vect-generic.c (lower_vec_perm): Likewise.
519 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
521 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
522 * selftest-run-tests.c (selftest::run_tests): Call it.
523 * vector-builder.h (vector_builder::operator ==): New function.
524 (vector_builder::operator !=): Likewise.
525 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
526 (vec_perm_indices::all_from_input_p): New function.
527 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
528 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
529 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
530 instead of reading the VECTOR_CST directly. Detect whether both
531 vector inputs are the same before constructing the vec_perm_indices,
532 and update the number of inputs argument accordingly. Use the
533 utility functions added above. Only construct sel2 if we need to.
535 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
537 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
538 the broadcast of the low byte.
539 (expand_mult_highpart): Use an explicit encoding for the permutes.
540 * optabs-query.c (can_mult_highpart_p): Likewise.
541 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
542 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
543 (vectorizable_bswap): Likewise.
544 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
545 explicit encoding for the power-of-2 permutes.
546 (vect_permute_store_chain): Likewise.
547 (vect_grouped_load_supported): Likewise.
548 (vect_permute_load_chain): Likewise.
550 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
552 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
553 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
554 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
555 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
556 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
557 (vect_gen_perm_mask_any): Likewise.
559 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
561 * int-vector-builder.h: New file.
562 * vec-perm-indices.h: Include int-vector-builder.h.
563 (vec_perm_indices): Redefine as an int_vector_builder.
564 (auto_vec_perm_indices): Delete.
565 (vec_perm_builder): Redefine as a stand-alone class.
566 (vec_perm_indices::vec_perm_indices): New function.
567 (vec_perm_indices::clamp): Likewise.
568 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
569 (vec_perm_indices::new_vector): New function.
570 (vec_perm_indices::new_expanded_vector): Update for new
571 vec_perm_indices class.
572 (vec_perm_indices::rotate_inputs): New function.
573 (vec_perm_indices::all_in_range_p): Operate directly on the
574 encoded form, without computing elided elements.
575 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
576 encoding. Update for new vec_perm_indices class.
577 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
578 the given vec_perm_builder.
579 (expand_vec_perm_var): Update vec_perm_builder constructor.
580 (expand_mult_highpart): Use vec_perm_builder instead of
581 auto_vec_perm_indices.
582 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
583 vec_perm_indices instead of auto_vec_perm_indices. Use a single
584 or double series encoding as appropriate.
585 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
586 vec_perm_indices instead of auto_vec_perm_indices.
587 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
588 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
589 (vect_permute_store_chain): Likewise.
590 (vect_grouped_load_supported): Likewise.
591 (vect_permute_load_chain): Likewise.
592 (vect_shift_permute_load_chain): Likewise.
593 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
594 (vect_transform_slp_perm_load): Likewise.
595 (vect_schedule_slp_instance): Likewise.
596 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
597 (vectorizable_mask_load_store): Likewise.
598 (vectorizable_bswap): Likewise.
599 (vectorizable_store): Likewise.
600 (vectorizable_load): Likewise.
601 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
602 vec_perm_indices instead of auto_vec_perm_indices. Use
603 tree_to_vec_perm_builder to read the vector from a tree.
604 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
605 vec_perm_builder instead of a vec_perm_indices.
606 (have_whole_vector_shift): Use vec_perm_builder and
607 vec_perm_indices instead of auto_vec_perm_indices. Leave the
608 truncation to calc_vec_perm_mask_for_shift.
609 (vect_create_epilog_for_reduction): Likewise.
610 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
611 from auto_vec_perm_indices to vec_perm_indices.
612 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
613 instead of changing individual elements.
614 (aarch64_vectorize_vec_perm_const): Use new_vector to install
615 the vector in d.perm.
616 * config/arm/arm.c (expand_vec_perm_d::perm): Change
617 from auto_vec_perm_indices to vec_perm_indices.
618 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
619 instead of changing individual elements.
620 (arm_vectorize_vec_perm_const): Use new_vector to install
621 the vector in d.perm.
622 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
623 Update vec_perm_builder constructor.
624 (rs6000_expand_interleave): Likewise.
625 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
626 (rs6000_expand_interleave): Likewise.
628 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
630 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
631 to qimode could truncate the indices.
632 * optabs.c (expand_vec_perm_var): Likewise.
634 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
636 * Makefile.in (OBJS): Add vec-perm-indices.o.
637 * vec-perm-indices.h: New file.
638 * vec-perm-indices.c: Likewise.
639 * target.h (vec_perm_indices): Replace with a forward class
641 (auto_vec_perm_indices): Move to vec-perm-indices.h.
642 * optabs.h: Include vec-perm-indices.h.
643 (expand_vec_perm): Delete.
644 (selector_fits_mode_p, expand_vec_perm_var): Declare.
645 (expand_vec_perm_const): Declare.
646 * target.def (vec_perm_const_ok): Replace with...
647 (vec_perm_const): ...this new hook.
648 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
649 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
650 * doc/tm.texi: Regenerate.
651 * optabs.def (vec_perm_const): Delete.
652 * doc/md.texi (vec_perm_const): Likewise.
653 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
654 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
655 expand_vec_perm for constant permutation vectors. Assert that
656 the mode of variable permutation vectors is the integer equivalent
657 of the mode that is being permuted.
658 * optabs-query.h (selector_fits_mode_p): Declare.
659 * optabs-query.c: Include vec-perm-indices.h.
660 (selector_fits_mode_p): New function.
661 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
662 is defined, instead of checking whether the vec_perm_const_optab
663 exists. Use targetm.vectorize.vec_perm_const instead of
664 targetm.vectorize.vec_perm_const_ok. Check whether the indices
665 fit in the vector mode before using a variable permute.
666 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
667 vec_perm_indices instead of an rtx.
668 (expand_vec_perm): Replace with...
669 (expand_vec_perm_const): ...this new function. Take the selector
670 as a vec_perm_indices rather than an rtx. Also take the mode of
671 the selector. Update call to shift_amt_for_vec_perm_mask.
672 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
673 Use vec_perm_indices::new_expanded_vector to expand the original
674 selector into bytes. Check whether the indices fit in the vector
675 mode before using a variable permute.
676 (expand_vec_perm_var): Make global.
677 (expand_mult_highpart): Use expand_vec_perm_const.
678 * fold-const.c: Includes vec-perm-indices.h.
679 * tree-ssa-forwprop.c: Likewise.
680 * tree-vect-data-refs.c: Likewise.
681 * tree-vect-generic.c: Likewise.
682 * tree-vect-loop.c: Likewise.
683 * tree-vect-slp.c: Likewise.
684 * tree-vect-stmts.c: Likewise.
685 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
687 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
688 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
689 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
690 (aarch64_vectorize_vec_perm_const): ...this new function.
691 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
692 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
693 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
694 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
695 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
696 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
697 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
699 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
700 check for NEON modes.
701 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
702 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
703 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
704 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
706 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
707 the old VEC_PERM_CONST conditions.
708 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
709 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
710 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
711 (ia64_vectorize_vec_perm_const_ok): Merge into...
712 (ia64_vectorize_vec_perm_const): ...this new function.
713 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
714 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
715 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
716 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
717 * config/mips/mips.c (mips_expand_vec_perm_const)
718 (mips_vectorize_vec_perm_const_ok): Merge into...
719 (mips_vectorize_vec_perm_const): ...this new function.
720 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
721 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
722 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
723 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
724 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
725 (rs6000_expand_vec_perm_const): Delete.
726 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
728 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
729 (altivec_expand_vec_perm_const_le): Take each operand individually.
730 Operate on constant selectors rather than rtxes.
731 (altivec_expand_vec_perm_const): Likewise. Update call to
732 altivec_expand_vec_perm_const_le.
733 (rs6000_expand_vec_perm_const): Delete.
734 (rs6000_vectorize_vec_perm_const_ok): Delete.
735 (rs6000_vectorize_vec_perm_const): New function.
736 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
737 an element count and rtx array.
738 (rs6000_expand_extract_even): Update call accordingly.
739 (rs6000_expand_interleave): Likewise.
740 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
741 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
742 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
743 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
744 (rs6000_expand_vec_perm_const): Delete.
745 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
746 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
747 (altivec_expand_vec_perm_const_le): Take each operand individually.
748 Operate on constant selectors rather than rtxes.
749 (altivec_expand_vec_perm_const): Likewise. Update call to
750 altivec_expand_vec_perm_const_le.
751 (rs6000_expand_vec_perm_const): Delete.
752 (rs6000_vectorize_vec_perm_const_ok): Delete.
753 (rs6000_vectorize_vec_perm_const): New function. Remove stray
754 reference to the SPE evmerge intructions.
755 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
756 an element count and rtx array.
757 (rs6000_expand_extract_even): Update call accordingly.
758 (rs6000_expand_interleave): Likewise.
759 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
760 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
762 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
764 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
766 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
767 vector mode and that that mode matches the mode of the data
769 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
770 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
771 directly using expand_vec_perm_1 when forcing selectors into
773 (expand_vec_perm_var): New function, split out from expand_vec_perm.
775 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
777 * optabs-query.h (can_vec_perm_p): Delete.
778 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
779 * optabs-query.c (can_vec_perm_p): Split into...
780 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
781 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
782 particular selector is valid.
783 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
784 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
785 (vect_grouped_load_supported): Likewise.
786 (vect_shift_permute_load_chain): Likewise.
787 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
788 (vect_transform_slp_perm_load): Likewise.
789 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
790 (vectorizable_bswap): Likewise.
791 (vect_gen_perm_mask_checked): Likewise.
792 * fold-const.c (fold_ternary_loc): Likewise. Don't take
793 implementations of variable permutation vectors into account
794 when deciding which selector to use.
795 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
796 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
797 with a false third argument.
798 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
799 to test whether the constant selector is valid and can_vec_perm_var_p
800 to test whether a variable selector is valid.
802 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
804 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
805 * optabs-query.c (can_vec_perm_p): Likewise.
806 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
807 instead of vec_perm_indices.
808 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
809 (vect_gen_perm_mask_checked): Likewise,
810 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
811 (vect_gen_perm_mask_checked): Likewise,
813 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
815 * optabs-query.h (qimode_for_vec_perm): Declare.
816 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
817 (qimode_for_vec_perm): ...this new function.
818 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
820 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
822 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
823 does not have a conditional at the top.
825 2018-01-02 Richard Biener <rguenther@suse.de>
827 * ipa-inline.c (big_speedup_p): Fix expression.
829 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
832 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
835 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
839 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
840 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
841 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
842 cond_taken_branch_cost 3->4.
844 2018-01-01 Jakub Jelinek <jakub@redhat.com>
846 PR tree-optimization/83581
847 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
848 TODO_cleanup_cfg if any changes have been made.
851 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
852 convert_modes if target mode has the right side, but different mode
856 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
857 last argument when extracting from CONCAT. If either from_real or
858 from_imag is NULL, use expansion through memory. If result is not
859 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
860 the parts directly to inner mode, if even that fails, use expansion
864 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
865 check for bswap in mode rather than HImode and use that in expand_unop
868 2017-12-31 Jakub Jelinek <jakub@redhat.com>
871 * config/i386/i386.c (ix86_attribute_table): Remove excess
872 initializer for "shared" attribute.
874 2017-12-30 Tom de Vries <tom@codesourcery.com>
877 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
879 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
880 offload_funcs entries that no longer have a corresponding cgraph_node.
881 Mark the remaining ones as DECL_PRESERVE_P.
882 (output_lto): Call prune_offload_funcs.
884 2017-12-30 Jakub Jelinek <jakub@redhat.com>
886 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
887 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
888 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
889 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
890 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
891 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
892 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
893 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
894 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
895 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
896 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
897 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
898 vaesenc_<mode>, vpclmulqdq_<mode>,
899 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
901 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
903 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
904 _Float<N> and _Float<N>X variants for rounding built-in
906 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
907 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
908 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
909 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
910 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
911 * builtins.c (mathfn_built_in_2): Likewise.
912 * internal-fn.def (CEIL): Likewise.
914 (NEARBYINT): Likewise.
918 * convert.c (convert_to_integer_1): Likewise.
919 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
920 (integer_valued_real_call_p): Likewise.
921 * fold-const-call.c (fold_const_call_ss): Likewise.
922 * gencfn-macros.c (print_case_cfn): Change CFN and operator
923 printers to take a const char * suffix instead of a bool.
924 (print_define_operator_list): Likewise.
925 (fltall_suffixes): New list of suffixes, that include the
926 traditional suffixes as well as all of the _Float<N> and
928 (main): For _Float<N> and _Float<N>X functions, emit both
929 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
930 has the _Float<N> and _Float<N>X case names or operators. The
931 <name>_ALL variant has both the traditional and the
932 _Float<N>/_Float<N>X case names or operators.
933 * match.pd (COPYSIGN optimizations): Provide optimizations for
934 _Float<N> and _Float<N>X types where possible.
935 (MIN/MAX optimizations): Likewise.
936 (sqrt optimizations): Likewise.
937 (rounding optimizations): Likewise.
939 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
941 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
942 gen_int_mode rather than GEN_INT.
943 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
945 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
947 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
948 instead of CONSTANT_P.
949 (gen_vec_series): Likewise.
950 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
952 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
954 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
957 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
959 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
960 (aarch64_ld1x2<VDC:mode>): Likewise.
961 (aarch64_simd_ld1<mode>_x2): Likewise.
962 (aarch64_simd_ld1<mode>_x2): Likewise.
963 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
964 (vld1_s8_x2): Likewise.
965 (vld1_u16_x2): Likewise.
966 (vld1_s16_x2): Likewise.
967 (vld1_u32_x2): Likewise.
968 (vld1_s32_x2): Likewise.
969 (vld1_u64_x2): Likewise.
970 (vld1_s64_x2): Likewise.
971 (vld1_f16_x2): Likewise.
972 (vld1_f32_x2): Likewise.
973 (vld1_f64_x2): Likewise.
974 (vld1_p8_x2): Likewise.
975 (vld1_p16_x2): Likewise.
976 (vld1_p64_x2): Likewise.
977 (vld1q_u8_x2): Likewise.
978 (vld1q_s8_x2): Likewise.
979 (vld1q_u16_x2): Likewise.
980 (vld1q_s16_x2): Likewise.
981 (vld1q_u32_x2): Likewise.
982 (vld1q_s32_x2): Likewise.
983 (vld1q_u64_x2): Likewise.
984 (vld1q_s64_x2): Likewise.
985 (vld1q_f16_x2): Likewise.
986 (vld1q_f32_x2): Likewise.
987 (vld1q_f64_x2): Likewise.
988 (vld1q_p8_x2): Likewise.
989 (vld1q_p16_x2): Likewise.
990 (vld1q_p64_x2): Likewise.
992 2017-12-27 Martin Liska <mliska@suse.cz>
994 PR tree-optimization/83552
995 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
996 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
998 2017-12-27 Tom de Vries <tom@codesourcery.com>
1000 * config/nvptx/nvptx.c (nvptx_option_override): Disable
1001 -gstatement-frontiers.
1003 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
1005 PR rtl-optimization/83513
1006 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
1007 before priority comparison.
1009 2017-12-25 Jakub Jelinek <jakub@redhat.com>
1012 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
1013 ix86_isa_flags2 to ix86_isa_flags.
1014 * config/i386/i386-c.c (ix86_target_macros_internal): Test
1015 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
1016 isa_flags rather than isa_flags2.
1017 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
1018 and -mavx512bitalg from isa2_opts to isa_opts.
1019 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
1020 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
1021 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
1023 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
1024 OPTION_MASK_ISA_AVX512F ored with another option similarly to
1025 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
1026 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
1027 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
1028 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
1029 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
1030 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
1031 builtins from bdesc_args2 to bdesc_args section.
1032 (__builtin_ia32_compressstoreuqi512_mask,
1033 __builtin_ia32_compressstoreuhi512_mask,
1034 __builtin_ia32_compressstoreuqi256_mask,
1035 __builtin_ia32_expandloadqi512_mask,
1036 __builtin_ia32_expandloadqi512_maskz,
1037 __builtin_ia32_expandloadhi512_mask,
1038 __builtin_ia32_expandloadhi512_maskz,
1039 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
1040 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
1041 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
1042 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
1043 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
1044 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
1045 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
1046 __builtin_ia32_vpshldv_v32hi_maskz,
1047 __builtin_ia32_vpopcountb_v64qi_mask,
1048 __builtin_ia32_vpopcountw_v32hi_mask,
1049 __builtin_ia32_vpshufbitqmb512_mask,
1050 __builtin_ia32_vpshufbitqmb256_mask): Add
1051 " | OPTION_MASK_ISA_AVX512BW".
1052 (__builtin_ia32_expandloadqi256_mask,
1053 __builtin_ia32_expandloadqi256_maskz,
1054 __builtin_ia32_vpopcountb_v32qi_mask): Add
1055 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
1056 (__builtin_ia32_expandloadhi256_mask,
1057 __builtin_ia32_expandloadhi256_maskz,
1058 __builtin_ia32_expandloadqi128_mask,
1059 __builtin_ia32_expandloadqi128_maskz,
1060 __builtin_ia32_expandloadhi128_mask,
1061 __builtin_ia32_expandloadhi128_maskz,
1062 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
1063 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
1064 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
1065 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
1066 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
1067 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
1068 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
1069 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
1070 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
1071 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
1072 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
1073 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
1074 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
1075 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
1076 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
1077 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
1078 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
1079 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
1080 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
1081 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
1082 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
1083 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
1084 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
1085 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
1086 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
1087 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
1088 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
1089 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
1090 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
1091 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
1092 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
1093 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
1094 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
1095 __builtin_ia32_vpopcountw_v8hi_mask): Add
1096 " | OPTION_MASK_ISA_AVX512VL".
1097 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1098 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
1099 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
1100 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
1101 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
1102 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
1103 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
1104 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
1105 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
1106 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
1107 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
1108 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
1109 _mm_mask_bitshuffle_epi64_mask): Likewise.
1110 * common/config/i386/i386-common.c
1111 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
1112 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
1113 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1114 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
1115 OPTION_MASK_ISA_AVX512BITALG_UNSET.
1116 (OPTION_MASK_ISA2_AVX512F_UNSET,
1117 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
1118 (ix86_handle_option): For -mno-general-regs-only, clear from
1119 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
1120 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
1121 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
1122 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
1123 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
1124 for moving from ix86_isa_flags2 to ix86_isa_flags.
1126 2017-12-23 Jakub Jelinek <jakub@redhat.com>
1129 * fold-const.c (struct contains_label_data): New type.
1130 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
1131 inside of a SWITCH_BODY seen during the walk.
1132 (contains_label_p): Use walk_tree instead of
1133 walk_tree_without_duplicates, prepare data for contains_label_1 and
1136 2017-12-22 Jakub Jelinek <jakub@redhat.com>
1139 * tree-iterator.c (alloc_stmt_list): Start with cleared
1140 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
1144 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
1145 enable avx512f explicitly in #pragma GCC target.
1146 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
1147 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
1148 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
1149 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
1150 __builtin_ia32_vpdpbusds_v8si_mask,
1151 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
1152 __builtin_ia32_vpdpbusds_v4si_mask,
1153 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
1154 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
1155 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
1156 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
1157 __builtin_ia32_vpdpwssds_v8si_mask,
1158 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
1159 __builtin_ia32_vpdpwssds_v4si_mask,
1160 __builtin_ia32_vpdpwssds_v4si_maskz): Use
1161 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
1162 just OPTION_MASK_ISA_AVX512VNNI.
1164 2017-12-22 Martin Jambor <mjambor@suse.cz>
1167 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
1170 2017-12-22 Julia Koval <julia.koval@intel.com>
1171 Sebastian Peryt <sebastian.peryt@intel.com>
1173 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
1174 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
1175 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
1176 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
1177 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
1178 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
1179 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
1180 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
1181 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
1182 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
1183 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
1184 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
1185 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
1186 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
1187 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
1188 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
1189 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
1190 _mm256_maskz_popcnt_epi64): New intrinsics.
1191 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
1192 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
1193 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
1194 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
1195 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
1196 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
1197 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
1198 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
1199 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
1200 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
1201 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
1202 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
1203 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
1204 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
1205 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
1206 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
1207 __builtin_ia32_vpshufbitqmb256_mask,
1208 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
1209 * config/i386/i386-c.c (__AVX512BITALG__): New.
1210 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
1211 (ix86_valid_target_attribute_inner_p): Ditto.
1212 (ix86_expand_args_builtin): Handle new types.
1213 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
1214 * config/i386/i386.opt: Add -mavx512bitalg.
1215 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
1216 avx512bitalgintrin.h.
1217 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
1218 (vpopcount<mode><mask_name>): Add more types.
1219 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
1220 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
1222 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1224 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
1225 Or in OPTION_MASK_ISA_AVX512F_SET.
1226 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1227 OPTION_MASK_ISA_AVX512VNNI_UNSET.
1228 (ix86_handle_option): Adjust for
1229 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
1230 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
1232 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
1233 isa_flag instead of isa_flag2.
1234 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
1235 isa_opts2 to isa_opts.
1236 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
1239 2017-12-22 Mike Stump <mikestump@comcast.net>
1240 Eric Botcazou <ebotcazou@adacore.com>
1242 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
1244 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1247 * config/i386/i386.c (ix86_function_arg_boundary): Return
1248 PARM_BOUNDARY for TYPE_EMPTY_P types.
1251 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
1252 if navail is >= dir.len.
1254 2017-12-21 Steve Ellcey <sellcey@cavium.com>
1256 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
1259 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1261 PR rtl-optimization/80747
1262 PR rtl-optimization/83512
1263 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
1264 succ edge from ENTRY, copy partition from e->dest to the newly
1266 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
1267 ENTRY, use BB_PARTITION of its successor block as current_partition.
1268 Don't copy partition when splitting succ edge from ENTRY.
1270 PR tree-optimization/83523
1271 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
1272 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
1273 (convert_mult_to_fma): Likewise.
1275 PR tree-optimization/83521
1276 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
1277 gimple_build_assign without code on result of
1278 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
1279 a VIEW_CONVERT_EXPR.
1281 2017-12-21 Andrew Pinski <apinski@cavium.com>
1282 Steve Ellcey <sellcey@cavium.com>
1284 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
1285 multi-arch for ilp32.
1287 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
1290 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
1291 constraints to operand 2.
1292 (*ashl<mode>3_mask_1): Ditto.
1293 (*<shift_insn><mode>3_mask): Ditto.
1294 (*<shift_insn><mode>3_mask_1): Ditto.
1295 (*<rotate_insn><mode>3_mask): Ditto.
1296 (*<rotate_insn><mode>3_mask_1): Ditto.
1298 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
1300 * reorg.c (make_return_insns): Reemit each insn with its own location.
1302 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1304 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
1305 generation for cases where splatting a value is not useful.
1306 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1307 across a vec_duplicate and a paradoxical subreg forming a vector
1308 mode to a vec_concat.
1310 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1312 * combine.c (simplify_set): Do not transform subregs to zero_extends
1313 if the destination is not a scalar int mode.
1315 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
1318 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
1319 shared zero if the input has overflowed.
1321 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1323 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
1324 support for Cortex-A55 and Cortex-A75.
1326 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1328 * common/config/arm/arm-common.c (compare_opt_names): Add function
1329 comment. Use strcmp instead of manual loop.
1331 2017-12-21 Martin Liska <mliska@suse.cz>
1333 PR gcov-profile/83509
1334 * gcov-dump.c (dump_gcov_file): Do not read info about
1335 support_unexecuted_blocks for gcda files.
1337 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1339 PR rtl-optimization/82973
1340 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
1341 (valid_for_const_vector_p): ... this.
1342 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
1343 (valid_for_const_vector_p): ... this. Adjust function comment.
1344 (gen_vec_duplicate): Adjust caller.
1345 * optabs.c (expand_vector_broadcast): Likewise.
1346 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
1347 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
1349 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
1350 instead of GET_CODE == CONST_FIXED.
1351 (simplify_subreg): Use CONST_FIXED_P macro instead of
1352 GET_CODE == CONST_FIXED.
1355 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
1356 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
1357 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
1358 (ix86_option_override_internal): Adjust for
1359 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
1360 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
1361 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
1362 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
1363 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
1364 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
1365 def_builtin2 for CET builtins.
1366 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
1367 up formatting in IX86_BUILTIN_RDPID code.
1368 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
1369 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
1370 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
1371 ix86_isa_flags2 to ix86_isa_flags.
1372 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
1374 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
1375 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
1376 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
1378 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
1379 Or in OPTION_MASK_ISA_AVX512F_SET.
1380 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1381 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
1382 (ix86_handle_option): Adjust for
1383 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
1384 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
1386 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1387 Alan Hayward <alan.hayward@arm.com>
1388 David Sherwood <david.sherwood@arm.com>
1390 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
1391 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1392 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
1393 Track polynomial offsets.
1395 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1396 Alan Hayward <alan.hayward@arm.com>
1397 David Sherwood <david.sherwood@arm.com>
1399 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1400 Treat drb->init as a poly_int. Fail if its misalignment wrt
1401 vector_alignment isn't known.
1403 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1404 Alan Hayward <alan.hayward@arm.com>
1405 David Sherwood <david.sherwood@arm.com>
1407 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
1409 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
1410 (vect_loop_versioning): Take the loop versioning threshold as a
1412 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
1413 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
1414 versioning_threshold.
1415 (vect_analyze_loop_2): Compute the loop versioning threshold
1416 whenever loop versioning is needed, and store it in the new
1417 field rather than combining it with the cost model threshold.
1418 (vect_transform_loop): Update call to vect_loop_versioning.
1419 Try to combine the loop versioning and cost thresholds here.
1421 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1422 Alan Hayward <alan.hayward@arm.com>
1423 David Sherwood <david.sherwood@arm.com>
1425 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
1426 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
1427 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
1428 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
1429 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
1430 (group_compare_offset): Update accordingly.
1431 (split_small_address_groups_p): Likewise.
1432 (record_use): Take addr_offset as a poly_uint64 rather than
1433 an unsigned HOST_WIDE_INT.
1434 (strip_offset): Return the offset as a poly_uint64 rather than
1435 an unsigned HOST_WIDE_INT.
1436 (record_group_use, split_address_groups): Track polynomial offsets.
1437 (add_iv_candidate_for_use): Likewise.
1438 (addr_offset_valid_p): Take the offset as a poly_int64 rather
1439 than a HOST_WIDE_INT.
1440 (strip_offset_1): Return the offset as a poly_int64 rather than
1443 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1444 Alan Hayward <alan.hayward@arm.com>
1445 David Sherwood <david.sherwood@arm.com>
1447 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
1448 rather than a HOST_WIDE_INT.
1449 * tree.c (get_binfo_at_offset): Likewise.
1451 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1452 Alan Hayward <alan.hayward@arm.com>
1453 David Sherwood <david.sherwood@arm.com>
1455 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
1456 rather than a HOST_WIDE_INT.
1457 * tree-sra.c (build_ref_for_offset): Likewise.
1459 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1460 Alan Hayward <alan.hayward@arm.com>
1461 David Sherwood <david.sherwood@arm.com>
1463 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
1465 * tree.c (mem_ref_offset): Likewise.
1466 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
1467 * builtins.c (get_object_alignment_2): Likewise.
1468 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
1469 * gimple-fold.c (get_base_constructor): Likewise.
1470 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
1471 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1473 * ipa-polymorphic-call.c
1474 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1475 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
1476 (get_ancestor_addr_info): Likewise.
1477 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
1478 * match.pd: Likewise.
1479 * tree-data-ref.c (dr_analyze_innermost): Likewise.
1480 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1481 * tree-eh.c (tree_could_trap_p): Likewise.
1482 * tree-object-size.c (addr_object_size): Likewise.
1483 * tree-ssa-address.c (copy_ref_info): Likewise.
1484 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
1485 (indirect_refs_may_alias_p): Likewise.
1486 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1487 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
1488 (non_rewritable_mem_ref_base): Likewise.
1489 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1490 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1491 * varasm.c (decode_addr_const): Likewise.
1493 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1494 Alan Hayward <alan.hayward@arm.com>
1495 David Sherwood <david.sherwood@arm.com>
1497 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
1498 offsets for MEM_REFs.
1500 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1501 Alan Hayward <alan.hayward@arm.com>
1502 David Sherwood <david.sherwood@arm.com>
1504 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
1505 as a poly_uint64 rather than an unsigned int.
1506 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
1508 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1509 Alan Hayward <alan.hayward@arm.com>
1510 David Sherwood <david.sherwood@arm.com>
1512 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
1514 (decode_addr_const): Update accordingly.
1516 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1517 Alan Hayward <alan.hayward@arm.com>
1518 David Sherwood <david.sherwood@arm.com>
1520 * tree.h (bit_field_size, bit_field_offset): New functions.
1521 * hsa-gen.c (gen_hsa_addr): Use them.
1522 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
1523 (simplify_vector_constructor): Likewise.
1524 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1525 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
1526 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
1527 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
1530 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1531 Alan Hayward <alan.hayward@arm.com>
1532 David Sherwood <david.sherwood@arm.com>
1534 * expr.h (emit_group_load, emit_group_load_into_temps)
1535 (emit_group_store): Take the size as a poly_int64 rather than an int.
1536 * expr.c (emit_group_load_1, emit_group_load): Likewise.
1537 (emit_group_load_into_temp, emit_group_store): Likewise.
1539 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1540 Alan Hayward <alan.hayward@arm.com>
1541 David Sherwood <david.sherwood@arm.com>
1543 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
1544 an unsigned int to a poly_uint64.
1545 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
1546 sizes as poly_uint64s rather than unsigned ints.
1547 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
1550 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1551 Alan Hayward <alan.hayward@arm.com>
1552 David Sherwood <david.sherwood@arm.com>
1554 * lra-constraints.c (emit_inc): Change inc_amount from an int
1557 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1558 Alan Hayward <alan.hayward@arm.com>
1559 David Sherwood <david.sherwood@arm.com>
1561 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
1563 (add_stack_var, stack_var_cmp, partition_stack_vars)
1564 (dump_stack_var_partition): Update accordingly.
1565 (alloc_stack_frame_space): Take the size as a poly_int64 rather
1566 than a HOST_WIDE_INT.
1567 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
1568 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
1569 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
1570 rather than a HOST_WIDE_INT.
1572 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1573 Alan Hayward <alan.hayward@arm.com>
1574 David Sherwood <david.sherwood@arm.com>
1576 * target.def (return_pops_args): Treat both the input and output
1577 sizes as poly_int64s rather than HOST_WIDE_INTS.
1578 * targhooks.h (default_return_pops_args): Update accordingly.
1579 * targhooks.c (default_return_pops_args): Likewise.
1580 * doc/tm.texi: Regenerate.
1581 * emit-rtl.h (incoming_args): Change pops_args, size and
1582 outgoing_args_size from int to poly_int64_pod.
1583 * function.h (expr_status): Change x_pending_stack_adjust and
1584 x_stack_pointer_delta from int to poly_int64.
1585 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
1586 (ARGS_SIZE_RTX): Update accordingly.
1587 * calls.c (highest_outgoing_arg_in_use): Change from int to
1589 (stack_usage_watermark, stored_args_watermark): New variables.
1590 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
1591 (emit_call_1): Change the stack_size and rounded_stack_size
1592 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
1594 (save_fixed_argument_area): Check stack_usage_watermark.
1595 (initialize_argument_information): Change old_pending_adj from
1596 a HOST_WIDE_INT * to a poly_int64_pod *.
1597 (compute_argument_block_size): Return the size as a poly_int64
1599 (finalize_must_preallocate): Track polynomial argument sizes.
1600 (compute_argument_addresses): Likewise.
1601 (internal_arg_pointer_based_exp): Track polynomial offsets.
1602 (mem_overlaps_already_clobbered_arg_p): Rename to...
1603 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
1604 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1605 Check stored_args_used_watermark.
1606 (load_register_parameters): Update accordingly.
1607 (check_sibcall_argument_overlap_1): Likewise.
1608 (combine_pending_stack_adjustment_and_call): Take the unadjusted
1609 args size as a poly_int64 rather than an int. Return a bool
1610 indicating whether the optimization was possible and return
1611 the new adjustment by reference.
1612 (check_sibcall_argument_overlap): Track polynomail argument sizes.
1613 Update stored_args_watermark.
1614 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
1615 (expand_call): Likewise. Maintain stack_usage_watermark and
1616 stored_args_watermark. Update calls to
1617 combine_pending_stack_adjustment_and_call.
1618 (emit_library_call_value_1): Handle polynomial argument sizes.
1619 Call stack_region_maybe_used_p and mark_stack_region_used.
1620 Maintain stack_usage_watermark.
1621 (store_one_arg): Likewise. Update call to
1622 mem_overlaps_already_clobbered_arg_p.
1623 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
1625 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
1626 * config/microblaze/microblaze.c (microblaze_function_prologue):
1628 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
1629 TARGET_RETURN_POPS_ARGS interface.
1630 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
1632 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
1633 * config/i386/i386.c (ix86_return_pops_args): Update for new
1634 TARGET_RETURN_POPS_ARGS interface.
1635 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
1636 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
1637 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
1638 TARGET_RETURN_POPS_ARGS interface.
1639 * config/vax/vax.c (vax_return_pops_args): Likewise.
1640 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
1641 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
1642 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
1643 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1644 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1645 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
1646 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1647 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1648 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
1649 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
1650 and x_stack_pointer_delta from int to poly_int64.
1651 * dojump.c (do_pending_stack_adjust): Update accordingly.
1652 * explow.c (allocate_dynamic_stack_space): Handle polynomial
1653 stack_pointer_deltas.
1654 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
1655 (pad_to_arg_alignment): Track polynomial offsets.
1656 (assign_parm_find_stack_rtl): Likewise.
1657 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
1658 * toplev.c (output_stack_usage): Update reference to
1659 current_function_pushed_stack_size.
1661 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1662 Alan Hayward <alan.hayward@arm.com>
1663 David Sherwood <david.sherwood@arm.com>
1665 * function.c (in_arg_offset, var_offset, dynamic_offset)
1666 (out_arg_offset, cfa_offset): Change from int to poly_int64.
1667 (instantiate_new_reg): Return the new offset as a poly_int64_pod
1668 rather than a HOST_WIDE_INT.
1669 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
1670 (instantiate_virtual_regs_in_insn): Likewise.
1672 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1673 Alan Hayward <alan.hayward@arm.com>
1674 David Sherwood <david.sherwood@arm.com>
1676 * rtl.h (get_args_size, add_args_size_note): New functions.
1677 (find_args_size_adjust): Return a poly_int64 rather than a
1679 (fixup_args_size_notes): Likewise. Make the same change to the
1680 end_args_size parameter.
1681 * rtlanal.c (get_args_size, add_args_size_note): New functions.
1682 * builtins.c (expand_builtin_trap): Use add_args_size_note.
1683 * calls.c (emit_call_1): Likewise.
1684 * explow.c (adjust_stack_1): Likewise.
1685 * cfgcleanup.c (old_insns_match_p): Update use of
1686 find_args_size_adjust.
1687 * combine.c (distribute_notes): Track polynomial arg sizes.
1688 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
1689 end_true_args_size, beg_delay_args_size and end_delay_args_size
1690 from HOST_WIDE_INT to poly_int64.
1691 (add_cfi_args_size): Take the args_size as a poly_int64 rather
1692 than a HOST_WIDE_INT.
1693 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
1694 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
1695 polynomial arg sizes.
1696 * emit-rtl.c (try_split): Use get_args_size.
1697 * recog.c (peep2_attempt): Likewise.
1698 * reload1.c (reload_as_needed): Likewise.
1699 * expr.c (find_args_size_adjust): Return the adjustment as a
1700 poly_int64 rather than a HOST_WIDE_INT.
1701 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
1702 to a poly_int64 and change the return type in the same way.
1703 (emit_single_push_insn): Track polynomial arg sizes.
1705 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1706 Alan Hayward <alan.hayward@arm.com>
1707 David Sherwood <david.sherwood@arm.com>
1709 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
1710 from HOST_WIDE_INT to poly_int64.
1711 * expr.c (push_block, emit_push_insn): Likewise.
1713 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1714 Alan Hayward <alan.hayward@arm.com>
1715 David Sherwood <david.sherwood@arm.com>
1717 * function.h (frame_space): Change start and length from HOST_WIDE_INT
1719 (get_frame_size): Return the size as a poly_int64 rather than a
1721 (frame_offset_overflow): Take the offset as a poly_int64 rather
1722 than a HOST_WIDE_INT.
1723 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1724 (assign_stack_temp): Likewise for the size.
1725 * function.c (get_frame_size): Return a poly_int64 rather than
1727 (frame_offset_overflow): Take the offset as a poly_int64 rather
1728 than a HOST_WIDE_INT.
1729 (try_fit_stack_local): Take the start, length and size as poly_int64s
1730 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
1731 rather than a HOST_WIDE_INT.
1732 (add_frame_space): Take the start and end as poly_int64s rather than
1734 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1735 (assign_stack_temp): Likewise for the size.
1736 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
1738 (find_temp_slot_from_address): Handle polynomial offsets.
1739 (combine_temp_slots): Likewise.
1740 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
1742 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
1743 poly_int64 rather than a HOST_WIDE_INT.
1744 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
1745 than a HOST_WIDE_INT.
1746 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
1747 polynomial frame offsets.
1748 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
1749 as a poly_int64 rather than an int.
1750 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
1751 * config/v850/v850-protos.h (compute_frame_size): Likewise.
1752 * config/v850/v850.c (compute_frame_size): Likewise.
1753 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
1754 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
1755 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
1756 * config/pa/pa.c (pa_compute_frame_size): Likewise.
1757 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
1758 rather than a HOST_WIDE_INT.
1759 * explow.c (get_dynamic_stack_base): Likewise.
1760 * final.c (final_start_function): Use the constant lower bound
1761 of the frame size for -Wframe-larger-than.
1762 * ira.c (do_reload): Adjust for new get_frame_size return type.
1763 * lra.c (lra): Likewise.
1764 * reload1.c (reload): Likewise.
1765 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
1766 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
1767 * rtlanal.c (get_initial_register_offset): Return the offset as
1768 a poly_int64 rather than a HOST_WIDE_INT.
1770 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1771 Alan Hayward <alan.hayward@arm.com>
1772 David Sherwood <david.sherwood@arm.com>
1774 * reload1.c (elim_table): Change initial_offset, offset and
1775 previous_offset from HOST_WIDE_INT to poly_int64_pod.
1776 (offsets_at): Change the target array's element type from
1777 HOST_WIDE_INT to poly_int64_pod.
1778 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
1779 (elimination_costs_in_insn, update_eliminable_offsets)
1780 (verify_initial_elim_offsets, set_offsets_for_label)
1781 (init_eliminable_invariants): Update after above changes.
1783 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1784 Alan Hayward <alan.hayward@arm.com>
1785 David Sherwood <david.sherwood@arm.com>
1787 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
1788 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
1789 (decomposition): Change start and end from HOST_WIDE_INT
1791 (decompose, immune_p): Update accordingly.
1792 (find_inc_amount): Return a poly_int64 rather than an int.
1793 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
1796 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1797 Alan Hayward <alan.hayward@arm.com>
1798 David Sherwood <david.sherwood@arm.com>
1800 * tree.h (get_inner_reference): Return the bitsize and bitpos
1801 as poly_int64_pods rather than HOST_WIDE_INT.
1802 * fold-const.h (ptr_difference_const): Return the pointer difference
1803 as a poly_int64_pod rather than a HOST_WIDE_INT.
1804 * expr.c (get_inner_reference): Return the bitsize and bitpos
1805 as poly_int64_pods rather than HOST_WIDE_INT.
1806 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
1808 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
1809 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
1810 (optimize_bit_field_compare): Update call to get_inner_reference.
1811 (decode_field_reference): Likewise.
1812 (fold_unary_loc): Track polynomial offsets and sizes.
1813 (split_address_to_core_and_offset): Return the bitpos as a
1814 poly_int64_pod rather than a HOST_WIDE_INT.
1815 (ptr_difference_const): Likewise for the pointer difference.
1816 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
1817 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
1818 * dbxout.c (dbxout_expand_expr): Likewise.
1819 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
1820 (loc_list_from_tree_1, fortran_common): Likewise.
1821 * gimple-laddress.c (pass_laddress::execute): Likewise.
1822 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
1823 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
1824 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1825 * tree-affine.c (tree_to_aff_combination): Likewise.
1826 (get_inner_reference_aff): Likewise.
1827 * tree-data-ref.c (split_constant_offset_1): Likewise.
1828 (dr_analyze_innermost): Likewise.
1829 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
1830 * tree-sra.c (ipa_sra_check_caller): Likewise.
1831 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1832 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
1833 (instrument_bool_enum_load, instrument_object_size): Likewise.
1834 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
1835 to get_inner_reference.
1836 * hsa-gen.c (gen_hsa_addr): Likewise.
1837 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
1838 * tsan.c (instrument_expr): Likewise.
1839 * match.pd: Update call to ptr_difference_const.
1841 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1842 Alan Hayward <alan.hayward@arm.com>
1843 David Sherwood <david.sherwood@arm.com>
1845 * fold-const.c (fold_comparison): Track sizes and offsets as
1846 poly_int64s rather than HOST_WIDE_INTs when folding address
1849 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1850 Alan Hayward <alan.hayward@arm.com>
1851 David Sherwood <david.sherwood@arm.com>
1853 * expr.h (get_bit_range): Return the bitstart and bitend as
1854 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1855 as a poly_int64 rather than a HOST_WIDE_INT.
1856 * expr.c (get_bit_range): Likewise.
1857 (expand_assignment): Update call accordingly.
1858 * fold-const.c (optimize_bit_field_compare): Likewise.
1860 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1861 Alan Hayward <alan.hayward@arm.com>
1862 David Sherwood <david.sherwood@arm.com>
1864 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1865 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1866 as an optional final parameter.
1867 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1868 (aarch64_legitimate_address_p): Likewise.
1869 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1870 instead of an rtx code.
1871 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1872 (aarch64_legitimate_address_hook_p): Likewise.
1873 (aarch64_print_ldpstp_address): Likewise.
1874 (aarch64_print_operand_address): Likewise.
1875 (aarch64_address_cost): Likewise.
1876 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1877 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1878 (aarch64_mem_pair_lanes_operand): Likewise.
1880 2017-12-20 Richard Biener <rguenther@suse.de>
1882 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1883 update_stmt_if_modified.
1885 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1887 PR tree-optimization/83491
1888 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1889 before walking uses. Improve coding style and comments.
1891 2017-12-20 Tom de Vries <tom@codesourcery.com>
1893 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1895 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1898 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1900 (pass_ipa_free_fn_summary): Add small_p private data member,
1901 initialize to false in the ctor.
1902 (pass_ipa_free_fn_summary::clone,
1903 pass_ipa_free_fn_summary::set_pass_param,
1904 pass_ipa_free_fn_summary::gate): New methods.
1905 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1906 | TODO_dump_symtab if small_p.
1907 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1908 entry and add another instance of the pass with false parm after
1910 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1911 ipa_free_fn_summary here.
1913 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1915 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1917 2017-12-20 Martin Sebor <msebor@redhat.com>
1920 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1921 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1922 (expand_builtin_strncmp): Same.
1924 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1927 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1928 allowed debug stmts before labels.
1929 (expand_gimple_basic_block): Likewise.
1930 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1931 * gimple-iterator.h (gsi_after_labels): Likewise.
1932 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1933 rename reused variable, and simplify using gsi_move_before.
1934 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1935 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1936 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1937 (gimple_verify_flow_info, gimple_block_label): Likewise.
1938 (make_blocks): Move debug markers after adjacent labels.
1939 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1940 allowed debug insns outside blocks.
1941 * df-scan.c (df_insn_delete): Likewise.
1942 * lra-constraints.c (update_ebb_live_info): Likewise.
1943 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1944 (vt_initialize, delete_vta_debug_insns): Likewise.
1945 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1947 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1948 Alan Hayward <alan.hayward@arm.com>
1949 David Sherwood <david.sherwood@arm.com>
1951 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1952 (round_up_to_byte_boundary): Likewise.
1953 * expr.h (get_bit_range): Add temporary shim.
1954 * gimple-ssa-store-merging.c (store_operand_info): Change the
1955 bitsize, bitpos, bitregion_start and bitregion_end fields from
1956 unsigned HOST_WIDE_INT to poly_uint64.
1957 (merged_store_group): Likewise load_align_base.
1958 (compatible_load_p, compatible_load_p): Update accordingly.
1959 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1960 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1961 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1962 bitregion_start and bitregion_end as poly_uint64s rather than
1963 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1964 (handled_load): Take the bitsize, bitpos,
1965 bitregion_start and bitregion_end as poly_uint64s rather than
1966 unsigned HOST_WIDE_INTs.
1967 (pass_store_merging::process_store): Update call to
1968 mem_valid_for_store_merging.
1970 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1971 Alan Hayward <alan.hayward@arm.com>
1972 David Sherwood <david.sherwood@arm.com>
1974 * builtins.c (get_object_alignment_2): Track polynomial offsets
1975 and sizes. Update the alignment handling.
1977 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1978 Alan Hayward <alan.hayward@arm.com>
1979 David Sherwood <david.sherwood@arm.com>
1981 * tree.h (get_inner_reference): Add a version that returns the
1982 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1983 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1984 the case in which bitpos is not associated with the first byte.
1986 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1987 Alan Hayward <alan.hayward@arm.com>
1988 David Sherwood <david.sherwood@arm.com>
1990 * tree-affine.h (get_inner_reference_aff): Return the size as a
1992 * tree-affine.c (get_inner_reference_aff): Likewise.
1993 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1994 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1996 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1997 Alan Hayward <alan.hayward@arm.com>
1998 David Sherwood <david.sherwood@arm.com>
2000 * fold-const.c (pointer_may_wrap_p): Take the offset as a
2001 HOST_WIDE_INT rather than a poly_int64.
2003 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2004 Alan Hayward <alan.hayward@arm.com>
2005 David Sherwood <david.sherwood@arm.com>
2007 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
2008 HOST_WIDE_INT to poly_int64_pod.
2009 (perform_symbolic_merge): Update accordingly.
2010 (bswap_replace): Likewise.
2012 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2013 Alan Hayward <alan.hayward@arm.com>
2014 David Sherwood <david.sherwood@arm.com>
2016 * tree-affine.h (aff_tree::offset): Change from widest_int
2018 (wide_int_ext_for_comb): Delete.
2019 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
2020 constants as poly_widest_int rather than widest_int.
2021 (aff_combination_constant_multiple_p): Return the multiplier
2022 as a poly_widest_int.
2023 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
2025 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
2026 version static and add an overload for poly_widest_int.
2027 (aff_combination_const, aff_combination_add_cst)
2028 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
2029 the constants as poly_widest_int rather than widest_int.
2030 (tree_to_aff_combination): Generalize INTEGER_CST case to
2032 (aff_combination_to_tree): Track offsets as poly_widest_ints.
2033 (aff_combination_add_product, aff_combination_mult): Handle
2035 (aff_combination_constant_multiple_p): Return the multiplier
2036 as a poly_widest_int.
2037 * tree-predcom.c (determine_offset): Return the offset as a
2039 (split_data_refs_to_components, suitable_component_p): Update
2041 (valid_initializer_p): Update call to
2042 aff_combination_constant_multiple_p.
2043 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
2044 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
2045 as a poly_int64 rather than a HOST_WIDE_INT.
2046 (get_address_cost): Handle polynomial offsets.
2047 (iv_elimination_compare_lt): Likewise.
2048 (rewrite_use_nonlinear_expr): Likewise.
2050 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2051 Alan Hayward <alan.hayward@arm.com>
2052 David Sherwood <david.sherwood@arm.com>
2054 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
2055 as a poly_int64_pod rather than a HOST_WIDE_INT.
2056 (get_addr_base_and_unit_offset): Likewise.
2057 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
2058 (get_addr_base_and_unit_offset): Likewise.
2059 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
2060 to poly_int64 in example.
2061 * fold-const.c (fold_binary_loc): Update call to
2062 get_addr_base_and_unit_offset.
2063 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
2064 (maybe_canonicalize_mem_ref_addr): Likewise.
2065 (gimple_fold_stmt_to_constant_1): Likewise.
2066 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2068 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
2069 * match.pd: Likewise.
2070 * omp-low.c (lower_omp_target): Likewise.
2071 * tree-sra.c (build_ref_for_offset): Likewise.
2072 (build_debug_ref_for_model): Likewise.
2073 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
2074 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
2075 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
2076 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
2077 (constant_pointer_difference): Likewise.
2078 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
2079 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
2080 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
2081 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
2082 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
2083 (set_ssa_val_to): Likewise.
2084 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
2085 (maybe_diag_stxncpy_trunc): Likewise.
2086 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
2087 * tree.c (build_simple_mem_ref_loc): Likewise.
2088 (array_at_struct_end_p): Likewise.
2090 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2091 Alan Hayward <alan.hayward@arm.com>
2092 David Sherwood <david.sherwood@arm.com>
2094 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
2095 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
2096 (get_ref_base_and_extent_hwi): Declare.
2097 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
2098 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
2099 (get_ref_base_and_extent_hwi): New function.
2100 * cfgexpand.c (expand_debug_expr): Update call to
2101 get_ref_base_and_extent.
2102 * dwarf2out.c (add_var_loc_to_decl): Likewise.
2103 * gimple-fold.c (get_base_constructor): Return the offset as a
2104 poly_int64_pod rather than a HOST_WIDE_INT.
2105 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
2106 * ipa-polymorphic-call.c
2107 (ipa_polymorphic_call_context::set_by_invariant)
2108 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
2109 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
2110 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
2111 rather than get_ref_base_and_extent.
2112 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
2113 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
2114 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
2116 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
2117 call to get_ref_base_and_extent.
2118 * tree-sra.c (create_access, get_access_for_expr): Likewise.
2119 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
2120 (stmt_kills_ref_p): Likewise.
2121 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2122 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
2124 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
2125 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
2126 when calling native_encode_expr.
2127 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
2128 call to get_ref_base_and_extent.
2129 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
2130 get_ref_base_and_extent.
2131 * var-tracking.c (track_expr_p): Likewise.
2133 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2134 Alan Hayward <alan.hayward@arm.com>
2135 David Sherwood <david.sherwood@arm.com>
2137 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
2138 HOST_WIDE_INT to poly_int64_pod.
2139 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
2140 polynomail parameter offsets.
2142 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2143 Alan Hayward <alan.hayward@arm.com>
2144 David Sherwood <david.sherwood@arm.com>
2146 * gengtype.c (main): Handle poly_int64_pod.
2147 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
2148 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
2149 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
2150 from HOST_WIDE_INT to poly_int64_pod.
2151 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
2152 (copy_cfa): New function.
2153 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
2154 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
2155 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
2156 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
2157 (initial_return_save): Treat offsets as poly_ints.
2158 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
2159 offset is nonconstant.
2160 (reg_save): Take the offset as a poly_int64. Fall back to
2161 DW_CFA_expression for nonconstant offsets.
2162 (queue_reg_save): Take the offset as a poly_int64.
2163 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
2165 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2166 Alan Hayward <alan.hayward@arm.com>
2167 David Sherwood <david.sherwood@arm.com>
2169 * rtl.h (operand_subword, operand_subword_force): Take the offset
2170 as a poly_uint64 an unsigned int.
2171 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
2173 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2174 Alan Hayward <alan.hayward@arm.com>
2175 David Sherwood <david.sherwood@arm.com>
2177 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
2178 'p' format code. Use INT_LIST rather than SUBREG as the example of
2179 a code with an XINT and an XEXP. Remove the implication that
2180 accessing an rtx field using XINT is expected to work.
2181 * rtl.def (SUBREG): Change format from "ei" to "ep".
2182 * rtl.h (rtunion::rt_subreg): New field.
2183 (XCSUBREG): New macro.
2184 (SUBREG_BYTE): Use it.
2185 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
2186 Update constructor accordingly.
2187 (subreg_shape::operator ==): Update accordingly.
2188 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
2189 than an unsigned int.
2190 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
2191 a poly_uint64 rather than an unsigned int.
2192 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
2193 than an unsigned int.
2194 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
2195 (subreg_size_highpart_offset): Return a poly_uint64 rather than
2196 an unsigned int. Take the sizes as poly_uint64s.
2197 (subreg_offset_from_lsb): Return a poly_uint64 rather than
2198 an unsigned int. Take the shift as a poly_uint64 rather than
2200 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
2201 as a poly_uint64 rather than an unsigned int.
2202 (simplify_subreg_regno): Likewise.
2203 (byte_lowpart_offset): Return the memory offset as a poly_int64
2205 (subreg_memory_offset): Likewise. Take the subreg offset as a
2206 poly_uint64 rather than an unsigned int.
2207 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
2208 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
2209 poly_uint64 rather than an unsigned int.
2210 * rtl.c (rtx_format): Describe 'p' in comment.
2211 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
2212 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
2213 offset as a poly_uint64 rather than an unsigned int.
2214 (byte_lowpart_offset): Return the memory offset as a poly_int64
2216 (subreg_memory_offset): Likewise. Take the subreg offset as a
2217 poly_uint64 rather than an unsigned int.
2218 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
2219 mode sizes as poly_uint64s rather than unsigned ints. Return a
2220 poly_uint64 rather than an unsigned int.
2221 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
2222 (copy_insn_1): Handle 'p'.
2223 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
2224 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
2225 an unsigned int. Return the shift in the same way.
2226 (subreg_lsb): Return the shift as a poly_uint64 rather than an
2228 (subreg_size_offset_from_lsb): Take the sizes and shift as
2229 poly_uint64s rather than unsigned ints. Return the offset as
2231 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
2232 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
2234 * rtlhash.c (add_rtx): Handle 'p'.
2235 * genemit.c (gen_exp): Likewise.
2236 * gengenrtl.c (type_from_format, gendef): Likewise.
2237 * gensupport.c (subst_pattern_match, get_alternatives_number)
2238 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
2239 (subst_dup): Likewise.
2240 * gengtype.c (adjust_field_rtx_def): Likewise.
2241 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
2242 (match_pattern_2): Likewise.
2243 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
2244 (rtx_test::subreg_field): New function.
2245 (operator ==, safe_to_hoist_p, transition_parameter_type)
2246 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
2247 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
2248 * genpeep.c (match_rtx): Likewise.
2249 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
2250 (rtx_writer::print_rtx_operand): Handle 'p'.
2251 (print_value): Handle SUBREG.
2252 * read-rtl.c (apply_int_iterator): Likewise.
2253 (rtx_reader::read_rtx_operand): Handle 'p'.
2254 * alias.c (rtx_equal_for_memref_p): Likewise.
2255 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
2256 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
2258 * calls.c (expand_call): Likewise.
2259 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
2260 (make_extraction, gen_lowpart_for_combine): Likewise.
2261 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
2263 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
2264 rather than an unsigned int. Treat subreg offsets as poly_ints.
2265 (exp_equiv_p): Handle 'p'.
2266 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
2267 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
2268 * dse.c (find_shift_sequence): Likewise.
2269 * dwarf2out.c (rtl_for_decl_location): Likewise.
2270 * expmed.c (extract_low_bits): Likewise.
2271 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
2272 (expand_expr_real_2): Likewise.
2273 * final.c (alter_subreg): Likewise.
2274 (leaf_renumber_regs_insn): Handle 'p'.
2275 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
2276 Treat subreg offsets as poly_ints.
2277 * fwprop.c (forward_propagate_and_simplify): Likewise.
2278 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
2279 * ira.c (get_subreg_tracking_sizes): Likewise.
2280 * ira-conflicts.c (go_through_subreg): Likewise.
2281 * ira-lives.c (process_single_reg_class_operands): Likewise.
2282 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
2283 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
2284 as a poly_uint64 rather than an unsigned int.
2285 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
2286 subreg offsets as poly_ints.
2287 * lra-constraints.c (operands_match_p): Handle 'p'.
2288 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
2289 * lra-spills.c (assign_mem_slot): Likewise.
2290 * postreload.c (move2add_valid_value_p): Likewise.
2291 * recog.c (general_operand, indirect_operand): Likewise.
2292 * regcprop.c (copy_value, maybe_mode_change): Likewise.
2293 (copyprop_hardreg_forward_1): Likewise.
2294 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
2295 (record_subregs_of_mode): Likewise.
2296 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
2297 * reload.c (operands_match_p): Handle 'p'.
2298 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
2299 * reload1.c (alter_reg, choose_reload_regs): Likewise.
2300 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
2301 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
2302 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
2303 (simplify_const_poly_int_tests<N>::run): Likewise.
2304 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
2305 a poly_uint64 rather than an unsigned int.
2306 * valtrack.c (debug_lowpart_subreg): Likewise.
2307 * var-tracking.c (var_lowpart): Likewise.
2308 (loc_cmp): Handle 'p'.
2310 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2311 Alan Hayward <alan.hayward@arm.com>
2312 David Sherwood <david.sherwood@arm.com>
2314 * ira.c (get_subreg_tracking_sizes): New function.
2315 (init_live_subregs): Take an integer size rather than a register.
2316 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
2317 to init_live_subregs.
2319 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2320 Alan Hayward <alan.hayward@arm.com>
2321 David Sherwood <david.sherwood@arm.com>
2323 * expr.c (store_constructor_field): Change bitsize from a
2324 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
2325 HOST_WIDE_INT to a poly_int64.
2326 (store_constructor): Change size from a HOST_WIDE_INT to
2328 (store_field): Likewise bitsize and bitpos.
2330 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2331 Alan Hayward <alan.hayward@arm.com>
2332 David Sherwood <david.sherwood@arm.com>
2334 * expmed.h (store_bit_field): Change bitregion_start and
2335 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
2336 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
2337 (store_bit_field_1, store_integral_bit_field, store_bit_field)
2338 (store_fixed_bit_field, store_split_bit_field): Likewise.
2339 * expr.c (store_constructor_field, store_field): Likewise.
2340 (optimize_bitfield_assignment_op): Likewise. Make the same change
2341 to bitsize and bitpos.
2342 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
2343 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
2344 same change in the constructor arguments.
2345 (get_best_mode): Change bitregion_start and bitregion_end from
2346 unsigned HOST_WIDE_INT to poly_uint64.
2347 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2348 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
2350 (bit_field_mode_iterator::next_mode): Update for new types
2351 of m_bitregion_start and m_bitregion_end.
2352 (get_best_mode): Change bitregion_start and bitregion_end from
2353 unsigned HOST_WIDE_INT to poly_uint64.
2355 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2356 Alan Hayward <alan.hayward@arm.com>
2357 David Sherwood <david.sherwood@arm.com>
2359 * rtl.h (simplify_gen_subreg): Add a temporary overload that
2360 accepts poly_uint64 offsets.
2361 * expmed.h (extract_bit_field): Take bitsize and bitnum as
2362 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2363 * expmed.c (lowpart_bit_field_p): Likewise.
2364 (extract_bit_field_as_subreg): New function, split out from...
2365 (extract_bit_field_1): ...here. Take bitsize and bitnum as
2366 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
2367 extractions, check that BITSIZE matches the size of the extracted
2368 value and that BITNUM is an exact multiple of that size.
2369 If all else fails, try forcing the value into memory if
2370 BITNUM is variable, and adjusting the address so that the
2371 offset is constant. Split the part that can only handle constant
2372 bitsize and bitnum out into...
2373 (extract_integral_bit_field): ...this new function.
2374 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
2375 rather than unsigned HOST_WIDE_INTs.
2377 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2378 Alan Hayward <alan.hayward@arm.com>
2379 David Sherwood <david.sherwood@arm.com>
2381 * expmed.h (store_bit_field): Take bitsize and bitnum as
2382 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2383 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
2384 that returns the byte size.
2385 (store_bit_field_1): Take bitsize and bitnum as
2386 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
2387 to simple_mem_bitfield_p. Split the part that can only handle
2388 constant bitsize and bitnum out into...
2389 (store_integral_bit_field): ...this new function.
2390 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
2391 than unsigned HOST_WIDE_INTs.
2392 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
2394 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2395 Alan Hayward <alan.hayward@arm.com>
2396 David Sherwood <david.sherwood@arm.com>
2398 * lra-int.h (lra_reg): Change offset from int to poly_int64.
2399 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
2401 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2402 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
2403 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
2404 offset as a poly_int64 rather than an int.
2405 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
2406 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2407 * lra-constraints.c (equiv_address_substitution): Track offsets
2409 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
2410 (curr_insn_transform): Handle the new form of sp_offset.
2411 * lra-eliminations.c (lra_elim_table): Change previous_offset
2412 and offset from HOST_WIDE_INT to poly_int64.
2413 (print_elim_table, update_reg_eliminate): Update accordingly.
2414 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
2415 (get_elimination): Update accordingly.
2416 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
2417 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2418 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
2419 poly_int64 offsets generally.
2420 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
2421 (mark_not_eliminable, init_elimination): Update accordingly.
2422 (remove_reg_equal_offset_note): Return a bool and pass the new
2423 offset back by pointer as a poly_int64.
2424 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
2425 rather than a HOST_WIDE_INT.
2426 (do_remat): Track offsets poly_int64s.
2427 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
2429 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2430 Alan Hayward <alan.hayward@arm.com>
2431 David Sherwood <david.sherwood@arm.com>
2433 * rtl.h (mem_attrs): Add a default constructor. Change size and
2434 offset from HOST_WIDE_INT to poly_int64.
2435 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
2436 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
2437 (widen_memory_access): Take the sizes and offsets as poly_int64s
2438 rather than HOST_WIDE_INTs.
2439 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
2440 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
2442 (adjust_offset_for_component_ref): Change the offset from a
2443 HOST_WIDE_INT to a poly_int64.
2444 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
2445 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
2446 * dce.c (find_call_stack_args): Likewise.
2447 * dse.c (record_store): Likewise.
2448 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
2449 * print-rtl.c (rtx_writer::print_rtx): Likewise.
2450 * read-rtl-function.c (test_loading_mem): Likewise.
2451 * rtlanal.c (may_trap_p_1): Likewise.
2452 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
2453 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
2454 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
2455 (mem_attrs::mem_attrs): New function.
2456 (set_mem_attributes_minus_bitpos): Change bitpos from a
2457 HOST_WIDE_INT to poly_int64.
2458 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
2459 (clear_mem_offset, clear_mem_size, change_address)
2460 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
2461 initialize mem_attrs.
2462 (set_mem_offset, set_mem_size, adjust_address_1)
2463 (adjust_automodify_address_1, offset_address, widen_memory_access):
2464 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
2466 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2467 Alan Hayward <alan.hayward@arm.com>
2468 David Sherwood <david.sherwood@arm.com>
2470 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
2471 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
2472 rather than 0 to represent an unknown size. Assert that the size
2473 is known when the mode isn't BLKmode.
2474 (may_trap_p_1): Use -1 for unknown sizes.
2475 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
2477 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2478 Alan Hayward <alan.hayward@arm.com>
2479 David Sherwood <david.sherwood@arm.com>
2481 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
2482 to poly_int64. Update commentary for positions_needed.large.
2483 (read_info_type): Change offset and width from HOST_WIDE_INT
2485 (set_usage_bits): Likewise.
2486 (canon_address): Return the offset as a poly_int64 rather than
2487 a HOST_WIDE_INT. Use strip_offset_and_add.
2488 (set_all_positions_unneeded, any_positions_needed_p): Use
2489 positions_needed.large to track stores with non-constant widths.
2490 (all_positions_needed_p): Likewise. Take the offset and width
2491 as poly_int64s rather than ints. Assert that rhs is nonnull.
2492 (record_store): Cope with non-constant offsets and widths.
2493 Nullify the rhs of an earlier store if we can't tell which bytes
2495 (find_shift_sequence): Take the access_size and shift as poly_int64s
2497 (get_stored_val): Take the read_offset and read_width as poly_int64s
2498 rather than HOST_WIDE_INTs.
2499 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
2500 non-constant offsets and widths.
2502 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2503 Alan Hayward <alan.hayward@arm.com>
2504 David Sherwood <david.sherwood@arm.com>
2506 * inchash.h (inchash::hash::add_poly_int): New function.
2507 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
2508 Use poly_int64 rather than HOST_WIDE_INT.
2509 (ao_ref::max_size_known_p): New function.
2510 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
2511 rather than HOST_WIDE_INT.
2512 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
2513 to temporaries until its interface is adjusted to match.
2514 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
2515 (aliasing_component_refs_p, decl_refs_may_alias_p)
2516 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
2517 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
2518 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
2520 * alias.c (ao_ref_from_mem): Likewise.
2521 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2522 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
2523 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
2524 (maybe_trim_complex_store, maybe_trim_constructor_store)
2525 (live_bytes_read, dse_classify_store): Likewise.
2526 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
2527 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
2528 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
2529 (vn_reference_lookup_3): Likewise.
2530 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
2532 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2533 Alan Hayward <alan.hayward@arm.com>
2534 David Sherwood <david.sherwood@arm.com>
2536 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
2537 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
2538 instead of ranges_overlap_p.
2540 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2541 Alan Hayward <alan.hayward@arm.com>
2542 David Sherwood <david.sherwood@arm.com>
2544 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
2545 sizes as poly_int64s rather than HOST_WIDE_INTs.
2547 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2548 Alan Hayward <alan.hayward@arm.com>
2549 David Sherwood <david.sherwood@arm.com>
2551 * gimple-fold.h (fold_ctor_reference): Take the offset and size
2552 as poly_uint64 rather than unsigned HOST_WIDE_INT.
2553 * gimple-fold.c (fold_ctor_reference): Likewise.
2555 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2556 Alan Hayward <alan.hayward@arm.com>
2557 David Sherwood <david.sherwood@arm.com>
2559 * target.def (dwarf_poly_indeterminate_value): New hook.
2560 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
2561 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
2562 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
2563 * doc/tm.texi: Regenerate.
2564 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
2565 offset as a poly_int64.
2566 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
2567 offset as a poly_int64.
2568 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
2569 Take the offset as a poly_int64.
2570 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
2571 (frame_pointer_fb_offset): Change to a poly_int64.
2572 (int_loc_descriptor): Take the offset as a poly_int64. Use
2573 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
2574 (based_loc_descr): Take the offset as a poly_int64.
2575 Use strip_offset_and_add to handle (plus X (const)).
2576 Use new_reg_loc_descr instead of an open-coded version of the
2577 previous implementation.
2578 (mem_loc_descriptor): Handle CONST_POLY_INT.
2579 (compute_frame_pointer_to_fb_displacement): Take the offset as a
2580 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
2582 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2583 Alan Hayward <alan.hayward@arm.com>
2584 David Sherwood <david.sherwood@arm.com>
2586 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
2587 (gen_rtx_REG_offset): Take the offset as a poly_int64.
2588 * inchash.h (inchash::hash::add_poly_hwi): New function.
2589 * gengtype.c (main): Register poly_int64.
2590 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
2591 offset as a poly_int.
2592 (reg_attr_hasher::equal): Use must_eq to compare offsets.
2593 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
2594 offset as a poly_int64.
2595 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
2596 * print-rtl.c (print_poly_int): New function.
2597 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
2599 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
2601 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
2602 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
2603 (vt_get_decl_and_offset): Return the offset as a poly_int64.
2604 Enforce track_offset_p for parts of a PARALLEL.
2605 (vt_add_function_parameter): Use const_offset for the final
2606 offset to track. Use get_tracked_reg_offset for the parts
2609 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2610 Alan Hayward <alan.hayward@arm.com>
2611 David Sherwood <david.sherwood@arm.com>
2613 * target.def (truly_noop_truncation): Take poly_uint64s instead of
2614 unsigned ints. Change default to hook_bool_puint64_puint64_true.
2615 * doc/tm.texi: Regenerate.
2616 * hooks.h (hook_bool_uint_uint_true): Delete.
2617 (hook_bool_puint64_puint64_true): Declare.
2618 * hooks.c (hook_bool_uint_uint_true): Delete.
2619 (hook_bool_puint64_puint64_true): New function.
2620 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
2621 instead of unsigned ints.
2622 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
2623 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
2625 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2626 Alan Hayward <alan.hayward@arm.com>
2627 David Sherwood <david.sherwood@arm.com>
2629 * optabs.h (expand_operand): Add an int_value field.
2630 (create_expand_operand): Add an int_value parameter and use it
2631 to initialize the new expand_operand field.
2632 (create_integer_operand): Replace with a declaration of a function
2633 that accepts poly_int64s. Move the implementation to...
2634 * optabs.c (create_integer_operand): ...here.
2635 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
2636 the mode preserves the value of int_value, instead of calling
2637 const_int_operand on the rtx. Use gen_int_mode to generate
2640 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2641 Alan Hayward <alan.hayward@arm.com>
2642 David Sherwood <david.sherwood@arm.com>
2644 * dumpfile.h (dump_dec): Declare.
2645 * dumpfile.c (dump_dec): New function.
2646 * pretty-print.h (pp_wide_integer): Turn into a function and
2647 declare a poly_int version.
2648 * pretty-print.c (pp_wide_integer): New function for poly_ints.
2650 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2651 Alan Hayward <alan.hayward@arm.com>
2652 David Sherwood <david.sherwood@arm.com>
2654 * doc/generic.texi (POLY_INT_CST): Document.
2655 * tree.def (POLY_INT_CST): New tree code.
2656 * treestruct.def (TS_POLY_INT_CST): New tree layout.
2657 * tree-core.h (tree_poly_int_cst): New struct.
2658 (tree_node): Add a poly_int_cst field.
2659 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
2660 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
2661 instead of a wide_int_ref.
2662 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
2664 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2665 instead of an unsigned HOST_WIDE_INT.
2666 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
2667 (ptrdiff_tree_p): Declare.
2668 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
2669 extern inline implementations if the target doesn't use POLY_INT_CST.
2670 (poly_int_tree_p): New function.
2671 (wi::unextended_tree): New class.
2672 (wi::int_traits <unextended_tree>): New override.
2673 (wi::extended_tree): Add a default constructor.
2674 (wi::extended_tree::get_tree): New function.
2675 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
2676 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
2677 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
2678 (wi::tree_to_poly_wide_ref): New typedefs.
2679 (wi::ints_for): Provide overloads for extended_tree and
2681 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
2682 (wi::to_wide): New functions.
2683 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
2684 * tree.c (poly_int_cst_hasher): New struct.
2685 (poly_int_cst_hash_table): New variable.
2686 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
2687 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
2689 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
2690 (init_ttree): Initialize poly_int_cst_hash_table.
2691 (build_int_cst, build_int_cst_type, build_invariant_address): Take
2692 a poly_int64 instead of a HOST_WIDE_INT.
2693 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2694 instead of an unsigned HOST_WIDE_INT.
2695 (wide_int_to_tree): Rename to...
2696 (wide_int_to_tree_1): ...this.
2697 (build_new_poly_int_cst, build_poly_int_cst): New functions.
2698 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
2699 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
2700 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
2701 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
2702 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
2704 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
2705 (streamer_read_tree_body): Likewise.
2706 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
2707 (streamer_write_tree_body): Likewise.
2708 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
2709 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
2710 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
2711 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
2712 * gimple-expr.h (is_gimple_constant): Likewise.
2713 * gimplify.c (maybe_with_size_expr): Likewise.
2714 * print-tree.c (print_node): Likewise.
2715 * tree-data-ref.c (data_ref_compare_tree): Likewise.
2716 * tree-pretty-print.c (dump_generic_node): Likewise.
2717 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2718 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
2719 * tree-vrp.c (compare_values_warnv): Likewise.
2720 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
2721 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
2722 (force_expr_to_var_cost): Likewise.
2723 * tree-ssa-loop.c (for_each_index): Likewise.
2724 * fold-const.h (build_invariant_address, size_int_kind): Take a
2725 poly_int64 instead of a HOST_WIDE_INT.
2726 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
2727 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
2729 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
2730 (int_const_binop_2): New function, split out from...
2731 (int_const_binop_1): ...here. Handle POLY_INT_CST.
2732 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
2733 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
2734 * gimple-ssa-strength-reduction.c (slsr_process_add)
2735 (slsr_process_mul): Check for INTEGER_CSTs before using them
2737 * stor-layout.c (bits_from_bytes): New function.
2738 (bit_from_pos): Use it.
2739 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
2740 by BITS_PER_UNIT to get the TYPE_SIZE.
2741 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
2742 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
2744 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2745 Alan Hayward <alan.hayward@arm.com>
2746 David Sherwood <david.sherwood@arm.com>
2748 * doc/rtl.texi (const_poly_int): Document. Also document the
2749 rtl sharing behavior.
2750 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
2751 * rtl.h (const_poly_int_def): New struct.
2752 (rtx_def::u): Add a cpi field.
2753 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
2754 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
2755 (wi::rtx_to_poly_wide_ref): New typedef
2756 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
2757 (poly_int_rtx_p): New functions.
2758 (trunc_int_for_mode): Declare a poly_int64 version.
2759 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2760 (immed_wide_int_const): Take a poly_wide_int_ref rather than
2762 (strip_offset): Declare.
2763 (strip_offset_and_add): New function.
2764 * rtl.def (CONST_POLY_INT): New rtx code.
2765 * rtl.c (rtx_size): Handle CONST_POLY_INT.
2766 (shared_const_p): Use poly_int_rtx_p.
2767 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
2769 (gen_int_shift_amount): Likewise.
2770 * emit-rtl.c (const_poly_int_hasher): New class.
2771 (const_poly_int_htab): New variable.
2772 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
2773 (const_poly_int_hasher::hash): New function.
2774 (const_poly_int_hasher::equal): Likewise.
2775 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
2776 (immed_wide_int_const): Rename to...
2777 (immed_wide_int_const_1): ...this and make static.
2778 (immed_wide_int_const): New function, taking a poly_wide_int_ref
2779 instead of a wide_int_ref.
2780 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
2781 (gen_lowpart_common): Handle CONST_POLY_INT.
2782 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
2783 * cselib.c (cselib_hash_rtx): Likewise.
2784 * dwarf2out.c (const_ok_for_output_1): Likewise.
2785 * expr.c (convert_modes): Likewise.
2786 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
2787 * rtlhash.c (add_rtx): Likewise.
2788 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
2789 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2790 Handle existing CONST_POLY_INT rtxes.
2791 * expmed.h (expand_shift): Take a poly_int64 instead of a
2793 * expmed.c (expand_shift): Likewise.
2794 * rtlanal.c (strip_offset): New function.
2795 (commutative_operand_precedence): Give CONST_POLY_INT the same
2796 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
2798 * rtl-tests.c (const_poly_int_tests): New struct.
2799 (rtl_tests_c_tests): Use it.
2800 * simplify-rtx.c (simplify_const_unary_operation): Handle
2802 (simplify_const_binary_operation): Likewise.
2803 (simplify_binary_operation_1): Fold additions of symbolic constants
2804 and CONST_POLY_INTs.
2805 (simplify_subreg): Handle extensions and truncations of
2807 (simplify_const_poly_int_tests): New struct.
2808 (simplify_rtx_c_tests): Use it.
2809 * wide-int.h (storage_ref): Add default constructor.
2810 (wide_int_ref_storage): Likewise.
2811 (trailing_wide_ints): Use GTY((user)).
2812 (trailing_wide_ints::operator[]): Add a const version.
2813 (trailing_wide_ints::get_precision): New function.
2814 (trailing_wide_ints::extra_size): Likewise.
2816 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2817 Alan Hayward <alan.hayward@arm.com>
2818 David Sherwood <david.sherwood@arm.com>
2820 * emit-rtl.h (gen_int_shift_amount): Declare.
2821 * emit-rtl.c (gen_int_shift_amount): New function.
2822 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
2824 * calls.c (shift_return_value): Likewise.
2825 * cse.c (fold_rtx): Likewise.
2826 * dse.c (find_shift_sequence): Likewise.
2827 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
2828 (expand_shift, expand_smod_pow2): Likewise.
2829 * lower-subreg.c (shift_cost): Likewise.
2830 * optabs.c (expand_superword_shift, expand_doubleword_mult)
2831 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
2832 (expand_vec_perm_var): Likewise.
2833 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2834 (simplify_binary_operation_1): Likewise.
2835 * combine.c (try_combine, find_split_point, force_int_to_mode)
2836 (simplify_shift_const_1, simplify_shift_const): Likewise.
2837 (change_zero_ext): Likewise. Use simplify_gen_binary.
2839 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2841 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2843 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2845 * doc/invoke.texi (ARM Options): Document accepted extension options
2846 for -march=armv8.3-a.
2848 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2851 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2852 or --with-float=softfp, set the default CPU to arm10e.
2854 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2856 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2857 * config/visium/predicates.md (const_shift_operand): Likewise.
2858 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2859 (visium_legitimize_reload_address): Likewise.
2861 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2864 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2866 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2869 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2871 2017-12-20 Julia Koval <julia.koval@intel.com>
2873 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2874 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2875 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2876 * config.gcc: Include vpclmulqdqintrin.h.
2877 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2878 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2880 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2881 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2882 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2883 * config/i386/i386.c (isa2_opts): Add -mcx16.
2884 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2885 (ix86_option_override_internal): Move mcx16 to flags2.
2886 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2887 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2888 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2889 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2890 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2891 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2892 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2893 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2894 * doc/invoke.texi: Add -mvpclmulqdq.
2896 2017-12-20 Tom de Vries <tom@codesourcery.com>
2899 * config/i386/i386.c (ix86_static_chain): Move
2900 DECL_STATIC_CHAIN test ...
2901 * calls.c (rtx_for_static_chain): ... here. New function.
2902 * calls.h (rtx_for_static_chain): Declare.
2903 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2904 instead of targetm.calls.static_chain.
2905 * df-scan.c (df_get_entry_block_def_set): Same.
2907 2017-12-19 Tom de Vries <tom@codesourcery.com>
2909 PR tree-optimization/83493
2910 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2913 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2915 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2916 inform with hardcoded english plural handling.
2918 2017-12-18 Jeff Law <law@redhat.com>
2920 PR tree-optimization/83477
2921 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2922 a non-virtual PHI, always push a new range.
2924 2017-12-19 Martin Sebor <msebor@redhat.com>
2927 * builtins.c (compute_objsize): Handle non-constant offsets.
2929 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2931 PR tree-optimization/83444
2932 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2933 character load case, if get_stridx on MEM_REF's operand doesn't
2934 look usable, retry with get_addr_stridx.
2936 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2939 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2940 (variable_tracking_main_1): Keep markers even when VTA fails.
2943 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2944 even if there are markers before it.
2945 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2947 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2949 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2950 typical order conditions.
2951 * sel-sched.c (extract_new_fences_from): Likewise.
2952 * config/visium/constraints.md (J, K, L): Likewise.
2953 * config/visium/predicates.md (const_shift_operand): Likewise.
2954 * config/visium/visium.c (visium_legitimize_address,
2955 visium_legitimize_reload_address): Likewise.
2956 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2957 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2958 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2959 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2960 SET_DUMP_DETAIL): Likewise.
2961 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2962 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2963 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2964 avr_set_core_architecture, avr_set_current_function,
2965 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2966 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2967 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2968 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2969 avr_map_decompose, avr_fold_builtin): Likewise.
2970 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2971 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2972 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2973 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2974 * config/m32c/m32c.c (m32c_conditional_register_usage,
2975 m32c_address_cost): Likewise.
2976 * config/m32c/predicates.md (shiftcount_operand,
2977 longshiftcount_operand): Likewise.
2978 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2979 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2980 can_use_cdx_ldstw): Likewise.
2981 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2982 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2984 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2985 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2986 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2987 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2988 * config/vax/vax.c (adjacent_operands_p): Likewise.
2989 * config/ft32/constraints.md (L, b, KA): Likewise.
2990 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2992 * cfgexpand.c (expand_stack_alignment): Likewise.
2993 * gcse.c (insert_expr_in_table): Likewise.
2994 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2995 * cgraphunit.c (cgraph_node::expand): Likewise.
2996 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2997 * emit-rtl.c (add_insn): Likewise.
2998 * input.c (dump_location_info): Likewise.
2999 * passes.c (NEXT_PASS): Likewise.
3000 * read-rtl-function.c (parse_note_insn_name,
3001 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
3003 * sched-rgn.c (sched_rgn_init): Likewise.
3004 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
3005 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
3006 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
3008 * explow.c (eliminate_constant_term): Likewise.
3009 * final.c (leaf_renumber_regs_insn): Likewise.
3010 * cfgrtl.c (print_rtl_with_bb): Likewise.
3011 * genhooks.c (emit_init_macros): Likewise.
3012 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
3013 * tree-data-ref.c (conflict_fn): Likewise.
3014 * selftest.c (assert_streq): Likewise.
3015 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
3016 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
3017 fold_binary_loc, multiple_of_p): Likewise.
3018 * reload.c (push_reload, find_equiv_reg): Likewise.
3019 * et-forest.c (et_nca, et_below): Likewise.
3020 * dbxout.c (dbxout_symbol_location): Likewise.
3021 * reorg.c (relax_delay_slots): Likewise.
3022 * dojump.c (do_compare_rtx_and_jump): Likewise.
3023 * gengtype-parse.c (type): Likewise.
3024 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
3025 simplify_const_relational_operation): Likewise.
3026 * reload1.c (do_output_reload): Likewise.
3027 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
3028 * gengtype.c (type_for_name): Likewise.
3029 * gimple-ssa-sprintf.c (format_directive): Likewise.
3031 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3034 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
3035 accessing it. Adjust comment.
3037 2017-12-19 Jakub Jelinek <jakub@redhat.com>
3040 * predict.c (zero_one_minusone): New function.
3041 (apply_return_prediction): Avoid return prediction for functions
3042 returning only -1, 0 and 1 values, unless they only return -1 and 0
3045 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
3047 * config/arc/arc.c (legitimate_scaled_address_p): Clean
3048 fall-through warning.
3049 (arc_compute_frame_size): Remove unused variables.
3050 (arc_print_operand): Fix fprintif format.
3051 (arc_can_follow_jump): Clean fall-through warning.
3053 2017-12-19 Marek Polacek <polacek@redhat.com>
3056 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
3059 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
3061 * config/arc/arc.c (overriderregs): New variable.
3062 (arc_override_options): Track fixed/call saved/call options.
3063 (arc_conditional_register_usage): Check against overrideregs
3064 variable whenever we change register properties.
3066 2017-12-19 Nathan Sidwell <nathan@acm.org>
3068 * opts.c (finish_options): Don't prefix dump_base_name if it
3069 already contains directories.
3071 2017-12-19 Martin Liska <mliska@suse.cz>
3073 PR rtl-optimization/82675
3074 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
3075 more element in sbitmap.
3077 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
3079 * gimplify.c (gimplify_expr): Use error_operand_p.
3081 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
3084 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
3087 2017-12-19 Daniel Cederman <cederman@gaisler.com>
3089 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
3090 the jump is to a label.
3092 2017-12-19 Jakub Jelinek <jakub@redhat.com>
3094 PR tree-optimization/83444
3095 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
3100 * ipa-inline.c (flatten_remove_node_hook): New function.
3101 (ipa_inline): Keep only nodes with flatten attribute at the end of
3102 the array in the order from ipa_reverse_postorder, only walk that
3103 portion of array for flattening, if there is more than one such
3104 node, temporarily register a removal hook and ignore removed nodes.
3106 PR tree-optimization/80631
3107 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
3108 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
3109 IFN_REDUC_MAX or IFN_REDUC_MIN.
3111 2017-12-18 Jeff Law <law@redhat.com>
3113 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
3114 record symbolic equivalences from backedges in the CFG.
3117 2017-11-19 Jeff Law <law@redhat.com>
3119 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
3120 of degenerates resulting from ignoring an edge.
3122 2017-12-18 Martin Sebor <msebor@redhat.com>
3125 PR tree-optimization/78450
3126 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
3127 (handle_builtin_strlen): Call it.
3129 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
3131 PR rtl-optimization/83424
3132 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
3134 2017-12-18 Marek Polacek <polacek@redhat.com>
3137 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
3138 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
3141 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
3143 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
3144 (umaddsidi4, umaddsidi_split): Likewise.
3146 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
3148 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
3151 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3154 * config/s390/s390.c (s390_option_override): Avoid strncpy.
3156 2017-12-18 Richard Biener <rguenther@suse.de>
3158 PR tree-optimization/81877
3159 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
3160 (outermost_indep_loop): Adjust.
3161 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
3162 (can_sm_ref_p): Adjust.
3164 2017-12-18 Richard Biener <rguenther@suse.de>
3167 * tree.c (array_at_struct_end_p): Return true if the underlying
3168 object has space for at least one element in excess of what
3169 the array domain specifies.
3171 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3173 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
3174 example to avoid overfull hbox.
3175 * doc/invoke.texi (Option Summary): Add missing @gol.
3176 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
3179 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3180 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3182 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
3183 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
3184 (AArch64 Options): Document -mverbose-cost-dump.
3185 (ARM Options): Likewise, plus -mflip-thumb.
3187 2017-12-17 Martin Sebor <msebor@redhat.com>
3190 * gimple-ssa-warn-restrict.c
3191 (builtin_memref::offset_out_of_bounds): Correct the handling of
3194 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3196 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
3198 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
3200 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
3202 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
3204 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
3207 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
3209 * doc/invoke.texi: Fix some typos.
3211 2017-12-16 Martin Sebor <msebor@redhat.com>
3213 PR tree-optimization/78918
3214 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
3215 * builtins.c (check_sizes): Rename...
3216 (check_access): ...to this. Rename function arguments for clarity.
3217 (check_memop_sizes): Adjust names.
3218 (expand_builtin_memchr, expand_builtin_memcpy): Same.
3219 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
3220 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
3221 (check_strncat_sizes, expand_builtin_strncat): Same.
3222 (expand_builtin_strncpy, expand_builtin_memset): Same.
3223 (expand_builtin_bzero, expand_builtin_memcmp): Same.
3224 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
3225 (maybe_emit_sprintf_chk_warning): Same.
3226 (expand_builtin_strcpy): Adjust.
3227 (expand_builtin_stpcpy): Same.
3228 (expand_builtin_with_bounds): Detect out-of-bounds accesses
3229 in pointer-checking forms of memcpy, memmove, and mempcpy.
3230 (gcall_to_tree_minimal, max_object_size): Define new functions.
3231 * builtins.h (max_object_size): Declare.
3232 * calls.c (alloc_max_size): Call max_object_size instead of
3233 hardcoding ssizetype limit.
3234 (get_size_range): Handle new argument.
3235 * calls.h (get_size_range): Add a new argument.
3236 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
3237 * doc/invoke.texi (-Wrestrict): Adjust, add example.
3238 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
3240 (gimple_fold_builtin_memory_chk): Same.
3241 (gimple_fold_builtin_stxcpy_chk): New function.
3242 * gimple-ssa-warn-restrict.c: New source.
3243 * gimple-ssa-warn-restrict.h: New header.
3244 * gimple.c (gimple_build_call_from_tree): Propagate location.
3245 * passes.def (pass_warn_restrict): Add new pass.
3246 * tree-pass.h (make_pass_warn_restrict): Declare.
3247 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
3249 (handle_builtin_strcat): Same.
3250 (strlen_optimize_stmt): Rename...
3251 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
3252 stpncpy, strncpy, and their checking forms.
3254 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
3256 PR rtl-optimization/82849
3257 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
3258 and get_max_loop_iterations_int.
3260 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3261 Alan Hayward <alan.hayward@arm.com>
3262 David Sherwood <david.sherwood@arm.com>
3264 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
3265 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
3266 size as a poly_uint64.
3267 (mode_for_vector, mode_for_int_vector): Take the number of vector
3268 elements as a poly_uint64.
3269 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
3270 size as a poly_uint64.
3271 (mode_for_vector, mode_for_int_vector): Take the number of vector
3272 elements as a poly_uint64.
3274 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3275 Alan Hayward <alan.hayward@arm.com>
3276 David Sherwood <david.sherwood@arm.com>
3278 * machmode.h (MACRO_MODE): New macro.
3279 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
3280 * targhooks.c (default_libcall_value, default_secondary_reload)
3281 (default_memory_move_cost, default_register_move_cost)
3282 (default_class_max_nregs): Likewise.
3284 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3285 Alan Hayward <alan.hayward@arm.com>
3286 David Sherwood <david.sherwood@arm.com>
3288 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
3289 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
3291 * genautomata.c (main): Likewise.
3292 * genconditions.c (write_header): Likewise.
3293 * genemit.c (main): Likewise.
3294 * genextract.c (print_header): Likewise.
3295 * genopinit.c (main): Likewise.
3296 * genoutput.c (output_prologue): Likewise.
3297 * genpeep.c (main): Likewise.
3298 * genpreds.c (write_insn_preds_c): Likewise.
3299 * genrecog.c (writer_header): Likewise.
3300 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
3301 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
3302 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
3303 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
3304 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
3305 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
3306 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
3307 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
3308 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
3309 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
3310 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
3311 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
3312 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
3313 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
3314 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
3315 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
3316 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
3317 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
3318 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
3319 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
3320 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
3321 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
3322 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
3323 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
3324 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
3325 * config/darwin.c (IN_TARGET_CODE): Likewise.
3326 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
3327 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
3328 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
3329 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
3330 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
3331 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
3332 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
3333 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
3334 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
3335 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
3336 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
3337 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
3338 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
3339 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
3340 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
3341 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
3342 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
3343 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
3344 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
3345 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
3346 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
3347 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
3348 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
3349 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
3350 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
3351 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
3352 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
3353 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
3354 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
3355 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
3356 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
3357 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
3358 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
3359 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
3360 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
3361 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
3362 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
3363 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
3364 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
3365 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
3366 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
3367 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
3368 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
3369 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
3370 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
3371 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
3372 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
3373 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
3374 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
3375 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
3376 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
3377 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
3378 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
3379 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
3380 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
3381 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
3382 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
3383 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
3384 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
3385 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
3386 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3387 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
3388 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
3389 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
3390 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
3391 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
3392 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
3393 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
3394 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
3395 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
3396 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
3397 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3398 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
3399 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
3400 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
3401 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
3402 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
3403 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
3404 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
3405 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
3406 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
3407 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
3408 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
3409 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
3410 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
3411 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
3412 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
3413 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
3414 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
3415 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
3416 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
3417 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
3418 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
3419 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
3420 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
3421 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
3422 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
3423 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
3424 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
3425 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
3426 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
3427 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
3428 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
3429 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
3430 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
3431 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
3433 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3434 Alan Hayward <alan.hayward@arm.com>
3435 David Sherwood <david.sherwood@arm.com>
3437 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
3440 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3441 Alan Hayward <alan.hayward@arm.com>
3442 David Sherwood <david.sherwood@arm.com>
3444 * doc/generic.texi (VEC_SERIES_EXPR): Document.
3445 * doc/md.texi (vec_series@var{m}): Document.
3446 * tree.def (VEC_SERIES_EXPR): New tree code.
3447 * tree.h (build_vec_series): Declare.
3448 * tree.c (build_vec_series): New function.
3449 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
3450 * tree-pretty-print.c (dump_generic_node): Likewise.
3451 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
3452 * tree-inline.c (estimate_operator_cost): Likewise.
3453 * expr.c (expand_expr_real_2): Likewise.
3454 * optabs-tree.c (optab_for_tree_code): Likewise.
3455 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
3456 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
3457 * expmed.c (make_tree): Handle VEC_SERIES.
3458 * optabs.def (vec_series_optab): New optab.
3459 * optabs.h (expand_vec_series_expr): Declare.
3460 * optabs.c (expand_vec_series_expr): New function.
3461 * tree-vect-generic.c (expand_vector_operations_1): Check that
3462 the operands also have vector type.
3464 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3465 Alan Hayward <alan.hawyard@arm.com>
3466 David Sherwood <david.sherwood@arm.com>
3468 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
3469 (VEC_COND_EXPR): Add missing @tindex.
3470 * doc/md.texi (vec_duplicate@var{m}): Document.
3471 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
3472 * tree.c (build_vector_from_val): Add stubbed-out handling of
3473 variable-length vectors, using VEC_DUPLICATE_EXPR.
3474 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
3475 * cfgexpand.c (expand_debug_expr): Likewise.
3476 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3477 * tree-inline.c (estimate_operator_cost): Likewise.
3478 * tree-pretty-print.c (dump_generic_node): Likewise.
3479 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
3480 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
3481 (test_vec_duplicate_folding): New function.
3482 (fold_const_c_tests): Call it.
3483 * optabs.def (vec_duplicate_optab): New optab.
3484 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
3485 * optabs.h (expand_vector_broadcast): Declare.
3486 * optabs.c (expand_vector_broadcast): Make non-static. Try using
3487 vec_duplicate_optab.
3488 * expr.c (store_constructor): Try using vec_duplicate_optab for
3490 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
3492 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
3495 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
3496 div/mod latencies a bit.
3498 2017-12-15 Jeff Law <law@redhat.com>
3500 PR tree-optimization/36550
3501 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
3502 (mark_threaded_blocks): Rewrite code to avoid block copying when
3503 optimizing for size. Don't pessimize blocks which will be
3504 copied, but all the statements will be dead.
3506 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
3508 PR tree-optimization/81165
3509 * tree-ssa-threadupdate.c (uses_in_bb): New.
3510 (estimate_threading_killed_stmts): New.
3511 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
3512 * tree-ssa-threadedge.c
3513 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
3516 2017-12-15 Jeff Law <law@redhat.com>
3518 PR tree-optimization/83410
3519 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
3520 threads when parallelizing loops.
3522 2017-12-15 Jakub Jelinek <jakub@redhat.com>
3524 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
3526 * config/alpha/alpha.c (vms_attribute_table): Swap
3527 affects_type_identity and handler fields, adjust comments.
3528 * config/mips/mips.c (mips_attribute_table): Likewise.
3529 * config/visium/visium.c (visium_attribute_table): Likewise.
3530 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
3531 * config/microblaze/microblaze.c (microblaze_attribute_table):
3533 * config/spu/spu.c (spu_attribute_table): Likewise.
3534 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
3535 * config/arc/arc.c (arc_attribute_table): Likewise.
3536 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
3537 * config/v850/v850.c (v850_handle_interrupt_attribute,
3538 v850_handle_data_area_attribute): Formatting fixes.
3539 (v850_attribute_table): Swap affects_type_identity and handler
3540 fields, adjust comments.
3541 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
3542 * config/arm/arm.c (arm_attribute_table): Likewise.
3543 * config/avr/avr.c (avr_attribute_table): Likewise.
3544 * config/s390/s390.c (s390_attribute_table): Likewise.
3545 * config/sh/sh.c (sh_attribute_table): Likewise.
3546 * config/i386/i386.c (ix86_handle_cconv_attribute,
3547 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
3548 (ix86_attribute_table): Swap affects_type_identity and handler
3549 fields, adjust comments.
3550 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3551 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
3552 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
3553 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
3554 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
3555 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3556 * config/rx/rx.c (rx_attribute_table): Likewise.
3557 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
3558 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
3559 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
3560 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
3561 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3562 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
3563 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
3564 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
3565 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3566 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3567 * doc/plugins.texi (user_attr): Likewise. Add NULL for
3569 * attribs.c (empty_attribute_table): Swap affects_type_identity and
3571 (register_scoped_attributes, decl_attributes): Formatting fixes.
3573 PR tree-optimization/83269
3574 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
3575 subtraction in arg0's type if type is signed and arg0 is unsigned.
3579 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
3580 convert? on @0 instead of convert. Check type of @1, not @0.
3581 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
3582 convert? on @0 instead of convert. Check type of @1, not @0.
3583 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
3584 only optimize if either both @1 and @2 types are narrower
3585 precision, or both are wider or equal precision, and in the former
3586 case only if both have undefined overflow.
3588 2017-12-15 Richard Biener <rguenther@suse.de>
3591 * internal-fn.def (IFN_NOP): Add.
3592 * internal-fn.c (expand_NOP): Do nothing.
3593 * lto-streamer-in.c (input_function): Instead of removing
3594 sanitizer calls replace them with IFN_NOP calls.
3596 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
3597 Alan Hayward <alan.hayward@arm.com>
3598 David Sherwood <david.sherwood@arm.com>
3600 * dse.c (store_info, read_info_type): Replace begin and end with
3602 (print_range): New function.
3603 (set_all_positions_unneeded, any_positions_needed_p)
3604 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
3606 (record_store): Likewise. Optimize the case in which all positions
3608 (get_stored_val): Replace read_begin and read_end with read_offset
3610 (replace_read): Update call accordingly.
3612 2017-12-15 Bin Cheng <bin.cheng@arm.com>
3614 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
3615 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
3616 (loop_cand::loop_cand): Initialize above members.
3617 (loop_cand::supported_operations): Delete.
3618 (loop_cand::can_interchange_p): Inline above function.
3619 (loop_cand::classify_simple_reduction): Record number of constant
3620 initialized simple reductions.
3621 (should_interchange_loops): New parameters. Check stmt cost of loops
3623 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
3624 Update call to should_interchange_loops.
3625 (should_interchange_loop_nest): Update call to
3626 should_interchange_loops.
3628 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
3631 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
3632 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
3633 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
3635 2017-12-15 Julia Koval <julia.koval@intel.com>
3637 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
3638 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
3640 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
3641 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
3642 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
3644 2017-12-15 Julia Koval <julia.koval@intel.com>
3646 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
3647 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
3648 * config/i386/sse.md (vaesenc_<mode>): New pattern.
3649 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
3650 _mm_aesenc_epi128): New intrinsics.
3652 2017-12-15 Julia Koval <julia.koval@intel.com>
3654 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
3655 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
3657 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
3658 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
3659 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
3661 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3663 * gimple-ssa-strength-reduction.c (analyze_increments):
3664 Distinguish replacement costs for constant strides from those for
3667 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3669 * var-tracking.c (variable_tracking_main_1): Formatting fix.
3671 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3673 * doc/invoke.texi: Document -Wcast-function-type.
3674 * recog.h (stored_funcptr): Change signature.
3675 * tree-dump.c (dump_node): Avoid warning.
3676 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
3678 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
3681 * gimple-fold.c (get_range_strlen): Add the handling of non-member
3684 2017-12-14 David Malcolm <dmalcolm@redhat.com>
3686 PR tree-optimization/83312
3687 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
3688 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
3689 "val" param, and to cope with arbitrary basic blocks.
3690 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
3691 handle NULL_TREE for "val", dropping "bb" param.
3692 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
3693 drop "bb" param. Handle NULL_TREE for "val".
3694 (find_case_label_for_value): Make "switch_stmt" param const.
3695 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
3697 (vrp_prop::check_all_array_refs): Reimplement as...
3698 (check_array_bounds_dom_walker::before_dom_children): ...this new
3699 vfunc. Replace linear search through BB block list, excluding
3700 those with non-executable in-edges via dominator walk.
3702 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3704 * config/arm/arm.opt (mverbose-cost-dump): New option.
3705 * config/arm/arm.c (arm_rtx_costs): Use it.
3707 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
3710 * reload1.c (emit_input_reload_insns): Skip debug markers.
3712 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
3714 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
3718 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
3719 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
3720 * config/sh/sh.c: Skip debug insns besides notes.
3721 * config/sh/sh.md: Likewise.
3722 * config/sh/sh_treg_combine.cc: Likewise.
3723 * config/sh/sync.md: Likewise.
3725 2017-12-14 Tom de Vries <tom@codesourcery.com>
3727 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3728 Add item for weak_undefined.
3730 2017-12-14 Richard Biener <rguenther@suse.de>
3732 PR tree-optimization/67842
3733 * sese.h (bb_in_region): Remove #if 0'ed code.
3735 2017-12-14 Richard Biener <rguenther@suse.de>
3737 PR tree-optimization/83326
3738 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3739 may_be_zero parameter and handle it by not marking the first
3740 peeled copy as not exiting the loop.
3741 (try_peel_loop): Likewise.
3742 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
3743 to handle the case of constant or zero iterations and perform
3744 loop header copying on-the-fly.
3746 2017-12-14 Richard Biener <rguenther@suse.de>
3748 PR tree-optimization/83418
3749 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
3750 Instead of asserting we don't get unfolded comparisons deal with them.
3752 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3755 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
3756 basic blocks. Assert debug bind insns don't appear outside of bbs,
3757 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
3760 PR tree-optimization/83198
3761 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
3762 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
3763 value if arg is a REAL_CST with incompatible type.
3765 2017-12-14 Sudakshina Das <sudi.das@arm.com>
3766 Bin Cheng <bin.cheng@arm.com>
3769 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
3771 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
3774 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3776 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
3777 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
3779 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
3780 Alan Hayward <alan.hayward@arm.com>
3781 David Sherwood <david.sherwood@arm.com>
3783 * poly-int.h: New file.
3784 * poly-int-types.h: Likewise.
3785 * coretypes.h: Include them.
3786 (POLY_INT_CONVERSION): Define.
3787 * target.def (estimated_poly_value): New hook.
3788 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
3789 * doc/tm.texi: Regenerate.
3790 * doc/poly-int.texi: New file.
3791 * doc/gccint.texi: Include it.
3792 * doc/rtl.texi: Describe restrictions on subreg modes.
3793 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
3794 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
3795 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
3796 * targhooks.h (default_estimated_poly_value): Declare.
3797 * targhooks.c (default_estimated_poly_value): New function.
3798 * target.h (estimated_poly_value): Likewise.
3799 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
3800 (wi::unary_traits): Delete.
3801 (wi::binary_traits::signed_shift_result_type): Define for
3802 offset_int << HOST_WIDE_INT, etc.
3803 (generic_wide_int::operator <<=): Define for all types and use
3804 wi::lshift instead of <<.
3805 (wi::hwi_with_prec): Add a default constructor.
3806 (wi::ints_for): New class.
3807 (operator <<): Define for all wide-int types.
3808 (operator /): New function.
3809 (operator %): Likewise.
3810 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
3811 (ASSERT_MAYBE_NE_AT): New macros.
3813 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
3814 Dominik Vogt <vogt@linux.vnet.ibm.com>
3817 * emit-rtl.c (init_emit): Remove ??? comment.
3818 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3819 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
3820 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
3821 alignment of 3 virtual registers to BITS_PER_WORD.
3823 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
3825 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
3827 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
3828 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
3830 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3833 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
3835 2017-12-13 Jakub Jelinek <jakub@redhat.com>
3837 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
3838 with the exception of debug begin stmt markers appear before
3842 * final.c (rest_of_handle_final): Call variable_tracking_main only
3843 if !flag_var_tracking.
3845 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3846 Jakub Jelinek <jakub@redhat.com>
3850 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3851 labels debug stmts that can only appear after labels.
3853 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3855 PR rtl-optimization/82398
3856 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3857 EXPR_USEFULNESS in priority comparison.
3859 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3861 PR rtl-optimization/83393
3862 * combine.c (move_deaths): If reg_stat points to a too new insn in
3863 last_death, do not use it: find the proper insn instead.
3865 2017-12-12 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3868 * config/rs6000/rs6000-string.c (expand_block_compare,
3869 expand_strn_compare): Fix set_mem_size() calls.
3871 2017-12-12 Jeff Law <law@redhat.com>
3873 PR tree-optimization/83298
3874 PR tree-optimization/83362
3875 PR tree-optimization/83383
3876 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3877 push_value_range a public interface. Add new argument to
3878 record_ranges_from_stmt.
3879 * gimple-ssa-evrp-analyze.c
3880 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3881 Update comments. Handle recording temporary equivalences.
3882 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3883 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3884 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3885 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3886 gimple-ssa-evrp-analyze.h.
3887 (record_temporary_equivalences_from_phis): Add new argument. When
3888 the PHI arg is an SSA_NAME, set the result's range to the range
3890 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3891 from statements too.
3892 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3893 Pass it down to children as needed.
3894 (thread_outgoing_edges): Likewise.
3895 (thread_across_edge): Likewise. Push/pop range state as needed.
3896 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3898 2017-12-12 Julia Koval <julia.koval@intel.com>
3900 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3901 (PTA_CANNONLAKE): Remove PTA_CLWB.
3903 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3906 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3907 (vcondv2div2df): Likewise.
3908 (vconduv2dfv2di): Likewise.
3910 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3912 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3913 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3914 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3915 ATTR_MATHFN_FPROUNDING_ERRNO.
3917 2017-12-12 Richard Biener <rguenther@suse.de>
3919 PR tree-optimization/83385
3920 * graphite-scop-detection.c (get_order, order): Remove.
3921 (bb_to_rpo): New global.
3923 (build_scops): Sort pbbs in RPO order.
3925 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3927 * combine.c (simplify_set): Do not transform subregs to zero_extends
3928 if the destination mode is a vector mode.
3930 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3932 PR tree-optimization/83359
3933 * tree-cfg.h (fold_loop_internal_call): Declare.
3934 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3935 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3936 (find_loop_dist_alias): New function.
3937 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3938 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3941 PR tree-optimization/80631
3942 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3943 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3944 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3945 hardcoding zero as the value if COND_EXPR is never true. For
3946 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3947 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3948 hardcoding MAX_EXPR as the reduction operation.
3949 (is_nonwrapping_integer_induction): Allow negative step.
3950 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3951 vect_create_epilog_for_reduction, if no value is suitable, don't
3952 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3954 2017-12-12 Richard Biener <rguenther@suse.de>
3956 PR tree-optimization/81889
3957 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3958 range info from the non-wrapping IV instead of just the range
3961 2017-12-12 Julia Koval <julia.koval@intel.com>
3963 * config.gcc: Add vaesintrin.h.
3964 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3966 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3967 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3969 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3970 * config/i386/immintrin.h: Include vaesintrin.h.
3971 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3972 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3973 _mm_aesdec_epi128): New intrinsics.
3975 2017-12-12 Julia Koval <julia.koval@intel.com>
3977 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3978 OPTION_MASK_ISA_VAES_UNSET): New.
3979 (ix86_handle_option): Handle -mvaes.
3980 * config/i386/cpuid.h: Define bit_VAES.
3981 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3982 * config/i386/i386-c.c (__VAES__): New.
3983 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3984 (ix86_valid_target_attribute_inner_p): Ditto.
3985 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3986 * config/i386/i386.opt: Add -mvaes.
3987 * doc/invoke.texi: Ditto.
3989 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3991 * debug.h (gcc_debug_hooks): Add inline_entry.
3992 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3993 * debug.c (do_nothing_debug_hooks): Likewise.
3994 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3995 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3996 (dwarf2_lineno_debug_hooks): Likewise.
3998 * common.opt (gstatement-frontiers): New, setting
3999 debug_nonbind_markers_p.
4000 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
4001 * toplev.c (process_options): Autodetect value for debug statement
4003 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
4004 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
4006 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
4007 markers. Integrate source bind into debug stmt expand loop.
4008 (pass_expand::execute): Check debug marker limit. Avoid deep
4009 TER and expand debug locations for debug bind insns only.
4010 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
4012 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
4013 * final.c (reemit_insn_block_notes): Take current block from
4014 nonbind markers. Declare note where it's first set.
4015 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
4016 begin stmt markers if enabled.
4017 (notice_source_line): Handle nonbind markers. Fail if their
4018 location is unknown or that of builtins.
4019 (rest_of_handle_final): Convert begin stmt markers to notes if
4020 var-tracking didn't run.
4021 (rest_of_clean_state): Skip begin stmt markers.
4022 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
4024 * function.c (allocate_struct_function): Set begin_stmt_markers.
4025 * function.h (struct function): Add debug_marker_count counter
4026 and debug_nonbind_markers flag.
4027 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
4028 * gimple-low.c (lower_function_body): Adjust
4029 debug_nonbind_markers.
4030 (lower_stmt): Drop or skip gimple debug stmts.
4031 (lower_try_catch): Skip debug stmts.
4032 * gimple.c (gimple_build_debug_begin_stmt): New.
4033 (gimple_copy): Increment debug_marker_count if copying one.
4034 * gimple.h (gimple_build_debug_begin_stmt): Declare.
4035 * gimplify.c (rexpr_location): New.
4036 (rexpr_has_location): New.
4037 (warn_switch_unreachable_r): Handle gimple debug stmts.
4038 (shortcut_cond_r): Call expr_location.
4040 (find_goto_label): New.
4041 (shortcut_cond_expr): Call expr_has_location, expr_location, and
4043 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
4045 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
4046 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
4047 (LANG_HOOKS_INITIALIZER): ... this.
4048 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
4049 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
4051 (update_ebb_live_info): Skip debug insn markers.
4052 * lra.c (debug_insn_static_data): Rename to...
4053 (debug_bind_static_data): ... this.
4054 (debug_marker_static_data): New.
4055 (lra_set_insn_recog_data): Select one of the above depending
4057 (lra_update_isn_regno_info): Don't assume debug insns have
4059 (push_insns): Skip debug insns.
4060 * lto-streamer-in.c (input_function): Drop debug stmts
4061 depending on active options. Adjust debug_nonbind_markers.
4062 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
4063 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
4064 begin stmt marker notes.
4065 (print_insn): Likewise.
4066 * recog.c (extract_insn): Recognize rtl for debug markers.
4067 * rtl.def (DEBUG_MARKER): New.
4068 * tree-inline.c: Include params.h.
4069 (remap_gimple_stmt): Handle nonbind markers.
4070 (maybe_move_debug_stmts_to_successors): Likewise.
4071 (copy_debug_stmt): Likewise.
4072 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
4073 markers regardless of no side effects.
4074 (tsi_link_before): Don't update container's side effects when adding
4075 a begin stmt marker.
4076 (tsi_link_after): Likewise.
4077 (expr_first): Skip begin stmt markers.
4078 (expr_last): Likewise.
4079 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
4080 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
4081 Disregard nonbind markers.
4082 * tree.c (make_node_stat): Don't set side effects for begin stmt
4084 (build1_stat): Likewise.
4085 * tree.def (DEBUG_BEGIN_STMT): New.
4086 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
4087 * var-tracking.c (delete_debug_insns): Renamed to...
4088 (delete_vta_debug_insns): ... this.
4089 (reemit_marker_as_note): New.
4090 (vt_initialize): Reemit markers.
4091 (delete_vta_debug_insns): Likewise.
4092 (vt_debug_insns_local): Reemit or delete markers.
4093 (variable_tracking_main_1): Likewise.
4094 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
4095 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
4096 (gimple_debug_nonbind_marker_p): New.
4097 (gimple_build_debug_bind): Adjust.
4098 (gimple_build_debug_begin_stmt): New.
4099 * doc/invoke.texi (max-debug-marker-count): New param.
4100 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
4101 (debug_parameter_ref, debug_marker): New.
4102 (NOTE_INSN_BEGIN_STMT): New.
4103 (DEBUG_INSN): Describe begin stmt markers.
4105 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
4106 without debug insns, we wouldn't, but clean up debug insns
4107 after a control flow insn nevertheless.
4109 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
4110 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
4111 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
4112 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
4113 (rtl_tidy_fallthru_edge): Likewise.
4114 (rtl_verify_fallthru): Likewise.
4115 (rtl_verify_bb_layout): Likewise.
4116 (skip_insns_after_block): Likewise.
4117 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
4118 * dwarf2out.c: Include print-rtl.h.
4119 (dwarf2out_next_real_insn): New.
4120 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
4121 Dump debug binds in asm comments.
4122 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
4123 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
4124 callers to use gsi_start_nondebug_bb instead.
4125 (gsi_after_labels): Skip gimple debug stmts.
4126 (gsi_start_nondebug): New.
4127 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
4128 (proper_loop_form_for_interchange): Adjust.
4129 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
4130 * gimple.h (gimple_seq_last_nondebug_stmt): New.
4131 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
4132 (collect_fallthrough_labels): Likewise.
4133 (should_warn_for_implicit_fallthrough): Likewise.
4134 (warn_implicit_fallthrough_r): Likewise.
4135 (expand_FALLTHROUGH_r): Likewise.
4136 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
4137 (graphite_copy_stmts_from_block): Skip nonbind markers.
4138 * haifa-sched.c (sched_extend_bb): Skip debug insns.
4139 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
4140 * jump.c (clean_barriers): Skip debug insns.
4141 * omp-expand.c (expand_parallel_call): Skip debug insns.
4142 (expand_task_call): Likewise.
4143 (remove_exit_barrier): Likewise.
4144 (expand_omp_taskreg): Likewise.
4145 (expand_omp_for_init_counts): Likewise.
4146 (expand_omp_for_generic): Likewise.
4147 (expand_omp_for_static_nochunk): Likewise.
4148 (expand_omp_for_static_chunk): Likewise.
4149 (expand_omp_simd): Likewise.
4150 (expand_omp_taskloop_for_outer): Likewise.
4151 (expand_omp_taskloop_for_inner): Likewise.
4152 (expand_oacc_for): Likewise.
4153 (expand_omp_sections): Likewise.
4154 (expand_omp_single): Likewise.
4155 (expand_omp_synch): Likewise.
4156 (expand_omp_atomic_load): Likewise.
4157 (expand_omp_atomic_store): Likewise.
4158 (expand_omp_atomic_fetch_op): Likewise.
4159 (expand_omp_atomic_pipeline): Likewise.
4160 (expand_omp_atomic_mutex): Likewise.
4161 (expand_omp_target): Likewise.
4162 (grid_expand_omp_for_loop): Likewise.
4163 (grid_expand_target_grid_body): Likewise.
4164 (build_omp_regions_1): Likewise.
4165 * omp-low.c (check_combined_parallel): Skip debug stmts.
4166 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
4167 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
4168 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
4170 * tree-cfg.c (make_blobs_1): Skip debug stmts.
4171 (make_edges): Likewise.
4172 (cleanup_dead_labels): Likewise.
4173 (gimple_can_merge_blocks_p): Likewise.
4174 (stmt_starts_bb_p): Likewise.
4175 (gimple_block_label): Likewise.
4176 (gimple_redirect_edge_and_branch): Likewise.
4177 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
4179 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
4181 * tree-pretty-print (print_declaration): Omit initializer in slim
4183 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
4185 (eliminate_unnecessary_stmts): Stabilize block removal order.
4186 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
4187 * var-tracking.c (get_first_insn): New.
4188 (vt_emit_notes): Call it.
4189 (vt_initialize): Walk any insns before the first BB.
4190 (delete_debug_insns): Likewise.
4192 * gimple.h (enum gimple_debug_subcode): Add
4193 GIMPLE_DEBUG_BEGIN_STMT.
4194 (gimple_debug_begin_stmt_p): New.
4195 (gimple_debug_nonbind_marker_p): New.
4196 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
4197 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
4198 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
4199 * insn-notes.def (BEGIN_STMT): New.
4200 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
4201 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
4202 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
4203 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
4204 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
4205 (INSN_DEBUG_MARKER_KIND): New.
4206 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
4207 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
4208 (INSN_VAR_LOCATION_PTR): New.
4209 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
4211 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
4212 deps only if debug bind insns are enabled.
4213 (pass_expand::execute): Avoid deep TER and expand
4214 debug locations for debug bind insns only.
4215 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
4216 debug stmts special handling down to debug bind stmts.
4217 * combine.c (try_combine): Narrow debug insns special handling
4218 down to debug bind insns.
4219 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
4220 Narrow debug insns preexisting special handling down to debug
4222 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
4223 handling down to debug bind insns.
4224 * function.c (instantiate_virtual_regs): Skip debug markers,
4225 adjust handling of debug binds.
4226 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
4227 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
4228 * haifa-sched.c (schedule_insn): Narrow special handling of debug
4229 insns to debug bind insns.
4230 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
4231 special handling of debug stmts to debug bind stmts.
4232 * ipa-split.c (split_function): Likewise.
4233 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
4234 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
4236 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
4237 * regrename.c (build_def_use): Likewise.
4238 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
4239 (pass_cprop_hardreg): Narrow special casing of debug insns to
4241 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
4242 * reload1.c (reload): Likewise.
4243 * sese.c (sese_insert_phis_for_liveouts): Narrow special
4244 casing of debug stmts to debug bind stmts.
4245 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
4246 * ssa-iterators.h (num_imm_uses): Likewise.
4247 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
4248 debug stmts to debug bind stmts.
4249 * tree-inline.c (tree_function_versioning): Narrow special casing
4250 of debug stmts to debug bind stmts.
4251 * tree-loop-distribution.c (generate_loops_for_partition):
4252 Narrow special casing of debug stmts to debug bind stmts.
4253 * tree-sra.c (analyze_access_subtree): Narrow special casing
4254 of debug stmts to debug bind stmts.
4255 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
4256 stmts to debug bind stmts.
4257 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
4258 casing of debug stmts to debug bind stmts.
4259 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
4260 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
4261 casing of debug stmts to debug bind stmts.
4262 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
4264 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
4265 debug stmts to debug bind stmts.
4266 (gimple_replace_ssa_lhs): Likewise.
4267 (insert_debug_temp_for_var_def): Likewise.
4268 (insert_debug_temps_for_defs): Likewise.
4269 (reset_debug_uses): Likewise.
4270 * tree-ssanames.c (release_ssa_name_fn): Likewise.
4271 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
4272 (adjust_debug_stmts): Likewise.
4273 (adjust_phi_and_debug_stmts): Likewise.
4274 (vect_do_peeling): Likewise.
4275 * tree-vect-loop.c (vect_transform_loop): Likewise.
4276 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
4277 * var-tracking.c (adjust_mems): Narrow special casing of debug
4278 insns to debug bind insns.
4279 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
4280 (compute_bb_dataflow, vt_find_locations): Likewise.
4281 (vt_expand_loc, emit_notes_for_changes): Likewise.
4282 (vt_init_cfa_base): Likewise.
4283 (vt_emit_notes): Likewise.
4284 (vt_initialize): Likewise.
4285 (vt_finalize): Likewise.
4287 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
4288 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
4289 (next_nonnote_nondebug_insn_bb): New.
4290 (prev_nonnote_nondebug_insn_bb): New.
4291 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
4292 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
4293 (prev_nonnote_nondebug_insn_bb): Declare.
4294 (next_nonnote_nondebug_insn_bb): Declare.
4295 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
4296 * cfgrtl.c (get_last_bb_insn): Likewise.
4297 * lra.c (push_insns): Likewise.
4299 2017-12-11 David Malcolm <dmalcolm@redhat.com>
4302 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
4304 * selftest.c (selftest::test_runner::test_runner): New ctor.
4305 (selftest::test_runner::~test_runner): New dtor.
4306 * selftest.h (class selftest::test_runner): New class.
4308 2017-12-11 Carl Love <cel@us.ibm.com>
4310 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
4311 vec_extract_fp32_from_shortl]): Add #defines.
4312 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
4313 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
4314 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
4315 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
4316 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
4317 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
4318 * doc/extend.texi: Add documentation for the added builtins.
4320 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
4322 PR rtl-optimization/80693
4323 PR rtl-optimization/81019
4324 PR rtl-optimization/81020
4325 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
4326 are not mentioned in i3. Place the REG_UNUSED note on i2,
4327 possibly modified to REG_DEAD, if it did not originate in i3.
4329 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4331 * recog.c (store_data_bypass_p_1): New function.
4332 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
4333 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
4335 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
4337 PR rtl-optimization/83361
4338 * ifcvt.c (if_convert): Call fixup_partitions.
4340 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
4342 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4343 early folding of splat_u{8,16,32}.
4345 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4347 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
4348 output_operand_lossage first argument with capital letter.
4349 (aarch64_override_options): Don't start error and sorry first argument
4350 with capital letter.
4352 2017-12-11 Andi Kleen <ak@linux.intel.com>
4354 PR gcov-profile/83355
4355 * auto-profile.c (string_table::get_index_by_decl): Don't
4356 recurse when abstract origin points to itself.
4358 2017-12-11 Bin Cheng <bin.cheng@arm.com>
4360 PR tree-optimization/83320
4361 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
4362 (prune_datarefs_not_in_loop): Ditto.
4364 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
4366 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
4368 2017-12-10 Jakub Jelinek <jakub@redhat.com>
4370 PR tree-optimization/83337
4371 * gimple-loop-interchange.cc (compute_access_stride): Handle
4372 bitfield DRs properly.
4374 2017-12-09 Jakub Jelinek <jakub@redhat.com>
4376 PR tree-optimization/83338
4377 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
4378 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
4379 vector of unsigned integers to vector of signed integers.
4381 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
4383 PR rtl-optimization/83317
4384 * lra-constraints.c (process_address_1): Add insn code check.
4386 2017-12-08 Michael Matz <matz@suse.de>
4388 Fix PR tree-optimization/83323
4389 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
4390 head-controlled loops and loop BBs.
4391 * common.opt (funroll-and-jam): Remove, instead ...
4392 (floop-unroll-and-jam): ... reuse this option.
4393 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
4394 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
4395 (-floop-unroll-and-jam): ... this option.
4397 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4399 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
4400 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
4401 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
4403 2017-12-08 Richard Biener <rguenther@suse.de>
4406 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
4407 handle accesses outside of zero-sized vars.
4409 2017-12-08 Martin Jambor <mjambor@suse.cz>
4411 PR tree-optimization/83141
4412 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
4413 test for MEM_REFs implicitely changing types with padding. Remove
4415 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
4417 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4419 * config/arc/arc.c (arc_attribute_table): Add exclusions to
4421 * config/avr/avr.c (avr_attribute_table): Likewise.
4422 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
4423 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
4424 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
4425 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
4426 of struct attribute_spec.
4427 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
4429 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
4432 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
4434 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
4437 * profile-count.c (profile_count::from_gcov_type): Move from
4438 profile-count.h; handle overflow.
4439 * profile-count.h (profile_count::from_gcov_type): Move offline.
4441 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
4443 PR rtl-optimization/83304
4444 * combine.c (move_deaths): If we do not know where a register died,
4447 2017-12-08 Richard Biener <rguenther@suse.de>
4449 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
4450 Provide -fopt-info-loop feedback when we interchange in a nest.
4452 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4454 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
4455 for armv6 ARM CPU IDs.
4457 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4459 * common/config/arm/arm-common.c: Include <algorithm>.
4460 (INCLUDE_VECTOR): Define.
4461 (compare_opt_names): New function.
4462 (arm_rewrite_selected_arch): Only strip out extensions that can be
4463 expressed through -mfpu. Sort the remaining extensions
4466 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4468 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
4469 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
4470 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
4471 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
4473 2017-06-08 Tristan Gingold <gindold@adacore.com>
4476 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
4477 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
4479 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4481 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
4482 of struct attribute_spec.
4484 2017-12-08 Julia Koval <julia.koval@intel.com>
4486 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
4487 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
4488 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
4489 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
4490 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
4491 _mm_maskz_dpwssds_epi32): Ditto.
4493 2017-12-08 Richard Biener <rguenther@suse.de>
4495 PR tree-optimization/81303
4496 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
4497 conditions try to create a comparison vector type matching
4498 the data vector type.
4499 (vectorizable_condition): Adjust.
4500 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
4501 Leave invariant conditions alone in case we can vectorize those.
4503 2017-12-08 Julia Koval <julia.koval@intel.com>
4505 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
4506 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
4507 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
4508 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
4509 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
4511 2017-12-08 Julia Koval <julia.koval@intel.com>
4513 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
4514 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
4515 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
4516 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
4517 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
4518 _mm_maskz_dpbusds_epi32): New intrinsics.
4520 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4522 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
4524 (ld<bh>io_signed): Likewise.
4525 (st<bhw>io): Likewise.
4526 * config/nios2/predicates.md (ldstio_memory_operand): Allow
4527 SMALL_INT12 constant integer operand.
4529 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4531 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
4533 (nios2_large_constant_allowed): ...this. Adjust uses.
4534 (nios2_plus_symbolic_constant_p): Rename to...
4535 (nios2_plus_large_constant_p): ...this. Adjust uses.
4536 (nios2_legitimate_address_p): Correct CONST_INT handling.
4537 (nios2_symbolic_memory_operand_p): Rename to...
4538 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
4539 (nios2_large_constant_p): Check for large constant integers too.
4540 (nios2_split_large_constant): Handle constant integers.
4541 (nios2_split_symbolic_memory_operand): Rename to...
4542 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
4543 (nios2_legitimize_constant_address): Handle constant integers.
4544 (r0rel_constant_p): Handle small constant integers.
4545 (nios2_print_operand_address): Handle r0-relative integer addresses.
4546 * config/nios2/nios2-protos.h: Adjust for renamed functions.
4547 * config/nios2/nios2.md: Adjust for renamed functions.
4549 2017-12-07 Andrew Waterman <andrew@sifive.com>
4551 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
4552 (TARGET_HAVE_SRODATA_SECTION): New define.
4553 (riscv_select_section): New function.
4555 2017-12-08 Joseph Myers <joseph@codesourcery.com>
4556 Alexander Monakov <amonakov@ispras.ru>
4557 Jakub Jelinek <jakub@redhat.com>
4560 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
4562 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4564 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
4565 a bool success value. Don't call output_operand_lossage here.
4566 (aarch64_print_ldpstp_address): Return a bool success value.
4567 (aarch64_print_operand_address): Call output_addr_const if
4568 aarch64_print_address_internal fails.
4569 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
4570 'y'; call output_operand_lossage instead. Call output_operand_lossage
4571 if aarch64_print_ldpstp_address fails.
4573 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4575 * tree-vector-builder.h
4576 (tree_vector_builder::binary_encoded_nelts): Declare.
4577 * tree-vector-builder.c
4578 (tree_vector_builder::binary_encoded_nelts): New function.
4579 * fold-const.c (negate_expr_p): Likewise.
4580 (operand_equal_p, fold_checksum_tree): Likewise.
4581 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
4582 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
4583 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
4584 (uniform_vector_p): Likewise.
4585 * varasm.c (const_hash_1, compare_constant): Likewise.
4586 * tree-ssa-ccp.c: Include tree-vector-builder.h.
4587 (valid_lattice_transition): Operate directly on the VECTOR_CST
4589 * ipa-icf.c: Include tree-vector-builder.h.
4590 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
4591 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
4593 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4595 * tree.c (build_vector): Delete.
4596 * tree.h (build_vector): Make static and move into the self-testing
4599 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4601 * vector-builder.h (vector_builder::derived): New const overload.
4602 (vector_builder::elt): New function.
4603 * tree-vector-builder.h (tree_vector_builder::type): New function.
4604 (tree_vector_builder::apply_step): Declare.
4605 * tree-vector-builder.c (tree_vector_builder::apply_step): New
4607 * gimple-fold.h (tree_vector_builder): Declare.
4608 (gimple_build_vector): Take a tree_vector_builder instead of a
4609 type and vector of elements.
4610 * gimple-fold.c (gimple_build_vector): Likewise.
4611 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
4613 (get_initial_defs_for_reduction): Likewise.
4614 (vectorizable_induction): Likewise.
4616 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4618 * tree-vector-builder.h
4619 (tree_vector_builder::new_binary_operation): Declare.
4620 * tree-vector-builder.c
4621 (tree_vector_builder::new_binary_operation): New function.
4622 * fold-const.c (fold_relational_const): Use it.
4623 (const_binop): Likewise. Check that both input vectors have
4624 the same number of elements, thus excluding things like WIDEN_SUM.
4625 Check whether it is possible to operate directly on the encodings
4628 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4630 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
4631 new_unary_operation, operating only on the encoded elements.
4632 (const_unop): Likewise.
4633 (exact_inverse): Likewise.
4634 (distributes_over_addition_p): New function.
4635 (const_binop): Use tree_vector_builder and new_unary_operation
4636 for combinations of VECTOR_CST and INTEGER_CST. Operate only
4637 on the encoded elements unless the encoding is strided and the
4638 operation does not distribute over addition.
4639 (fold_convert_const): Use tree_vector_builder and
4640 new_unary_operation. Operate only on the encoded elements
4641 for truncating integer conversions, or for non-stepped encodings.
4643 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4645 * config/sparc/sparc.c: Include tree-vector-builder.h.
4646 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
4647 * expmed.c: Include tree-vector-builder.h.
4648 (make_tree): Use tree_vector_builder instead of build_vector.
4649 * fold-const.c: Include tree-vector-builder.h.
4650 (const_binop): Use tree_vector_builder instead of build_vector.
4651 (const_unop): Likewise.
4652 (native_interpret_vector): Likewise.
4653 (fold_vec_perm): Likewise.
4654 (fold_ternary_loc): Likewise.
4655 * gimple-fold.c: Include tree-vector-builder.h.
4656 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
4658 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
4659 (simplify_vector_constructor): Use tree_vector_builder instead
4661 * tree-vect-generic.c: Include tree-vector-builder.h.
4662 (add_rshift): Use tree_vector_builder instead of build_vector.
4663 (expand_vector_divmod): Likewise.
4664 (optimize_vector_constructor): Likewise.
4665 * tree-vect-loop.c: Include tree-vector-builder.h.
4666 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
4667 of build_vector. Explicitly use a stepped encoding for
4669 * tree-vect-slp.c: Include tree-vector-builder.h.
4670 (vect_get_constant_vectors): Use tree_vector_builder instead
4672 (vect_transform_slp_perm_load): Likewise.
4673 (vect_schedule_slp_instance): Likewise.
4674 * tree-vect-stmts.c: Include tree-vector-builder.h.
4675 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
4676 (vect_gen_perm_mask_any): Likewise.
4677 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
4678 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
4680 (build_vector_from_val): Likewise. Explicitly use a duplicate
4683 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
4685 * doc/generic.texi (VECTOR_CST): Describe new representation of
4687 * vector-builder.h: New file.
4688 * tree-vector-builder.h: Likewise.
4689 * tree-vector-builder.c: Likewise.
4690 * Makefile.in (OBJS): Add tree-vector-builder.o.
4691 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
4692 * tree-core.h (tree_base): Add a vector_cst field to the u union.
4693 (tree_vector): Change the number of elements to
4694 vector_cst_encoded_nelts.
4695 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
4696 (VECTOR_CST_ELTS): Delete.
4697 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
4698 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
4699 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
4700 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
4701 (VECTOR_CST_ENCODED_ELT): Likewise.
4702 (vector_cst_encoded_nelts): New function.
4703 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4704 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4705 (vector_cst_int_elt, vector_cst_elt): Declare.
4706 * tree.c: Include tree-vector-builder.h.
4707 (tree_code_size): Abort if passed VECTOR_CST.
4708 (tree_size): Update for new VECTOR_CST layout.
4709 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4710 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4711 (build_vector): Use tree_vector_builder.
4712 (vector_cst_int_elt, vector_cst_elt): New functions.
4713 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
4714 encoded elements and then create the vector in the canonical form.
4715 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
4716 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
4717 (tree_c_tests): Call test_vector_cst_patterns.
4718 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
4720 (hash_tree): Likewise.
4721 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
4722 (streamer_write_tree_header): Likewise.
4723 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
4724 (streamer_alloc_tree): Likewise. Update call to make_vector.
4725 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
4727 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4729 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
4730 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
4731 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
4733 2017-12-07 Bin Cheng <bin.cheng@arm.com>
4734 Richard Biener <rguenther@suse.de>
4736 PR tree-optimization/81303
4737 * Makefile.in (gimple-loop-interchange.o): New object file.
4738 * common.opt (floop-interchange): Reuse the option from graphite.
4739 * doc/invoke.texi (-floop-interchange): Ditto. New document for
4740 -floop-interchange and mention it for -O3.
4741 * opts.c (default_options_table): Enable -floop-interchange at -O3.
4742 * gimple-loop-interchange.cc: New file.
4743 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
4744 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
4745 * passes.def (pass_linterchange): New pass.
4746 * timevar.def (TV_LINTERCHANGE): New time var.
4747 * tree-pass.h (make_pass_linterchange): New declaration.
4748 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
4749 interchange. Record IV before/after increment in new parameters.
4750 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
4751 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
4753 (check_reduction_path): ...New function here.
4754 * tree-vectorizer.h (check_reduction_path): New declaration.
4756 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
4759 PR rtl-optimization/80818
4760 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
4761 always early clobbered.
4762 * lra-lives.c (process_bb_lives): Check input hard regs for early
4763 clobbered non-operand hard reg.
4765 2017-12-07 Jakub Jelinek <jakub@redhat.com>
4768 * tree-cfg.c (verify_gimple_assign_binary): Don't require
4769 types_compatible_p, just that TYPE_MODE is the same.
4771 2017-12-07 Martin Sebor <msebor@redhat.com>
4774 * attribs.c (empty_attribute_table): Initialize new member of
4775 struct attribute_spec.
4776 (decl_attributes): Add argument. Handle mutually exclusive
4777 combinations of attributes.
4778 (selftests::test_attribute_exclusions): New function.
4779 (selftests::attribute_c_tests): Ditto.
4780 * attribs.h (decl_attributes): Add default argument.
4781 * selftest.h (attribute_c_tests): Declare.
4782 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
4783 * tree-core.h (attribute_spec::exclusions, exclude): New type and
4785 * doc/extend.texi (Common Function Attributes): Update const and pure.
4786 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
4787 of struct attribute_spec.
4788 * config/arc/arc.c (arc_attribute_table): Same.
4789 * config/arm/arm.c (arm_attribute_table): Same.
4790 * config/avr/avr.c ( avr_attribute_table): Same.
4791 * config/bfin/bfin.c (bfin_attribute_table): Same.
4792 * config/cr16/cr16.c (cr16_attribute_table): Same.
4793 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
4794 * config/h8300/h8300.c (h8300_attribute_table): Same.
4795 * config/i386/i386.c (ix86_attribute_table): Same.
4796 * config/ia64/ia64.c (ia64_attribute_table): Same.
4797 * config/m32c/m32c.c (m32c_attribute_table): Same.
4798 * config/m32r/m32r.c (m32r_attribute_table): Same.
4799 * config/m68k/m68k.c (m68k_attribute_table): Same.
4800 * config/mcore/mcore.c (mcore_attribute_table): Same.
4801 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
4802 * config/mips/mips.c (mips_attribute_table): Same.
4803 * config/msp430/msp430.c (msp430_attribute_table): Same.
4804 * config/nds32/nds32.c (nds32_attribute_table): Same.
4805 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
4806 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
4807 * config/rl78/rl78.c (rl78__attribute_table): Same.
4808 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
4809 * onfig/rx/rx.c (rx_attribute_table): Same.
4810 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
4811 * config/sh/sh.c (sh_attribute_table): Same.
4812 * config/sparc/sparc.c (sparc_attribute_table): Same.
4813 * config/spu/spu.c (spu_attribute_table): Same.
4814 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
4815 * config/v850/v850.c (v850_attribute_table): Same.
4816 * config/visium/visium.c (visium_attribute_table): Same.
4818 2017-12-07 Tamar Christina <tamar.christina@arm.com>
4821 * config/arm/arm.c (INCLUDE_STRING): Define.
4822 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
4823 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
4826 2017-12-07 Michael Matz <matz@suse.de>
4828 Add unroll and jam pass
4830 * gimple-loop-jam.c: New file.
4831 * Makefile.in (OBJS): Add gimple-loop-jam.o.
4832 * common.opt (funroll-and-jam): New option.
4833 * opts.c (default_options_table): Add unroll-and-jam at -O3.
4834 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
4835 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
4836 * passes.def: Add pass_loop_jam.
4837 * timevar.def (TV_LOOP_JAM): Add.
4838 * tree-pass.h (make_pass_loop_jam): Declare.
4839 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
4840 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
4841 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
4842 to flow_loop_tree_node_add.
4843 (duplicate_subloops, copy_loops_to): Append to sibling list.
4844 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
4845 * doc/invoke.texi (-funroll-and-jam): Document new option.
4846 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
4848 2017-12-07 Richard Biener <rguenther@suse.de>
4850 PR tree-optimization/83296
4851 PR tree-optimization/67769
4852 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4853 flow sensitive info in an unrelated BB.
4854 (value_replacement): Use reset_flow_sensitive_info.
4855 (minmax_replacement): Reset flow sensitive info on the def
4856 we move. Do not reset flow sensitive info in the whole BB
4857 we move the stmt to.
4858 (abs_replacement): Likewise.
4860 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4863 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4864 rs6000_cpu to the given -mcpu=, or to the default processor.
4866 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4868 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4869 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4870 instead of rs6000_cpu_attr.
4871 (rs6000_adjust_cost): Ditto.
4872 (is_microcoded_insn): Ditto.
4873 (rs6000_adjust_priority): Ditto.
4874 (rs6000_issue_rate): Ditto.
4875 (rs6000_use_sched_lookahead): Ditto.
4876 (rs6000_use_sched_lookahead_guard): Ditto.
4877 (rs6000_sched_reorder): Ditto.
4878 (force_new_group): Ditto.
4879 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4880 (group_ending_nop): Ditto.
4882 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4884 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4885 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4886 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4887 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4889 (rs6000_reassociation_width): Ditto.
4890 (rs6000_emit_epilogue): Ditto.
4891 (rs6000_adjust_cost): Ditto.
4892 (is_microcoded_insn): Ditto.
4893 (is_cracked_insn): Ditto.
4894 (rs6000_adjust_priority): Ditto.
4895 (rs6000_sched_reorder): Ditto.
4896 (rs6000_sched_reorder2): Ditto.
4897 (insn_must_be_first_in_group): Ditto.
4898 (insn_must_be_last_in_group): Ditto.
4899 (rs6000_register_move_cost): Ditto.
4900 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4903 2017-12-07 Julia Koval <julia.koval@intel.com>
4905 * config.gcc: Add vaesintrin.h.
4906 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4908 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4909 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4911 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4912 * config/i386/immintrin.h: Include vaesintrin.h.
4913 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4914 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4915 _mm_aesdec_epi128): New intrinsics.
4917 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4919 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4920 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4921 c-family/c-spellcheck.cc.
4922 (best_macro_match::best_macro_match): Likewise.
4924 (struct edit_distance_traits<cpp_hashnode *>): Move to
4925 c-family/c-spellcheck.h.
4926 (class best_macro_match): Likewise.
4928 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4930 PR tree-optimization/83293
4931 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4932 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4933 might insert into empty bb.
4936 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4937 simplify for plus with :c added, and pointer_plus without that.
4938 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4939 with undefined overflow and the conversion is not widening,
4940 perform negation in utype and only convert to type afterwards.
4941 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4942 simplify for plus with :c added, and pointer_plus without that.
4943 If type is integral with undefined overflow and the conversion is
4944 not widening, perform minus in utype and only convert to type
4945 afterwards. Move the last pointer_diff_expr simplify into the
4948 2017-12-06 Martin Sebor <msebor@redhat.com>
4950 PR tree-optimization/82646
4951 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4952 strncpy, not maxlen.
4954 2017-12-06 Martin Sebor <msebor@redhat.com>
4956 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4959 PR tree-optimization/83075
4960 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4961 strncat/strncpy don't change length of source string.
4963 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4966 2017-11-29 Martin Aberg <maberg@gaisler.com>
4968 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4969 to prevent b2bst errata sequence.
4970 (sqrtdf2_fix): Likewise.
4972 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4974 PR tree-optimization/81945
4975 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4976 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4977 to dest_cfun has orig_loop_num set, either remap it to the new
4978 loop number if the loop got moved too, or clear it.
4980 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4982 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4984 (thunderx2t99_nothing): New insn reservation.
4985 (thunderx2t99_mrs): New insn reservation.
4986 (thunderx2t99_multiple): New insn reservation.
4987 (thunderx2t99_alu_basi): Add bfx to reservation.
4988 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4990 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4993 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4995 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4997 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4998 tree-ssa-pre.c::remove_dead_inserted_code.
4999 * tree-ssa-dce.h: New file.
5000 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
5001 (remove_dead_inserted_code): Move and rename to function
5002 tree-ssa-dce.c::simple_dce_from_worklist.
5003 (pass_pre::execute): Update use.
5005 2017-12-05 Richard Biener <rguenther@suse.de>
5007 PR tree-optimization/83277
5008 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
5009 to code-gen liveout vars.
5011 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
5013 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
5014 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
5015 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
5018 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
5020 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
5022 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
5023 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
5026 2017-12-05 Richard Biener <rguenther@suse.de>
5028 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
5029 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
5030 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
5031 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
5033 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5035 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
5037 2017-12-05 Martin Liska <mliska@suse.cz>
5038 Jakub Jelinek <jakub@redhat.com>
5040 * doc/invoke.texi: Document the options.
5041 * flag-types.h (enum sanitize_code): Add
5042 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
5043 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
5044 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
5045 * opts.c: Define new sanitizer options.
5046 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
5047 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
5049 2017-12-05 Julia Koval <julia.koval@intel.com>
5051 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
5052 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
5053 (ix86_handle_option): Handle -mavx512vnni.
5054 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
5055 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
5056 * config/i386/i386-c (__AVX512VNNI__): New.
5057 * config/i386/i386.c (ix86_target_string): Handle new option.
5058 (ix86_valid_target_attribute_inner_p): Handle new option.
5059 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
5060 * config/i386/i386.opt (mavx512vnni): New option.
5062 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
5065 * athlon.md: Disable for generic.
5066 * haswell.md: Enable for generic.
5067 * i386.c (ix86_sched_init_global): Add core hooks for generic.
5068 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
5070 (ix86_adjust_cost): Move generic to haswell path.
5072 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
5074 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
5075 instead of MEM_P in a couple more places. Fix formatting issues.
5077 2017-12-04 Jim Wilson <jimw@sifive.com>
5079 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
5080 instead of GP_REG_LAST-1.
5081 (riscv_adjust_libcall_cfi_prologue): Likewise.
5082 (riscv_adjust_libcall_cri_epilogue): Likewise.
5083 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
5086 2017-12-04 Luis Machado <luis.machado@linaro.org>
5088 * ipa-pure-const.c (check_decl): Add missing newline.
5089 (state_from_flags): Likewise.
5091 2017-12-04 Jeff Law <law@redhat.com>
5093 PR tree-optimizatin/78496
5094 * gimple-ssa-evrp-analyze.h
5095 (evrp_range_analyzer::get_vr_values): Simplify.
5096 * gimple-ssa-evrp-analyze.c: Corresponding changes.
5097 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
5098 and gimple-ssa-evrp-analyze.h.
5099 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
5100 (simplify_stmt_for_jump_threading): Copy a blob of code from
5101 tree-vrp.c to use ranges to simplify statements.
5102 (dom_opt_dom_walker::before_dom_children): Call
5103 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
5104 (dom_opt_dom_walker::after_dom_children): Similarly for
5105 evrp_range_analyzer::leave.
5106 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
5109 * gimple-ssa-evrp-analyze.c
5110 (evrp_range_analyzer::extract_range_from_stmt): Always use
5111 vr_values::update_value_range so preexisting range info is
5112 medged with new range info, even if the new range is VR_VARYING.
5114 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
5116 * combine.c: Adjust comment.
5117 (use_crosses_set_p): Delete.
5118 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
5119 (try_combine): Ditto.
5121 2017-12-04 Richard Biener <rguenther@suse.de>
5123 PR tree-optimization/83255
5124 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
5125 Re-add zero-iteration check.
5127 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
5129 PR rtl-optimization/83245
5130 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
5131 hard registers as earlyclobber, also if not in an asm.
5133 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
5137 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5140 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
5141 rs6000_cpu based on cpu_index, not tune_index.
5143 2017-12-04 Richard Biener <rguenther@suse.de>
5145 PR tree-optimization/83238
5146 * graphite-scop-detection.c (scop_detection::merge_sese): Make
5147 code match comment, rejecting invalid SESE regions.
5149 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
5151 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
5152 require base operand is a REG_POINTER prior to reload on targets
5153 with non-equivalent space registers.
5155 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
5157 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
5158 (update_specialized_profile): Fix updating of counts.
5159 (perhaps_add_new_callers): Likewise.
5161 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
5164 * x86-tune.def: Remove obsolette FIXMEs.
5165 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
5166 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
5167 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
5169 (X86_TUNE_PAD_RETURNS): Disable for generic.
5171 2017-12-02 Jakub Jelinek <jakub@redhat.com>
5173 PR tree-optimization/83170
5174 PR tree-optimization/83241
5175 * gimple-ssa-store-merging.c
5176 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
5177 gimple_vuse (ins_stmt) in case it has changed.
5178 (imm_store_chain_info::output_merged_store): Likewise.
5180 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
5184 * tree-cfg.c (pass_warn_function_return::execute): Handle
5185 __builtin_ubsan_handle_missing_return like __builtin_unreachable
5186 with BUILTINS_LOCATION.
5190 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
5191 is BLKmode for vector field with vector raw mode, use TYPE_MODE
5192 instead of DECL_MODE.
5194 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
5195 last argument to rtx pointer.
5196 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
5197 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
5198 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
5199 depending on the chosen ISAs.
5200 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
5201 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
5202 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
5204 * config/i386/sse.md (mov<mode>_internal): Likewise.
5205 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
5207 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5209 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
5211 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5213 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
5214 parameters from prototype.
5215 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
5216 parameters. Don't print a comment.
5217 (emit_fusion_gpr_load): Adjust.
5218 (emit_fusion_load_store): Adjust.
5219 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
5220 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
5221 comment on the second line.
5223 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5226 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
5227 rs6000_cpu based on cpu_index, not tune_index.
5229 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5231 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
5232 which_alternative instead of which_alternative + 1.
5233 (output_asm_insn): Print an extra tab if the template is short.
5235 2017-12-01 Jim Wilson <jimw@sifive.com>
5237 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
5239 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
5240 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
5242 * doc/tm.texi: Regenerate.
5244 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
5247 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
5248 whether we can allocate pseudos before trying to fix an address.
5249 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
5250 memory address is indexed or indirect.
5251 (floatuns_<mode>si2_hw2): Likewise.
5253 2017-12-01 Jason Merrill <jason@redhat.com>
5255 * Makefile.in (TAGS): Add c-family/*.cc.
5257 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
5259 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
5260 (call_value_insn): Likewise.
5261 (sibcall_insn): Likewise.
5262 (sibcall_value_insn): Likewise.
5263 (movsi_aarch64): Likewise.
5264 (movdi_aarch64): Likewise.
5265 (add_losym_): Likewise.
5266 (ldr_got_small_): Likewise.
5267 (ldr_got_small_sidi): Likewise.
5268 (ldr_got_small_28k_): Likewise.
5269 (ldr_got_small_28k_sidi): Likewise.
5270 * config/aarch64/aarch64.c (aarch64_print_address_internal):
5271 Move output_addr_const to symbolic case. Add error check.
5273 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5275 * config/s390/predicates.md (plus16_Q_operand): New predicate.
5276 * config/s390/s390.md: Disable MVC merging peephole if it would
5277 disable operand forwarding.
5278 (new peephole2): Split MVCs if it would turn them into up to 2
5281 2017-12-01 Richard Biener <rguenther@suse.de>
5283 PR tree-optimization/83232
5284 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
5285 detection of same access. Instead of breaking the group here
5286 do not consider the duplicate. Add comment explaining real fix.
5288 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
5290 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
5292 2017-12-01 Sudakshina Das <sudi.das@arm.com>
5294 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
5296 2017-12-01 Jakub Jelinek <jakub@redhat.com>
5298 * function.h (struct function): Remove cilk_frame_decl,
5299 is_cilk_function and calls_cilk_spawn fields.
5300 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
5302 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
5303 cilk_elemental field.
5304 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
5305 * target.def: Adjust comment.
5306 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
5307 Don't test cilk_elemental.
5309 PR tree-optimization/83233
5310 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
5311 bswap_stat name for the struct.
5314 * tree.h (SWITCH_BREAK_LABEL_P): Define.
5315 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
5316 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
5317 SWITCH_BREAK_LABEL_P set on the label.
5318 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
5319 added for default case if it was missing and not all cases covered.
5320 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
5321 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
5323 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
5324 to avoid -Wimplicit-fallthrough warning.
5325 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
5326 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
5328 PR tree-optimization/83221
5329 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
5331 (init_reassoc): Formatting fix.
5334 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
5335 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
5339 * tree-cfg.c: Include asan.h.
5340 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
5342 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5344 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
5346 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5348 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
5349 vpcmpeqd instruction.
5351 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5353 * config/i386/i386.c (standard_sse_constant_opcode): Fix
5354 registers type for 128bit mode.
5356 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5358 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
5360 * gengtype-state.c (read_state_pair): Likewise.
5361 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
5362 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
5363 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
5364 * ifcvt.c (noce_try_store_flag_constants): Likewise.
5365 * tree-ssa-ccp.c (ccp_finalize): Likewise.
5366 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
5367 * builtins.c (fold_builtin_3): Likewise.
5368 * graphite-scop-detection.c
5369 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
5370 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
5373 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
5376 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
5377 parameter. Return true if ignore_decl_rtl_set_p is true and other
5378 conditions are satisfied.
5379 * asan.h (asan_protect_global): Add new parameter.
5380 * varasm.c (categorize_decl_for_section): Pass true as second parameter
5381 to asan_protect_global calls.
5383 2017-11-30 Jim Wilson <jimw@sifive.com>
5385 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
5386 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
5387 -mstrict-align, add info on default value. Delete redundant lines for
5388 -mabi. Add missing -mexplicit-relocs docs.
5390 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5392 * config/arc/arc.md (trap): New pattern.
5394 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5396 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
5397 instruction to end into a delay slot.
5398 * config/arc/arc.md (cond_delay_insn): Check if the instruction
5399 can be placed into a delay slot against reg_note.
5401 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5403 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
5404 labels number of usages.
5406 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5408 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
5410 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
5412 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5415 * internal-fn.c (expand_mul_overflow): Optimize unsigned
5416 multiplication by power of 2 constant into two shifts + comparison.
5418 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
5421 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
5423 2017-11-30 Richard Biener <rguenther@suse.de>
5425 PR tree-optimization/83202
5426 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
5427 allow_peel argument and guard peeling.
5428 (canonicalize_loop_induction_variables): Likewise.
5429 (canonicalize_induction_variables): Pass false.
5430 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
5431 peeling from cunrolli.
5433 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5435 * combine.c (try_combine): Print a message to dump file whenever
5436 I0, I1, or I2 cannot be combined into I3.
5438 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5440 PR rtl-optimization/83156
5441 PR rtl-optimization/82621
5442 * combine.c (try_combine): Don't split an I2 if one of the dests is
5443 set again before I3. Allow unused dests.
5445 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5447 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
5449 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
5451 PR rtl-optimization/80818
5452 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
5453 recursively. Use insn code for clobber.
5454 (lra_set_insn_recog_data): Pass the new arg to
5455 collect_non_operand_hard_regs.
5456 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
5458 (lra_update_insn_regno_info): Pass insn to
5459 add_regs_to_insn_regno_info.
5461 2017-11-29 Jim Wilson <jimw@sifive.com>
5462 Andrew Waterman <andrew@sifive.com>
5464 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
5465 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
5466 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
5467 use SINGLE_SHIFT_COST.
5468 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
5469 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
5471 2017-11-29 Julia Koval <julia.koval@intel.com>
5473 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
5474 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
5475 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
5476 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
5477 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
5478 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
5479 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
5480 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
5481 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
5482 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
5483 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
5484 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
5485 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
5486 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
5487 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
5488 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
5489 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
5490 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
5491 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
5492 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
5493 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
5494 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
5495 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
5496 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
5497 __builtin_ia32_vpshldv_v2di_mask,
5498 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
5499 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
5500 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
5502 2017-11-29 Julia Koval <julia.koval@intel.com>
5504 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
5505 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
5506 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
5507 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
5508 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
5509 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
5510 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
5511 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
5512 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
5513 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
5514 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
5515 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
5516 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
5517 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
5518 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
5519 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
5520 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
5521 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
5522 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5523 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
5524 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
5526 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5528 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
5529 movsi_pic_gotdata_op instruction as a load for the UT699 errata
5532 2017-11-29 Martin Aberg <maberg@gaisler.com>
5534 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
5535 to prevent b2bst errata sequence.
5536 (sqrtdf2_fix): Likewise.
5538 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5540 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
5541 (div_sqrt_insn_p): New function.
5542 (sparc_do_work_around_errata): Insert NOP instructions to
5543 prevent sequences that could trigger the TN-0013 errata for
5544 certain LEON3 processors.
5545 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
5546 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
5547 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
5548 (in_branch_delay): Prevent div and sqrt in delay slot if
5550 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
5552 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5554 * config/sparc/sparc.c (atomic_insn_p): New function.
5555 (sparc_do_work_around_errata): Insert NOP instructions to
5556 prevent sequences that could trigger the TN-0010 errata for
5558 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
5559 instruction referable in atomic_insns_p.
5561 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5563 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
5564 (atomic_compare_and_swap_leon3_1): Likewise.
5567 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5569 * config/sparc/sparc.c (fpop_insn_p): New function.
5570 (sparc_do_work_around_errata): Insert NOP instructions to
5571 prevent sequences that could trigger the TN-0012 errata for
5573 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
5574 * config/sparc/sparc.md (fix_gr712rc): New attribute.
5575 (in_branch_annul_delay): Prevent floating-point instructions
5576 in delay slot of annulled integer branch.
5578 2017-11-29 Richard Biener <rguenther@suse.de>
5580 PR tree-optimization/83202
5581 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
5583 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
5584 nodes vectorized to the same stmts multiple times.
5585 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
5586 (vect_analyze_slp_instance): Adjust.
5587 (scalar_stmts_to_slp_tree_map_t): New typedef.
5588 (vect_schedule_slp_instance): Add a map recording the SLP node
5589 representing the vectorized stmts for a set of scalar stmts.
5590 Avoid code-generating redundancies.
5591 (vect_schedule_slp): Allocate map and pass it down.
5593 2017-11-29 Nathan Sidwell <nathan@acm.org>
5596 * tree.c (build_complex_type): Fix canonicalization. Only fill in
5599 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
5601 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
5602 cases for printing LDP/STP memory addresses.
5603 (aarch64_print_address_internal): Renamed from
5604 aarch64_print_operand_address, added parameter, add Pmode check.
5605 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
5606 (aarch64_print_operand_address): Indirect to
5607 aarch64_print_address_internal.
5608 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
5609 'y' operand output specifier.
5611 2017-11-29 Jakub Jelinek <jakub@redhat.com>
5614 * tree.c (build_simple_mem_ref_loc): Handle
5615 get_addr_base_and_unit_offset returning a MEM_REF.
5618 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
5621 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
5624 2017-11-28 David Malcolm <dmalcolm@redhat.com>
5626 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
5628 (layout::move_to_column): Likewise.
5630 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5633 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
5634 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
5636 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5637 Martin Jambor <mjambor@suse.cz>
5640 * tree.h (expr_type_first_operand_type_p): Declare
5641 * tree.c (expr_type_first_operand_type_p): New function.
5642 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
5643 (ipa_value_from_jfunc): Adjust declaration.
5644 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
5645 Use it as result type for arithmetics, unless it is NULL in which case
5646 be more conservative.
5647 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
5648 ipa_get_jf_pass_through_result.
5649 (propagate_vals_across_pass_through): Likewise.
5650 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
5651 is to propagate_vals_across_pass_through.
5652 (propagate_constants_across_call): Pass PARM_TYPE to
5653 propagate_scalar_across_jump_function.
5654 (find_more_scalar_values_for_callers_subset): Pass parameter type to
5655 ipa_value_from_jfunc.
5656 (cgraph_edge_brings_all_scalars_for_node): Likewise.
5657 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
5658 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
5659 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
5660 target_type, pass it to ipa_value_from_jfunc.
5661 (update_indirect_edges_after_inlining): Pass parameter type to
5662 try_make_edge_direct_simple_call.
5664 2017-11-28 Jeff Law <law@redhat.com>
5666 * gimple-ssa-evrp-analyze.c
5667 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
5668 refine ranges if scev_initialized_p returns true.
5669 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
5671 2017-11-28 Julia Koval <julia.koval@intel.com>
5673 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
5674 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
5675 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
5676 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
5677 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
5678 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
5679 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
5680 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
5681 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
5682 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
5683 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
5684 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
5685 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
5686 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
5687 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
5688 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
5689 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
5690 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
5691 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
5692 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
5693 __builtin_ia32_vpshrd_v2di_mask): New builtins.
5694 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
5696 2017-11-28 Julia Koval <julia.koval@intel.com>
5698 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
5699 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
5700 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
5701 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
5702 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
5703 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
5704 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
5705 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
5706 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
5707 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
5708 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
5709 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
5710 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
5711 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
5712 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
5713 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
5714 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
5715 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
5716 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
5717 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
5718 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
5719 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
5720 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
5721 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
5722 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
5723 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
5724 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
5725 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
5726 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
5727 __builtin_ia32_vpshld_v2di_mask): New builtins.
5728 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5729 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
5731 2017-11-28 Richard Biener <rguenther@suse.de>
5733 PR tree-optimization/80776
5734 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
5736 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
5738 (evrp_range_analyzer::record_ranges_from_incoming_edges):
5739 If the incoming edge is an effective fallthru because the other
5740 edge only reaches a __builtin_unreachable () then record ranges
5741 derived from the controlling condition in SSA info.
5742 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
5743 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5745 2017-11-28 Olivier Hainque <hainque@adacore.com>
5747 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
5748 on mingw build hosts.
5750 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5753 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
5755 2017-11-28 Richard Biener <rguenther@suse.de>
5758 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
5759 copies generated from memcpy use a character array as reference
5762 2017-11-28 Julia Koval <julia.koval@intel.com>
5763 Sebastian Peryt <sebastian.peryt@intel.com>
5765 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
5766 c-family/c-cilkplus.o, c-family/array-notation-common.o,
5767 cilk-common.o, cilk.h, cilk-common.c): Remove.
5769 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
5770 * builtins.c (is_builtin_name): Remove cilkplus condition.
5771 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
5772 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
5773 cilk-builtins.def, cilkplus.def): Remove.
5774 * cif-code.def (CILK_SPAWN): Remove.
5775 * cilk-builtins.def: Delete.
5776 * cilk-common.c: Ditto.
5778 * cilkplus.def: Ditto.
5779 * config/darwin.h (fcilkplus): Delete.
5780 * cppbuiltin.c: Ditto.
5781 * doc/extend.texi: Remove cilkplus doc.
5782 * doc/generic.texi: Ditto.
5783 * doc/invoke.texi: Ditto.
5784 * doc/passes.texi: Ditto.
5785 * gcc.c (fcilkplus): Remove.
5786 * gengtype.c (cilk.h): Remove.
5787 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
5789 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
5791 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
5793 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
5794 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
5795 cilkplus conditions.
5796 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
5797 inline_read_section): Ditto.
5798 * ipa-inline-analysis.c (cilk.h): Remove.
5799 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
5800 * lto-wrapper.c (merge_and_complain, append_compiler_options,
5801 append_linker_options): Remove condition for fcilkplus.
5802 * lto/lto-lang.c (cilk.h): Remove.
5803 (lto_init): Remove condition for fcilkplus.
5804 * omp-expand.c (expand_cilk_for_call): Delete.
5805 (expand_omp_taskreg, expand_omp_for_static_chunk,
5806 expand_omp_for): Remove cilkplus
5808 (expand_cilk_for): Delete.
5809 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
5810 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
5811 execute_lower_omp, diagnose_sb_0): Ditto.
5812 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
5813 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5814 * tree-nested.c: Ditto.
5815 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
5816 (dump_generic_node): Ditto.
5817 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5818 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
5820 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
5822 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5824 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
5825 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
5826 for vector mode and !TARGET_SIMD.
5828 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5830 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
5832 * tree.h (SWITCH_LABELS): Remove.
5833 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
5834 assert SWITCH_BODY is non-NULL.
5835 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
5837 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
5839 PR tree-optimization/80788
5840 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
5841 has TREE_OVERFLOW set, call drop_tree_overflow.
5843 2017-11-28 Richard Biener <rguenther@suse.de>
5845 PR tree-optimization/83158
5846 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
5848 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5851 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5852 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5854 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5856 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5857 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5859 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5862 * varasm.c (compare_constant): Take the mode of the constants into
5863 account when comparing floating point constants.
5865 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5867 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5868 from explicit instantiation of debug_helper.
5869 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5871 2017-11-27 Richard Biener <rguenther@suse.de>
5873 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5876 2017-11-27 Richard Biener <rguenther@suse.de>
5878 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5879 (type_hash_canon): Also clear int_cst_hash_table entry for
5881 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5883 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5885 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5886 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5887 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5889 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5891 * hash-map.h (gt_cleare_cache): Avoid UB.
5893 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5895 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5896 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5897 (decide_unroll_runtime_iterations): Likewise.
5898 (decide_unroll_stupid): Likewise.
5900 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5903 * config/i386/i386.md: Add a loop with incssp.
5905 2017-11-27 Martin Jambor <mjambor@suse.cz>
5907 PR tree-optimization/81248
5908 * tree-sra.c (splice_param_accesses): Remove size check.
5909 (decide_one_param_reduction): Fix size check.
5910 * gimple-pretty-print.c (dump_profile): Silence warning.
5911 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5913 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5916 * dbxout.c (lastlineno): New variable.
5917 (dbx_debug_hooks): Use dbxout_switch_text_section as
5918 switch_text_section debug hook.
5919 (dbxout_function_end): Switch to current_function_section
5920 rather than function_section. If crtl->has_bb_partition,
5921 output just one N_FUN, depending on in_cold_section_p.
5922 (dbxout_source_line): Remember last lineno in lastlineno.
5923 (dbxout_switch_text_section): New function.
5924 (dbxout_function_decl): Adjust dbxout_block caller.
5925 (dbx_block_with_cold_children): New function.
5926 (dbxout_block): Return true if any LBRAC/RBRAC have been
5927 emitted. Use dbx_block_with_cold_children at depth == 0
5928 in second partition. Add PARENT_BLOCKNUM argument, pass
5929 it optionally adjusted to children. Output LBRAC/RBRAC
5930 around recursive call only if the block is in the current
5931 partition, if not and anything was output, emit empty
5933 * final.c (final_scan_insn): Compute cold_function_name
5934 before calling switch_text_section debug hook. Call
5935 that hook even if dwarf2out_do_frame if not emitting
5939 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5940 TREE_READONLY decls.
5942 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5944 PR rtl-optimization/82488
5945 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5947 2017-11-26 Julia Koval <julia.koval@intel.com>
5949 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5951 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5954 2017-11-26 Julia Koval <julia.koval@intel.com>
5956 * config/i386/driver-i386.c (host_detect_local_cpu):
5957 Detect skylake-avx512.
5959 2017-11-26 Julia Koval <julia.koval@intel.com>
5961 * config.gcc: Add -march=cannonlake.
5962 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5963 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5964 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5965 (PTA_CANNONLAKE): New.
5966 (processor_target_table): Add cannonlake.
5967 (ix86_option_override_internal): Ditto.
5968 (fold_builtin_cpu): Ditto.
5969 (get_builtin_code_for_version): Handle cannonlake.
5970 (M_INTEL_COREI7_CANNONLAKE): New.
5971 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5972 * doc/invoke.texi: Add -march=cannonlake.
5974 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5976 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5977 (try_init_one_plugin): Alternative implementation for MinGW.
5978 * Makefile.in (plugin_implib): New.
5979 (gengtype-lex.c): Fix broken AIX workaround.
5980 * configure: Regenerate.
5981 * doc/plugins.texi: Document support for MinGW.
5983 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5985 PR rtl-optimization/81553
5986 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5987 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5988 is a shift where C1 has different mode than the whole shift, use C1's
5989 mode for MULT rather than the shift's mode.
5992 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5993 builtins not enabled in the currently selected ISA.
5995 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5997 PR tree-optimization/71026
5998 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5999 (insert_reciprocals): Change to insert reciprocals before a division
6000 by a square and to insert the square of a reciprocal.
6001 (execute_cse_reciprocals_1): Change to consider division by a square.
6002 (register_division_in): Add importance parameter.
6004 2017-11-24 Richard Biener <rguenther@suse.de>
6006 PR tree-optimization/82402
6007 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
6008 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
6010 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
6012 * match.pd (0-ptr): New transformation.
6014 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
6017 * ipa-inline.c (inline_small_functions): Set current badnes correctly
6018 when skipping checking.
6020 2017-11-24 Richard Biener <rguenther@suse.de>
6022 PR tree-optimization/83128
6023 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
6024 (vn_reference_lookup_3): Likewise.
6026 2017-11-24 Jakub Jelinek <jakub@redhat.com>
6029 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
6030 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
6033 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
6034 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
6035 check for the latter with gimple_call_builtin_p. Do not handle
6036 BUILT_IN_STPNCPY_CHK which is not a pass through call.
6038 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
6040 * config/arm/arm_neon.h: Fix pragma GCC push_options before
6043 2017-11-23 Julia Koval <julia.koval@intel.com>
6045 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
6046 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
6047 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
6048 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
6049 _mm512_maskz_expandloadu_epi16): New intrinsics.
6050 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
6051 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
6052 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
6053 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
6054 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
6055 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
6056 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
6057 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
6058 _mm256_maskz_expandloadu_epi8): New intrinsics.
6059 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
6060 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
6061 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
6062 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
6063 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
6064 * config/i386/sse.md (VI248_VLBW): New iterator.
6065 (expand<mode>_mask, expand<mode>_maskz): New patterns.
6067 2017-11-23 Julia Koval <julia.koval@intel.com>
6069 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
6070 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
6071 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
6072 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
6073 _mm512_mask_compressstoreu_epi16): New.
6074 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
6075 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
6076 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
6077 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
6078 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
6079 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
6080 _mm256_mask_compressstoreu_epi8): New.
6081 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
6082 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
6083 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
6084 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
6085 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
6086 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
6087 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
6088 __builtin_ia32_compresshi128_mask,
6089 __builtin_ia32_compressstoreuqi512_mask,
6090 __builtin_ia32_compressstoreuhi512_mask,
6091 __builtin_ia32_compressstoreuqi256_mask,
6092 __builtin_ia32_compressstoreuqi128_mask,
6093 __builtin_ia32_compressstoreuhi256_mask,
6094 __builtin_ia32_compressstoreuhi128_mask): New builtins.
6095 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
6097 (ix86_expand_special_args_builtin): Handle new types.
6098 (s4fma_expand): Handle new builtin array.
6099 * config/i386/immintrin.h: Include new headers.
6100 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
6101 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
6103 2017-11-23 Jakub Jelinek <jakub@redhat.com>
6105 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
6106 fixes. Declare temp and g variables at the top in order to avoid
6107 {} in most of the cases.
6109 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
6111 * match.pd (ptr-0): New transformation.
6113 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
6115 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
6116 (struct addr_mode_cost_table): New.
6117 (struct tune_params): Add field addr_mode_costs.
6118 * config/arm/arm.c (generic_addr_mode_costs): New.
6119 (arm_slowmul_tune): Initialise addr_mode_costs field.
6120 (arm_fastmul_tune): Likewise.
6121 (arm_strongarm_tune): Likewise.
6122 (arm_xscale_tune): Likewise.
6123 (arm_9e_tune): Likewise.
6124 (arm_marvell_pj4_tune): Likewise.
6125 (arm_v6t2_tune): Likewise.
6126 (arm_cortex_tune): Likewise.
6127 (arm_cortex_a8_tune): Likewise.
6128 (arm_cortex_a7_tune): Likewise.
6129 (arm_cortex_a15_tune): Likewise.
6130 (arm_cortex_a35_tune): Likewise.
6131 (arm_cortex_a53_tune): Likewise.
6132 (arm_cortex_a57_tune): Likewise.
6133 (arm_exynosm1_tune): Likewise.
6134 (arm_xgene1_tune): Likewise.
6135 (arm_cortex_a5_tune): Likewise.
6136 (arm_cortex_a9_tune): Likewise.
6137 (arm_cortex_a12_tune): Likewise.
6138 (arm_cortex_a73_tune): Likewise.
6139 (arm_v7m_tune): Likewise.
6140 (arm_cortex_m7_tune): Likewise.
6141 (arm_v6m_tune): Likewise.
6142 (arm_fa726te_tune): Likewise.
6143 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
6145 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
6147 * config/arm/arm.c (arm_mem_costs): New function.
6148 (arm_rtx_costs_internal): Use arm_mem_costs.
6150 2017-11-23 Mark Wielaard <mark@klomp.org>
6152 * dwarf2out.c (init_sections_and_labels): Use generation to create
6153 unique ranges_section_label and ranges_base_label. Return generation.
6154 (output_rnglists): Add generation argument. Use generation to create
6155 unique ranges labels.
6156 (dwarf2out_finish): Get generation from init_sections_and_labels
6157 and pass generation to output_rnglists.
6159 2017-11-23 Mike Stump <mikestump@comcast.net>
6160 Eric Botcazou <ebotcazou@adacore.com>
6162 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
6163 * cfgloop.h (struct loop): Add unroll field.
6164 * function.h (struct function): Add has_unroll bitfield.
6165 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
6166 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
6167 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
6168 (pass_rtl_unroll_loops::gate): Likewise.
6169 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
6170 for which loop->unroll==1.
6171 (decide_unroll_constant_iterations): Use note for consistency and
6172 take loop->unroll into account. Return early if loop->unroll is set.
6173 Fix thinko in existing test.
6174 (decide_unroll_runtime_iterations): Use note for consistency and
6175 take loop->unroll into account.
6176 (decide_unroll_stupid): Likewise.
6177 * lto-streamer-in.c (input_cfg): Read loop->unroll.
6178 * lto-streamer-out.c (output_cfg): Write loop->unroll.
6179 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
6181 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
6182 (print_loop): Print loop->unroll if set.
6183 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
6184 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
6185 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
6187 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
6188 loop->unroll is set and smaller than the trip count. Otherwise bypass
6189 entirely the heuristics if loop->unroll is set. Remove dead note.
6190 Fix off-by-one bug in other note.
6191 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
6192 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
6194 (tree_unroll_loops_completely): Make static.
6195 (pass_complete_unroll::execute): Use correct type for variable.
6196 (pass_complete_unrolli::execute): Fix formatting.
6197 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
6199 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
6201 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
6202 enable when TARGET_PREFER_AVX128 is set.
6204 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6206 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
6208 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
6209 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
6211 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6213 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
6214 frequencies. Fix estimation of aggregate parameters.
6216 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6218 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
6221 2017-11-23 Tom de Vries <tom@codesourcery.com>
6223 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
6225 2017-11-23 Tom de Vries <tom@codesourcery.com>
6227 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
6230 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
6233 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
6234 sibcall_value_pcrel_fdpic): Use local variable instead of
6236 (calli_tbr_rel): Add missing operand 2.
6237 (call_valuei_tbr_rel): Add missing operand 3.
6239 2017-11-23 Jakub Jelinek <jakub@redhat.com>
6242 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
6243 bitpos/bitsize covering the whole destination, use store_expr only if
6244 the complex mode is the same. Otherwise, use expand_normal and if
6245 it returns CONCAT, subreg each part separately instead of trying to
6246 subreg the whole result.
6248 2017-11-23 Richard Biener <rguenther@suse.de>
6250 PR tree-optimization/23094
6251 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
6252 come here from walking over backedges in the first iteration.
6253 (vn_reference_lookup_3): Skip clobbers that store the same value.
6255 2017-11-23 Richard Biener <rguenther@suse.de>
6257 PR tree-optimization/81403
6258 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
6259 a block we need a leader relative to.
6260 (phi_translate_1): For nary processing require a leader from
6261 get_representative_for given we run expression simplification
6262 using match-and-simplify. Remove previous fix.
6264 2017-11-22 Jeff Law <law@redhat.com>
6266 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
6267 Use new method allocate_value_range rather than accessing the
6268 vrp_value_range_pool data member directly.
6269 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
6270 to use extract_range_from_stmt method to avoid need for
6271 extract_range_from_assignment method.
6272 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
6273 method rather than setting values_propgated data member directly.
6274 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
6275 and values propagated data members and extract_range_from_assignment
6276 method. Reorder private data members to conform to standards.
6277 Add new methods set_lattice_propagation_complete and
6278 allocate_value_range.
6280 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
6282 PR rtl-optimization/83030
6283 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
6284 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
6285 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
6286 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
6287 CROSSING_JUMP_P flag is already set before setting it.
6288 * cfgrtl.c (fixup_partition_crossing): Likewise.
6289 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
6292 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6294 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
6295 Handle the case where both arguments are using gen_const_vec_series.
6297 2017-11-22 David Malcolm <dmalcolm@redhat.com>
6300 * pretty-print.c (pp_format): Move quoting implementation to
6301 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
6302 to pass address of "quote" local.
6303 (pp_begin_quote): New function.
6304 (pp_end_quote): New function.
6305 * pretty-print.h (printer_fn): Convert penultimate param from bool
6307 (pp_begin_quote): New decl.
6308 (pp_end_quote): New decl.
6309 * tree-diagnostic.c (default_tree_printer): Convert penultimate
6310 param from bool to bool *.
6311 * tree-diagnostic.h (default_tree_printer): Likewise.
6313 2017-11-22 Jeff Law <law@redhat.com>
6315 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
6316 Initialize vr_values.
6317 (evrp_range_analyzer::try_find_new_range): Call methods attached to
6318 vr_values via vr_values class instance rather than delegators.
6319 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
6320 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
6321 (evrp_range_analyzer::push_value_range): Likewise.
6322 (evrp_range_analyzer::pop_value_range): Likewise.
6323 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
6324 most delegators. Those remaining are exposed as public interfaces.
6325 Make vr_values a pointer and private.
6326 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
6328 (evrp_range_analyzer::get_vr_value): New method.
6329 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
6330 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
6331 (class evrp_dom_walker): Attach evrp_folder class, initialize
6332 it in the ctor. Remove temporary delegators.
6333 (evrp_dom_walker::before_dom_children): Call methods in attached
6334 evrp_range_analyzer class via class instance pointer. Use
6335 free value_range_constant_singleton to remove need for
6336 op_with_constant_singleton_value delegator method. Do not
6337 create a vrp_prop class instance for every call! Narrow
6338 scope of a couple variables.
6339 (evrp_dom_walker::cleanup): Call methods in attached
6340 evrp_range_analyzer class via class instance pointer.
6341 * vr-values.h (class vr_values): Privatize many methods and
6344 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6346 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
6347 * cfgexpand.c (expand_debug_expr): Remove handling for them.
6348 * expr.c (expand_expr_real_2): Likewise.
6349 * fold-const.c (const_unop): Likewise.
6350 * optabs-tree.c (optab_for_tree_code): Likewise.
6351 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
6352 * tree-inline.c (estimate_operator_cost): Likewise.
6353 * tree-pretty-print.c (dump_generic_node): Likewise.
6354 (op_code_prio): Likewise.
6355 (op_symbol_code): Likewise.
6356 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
6357 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
6358 * internal-fn.c (direct_internal_fn_optab): New function.
6359 (direct_internal_fn_array, direct_internal_fn_supported_p
6360 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
6361 * fold-const-call.c (fold_const_reduction): New function.
6362 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
6364 * tree-vect-loop.c: Include internal-fn.h.
6365 (reduction_code_for_scalar_code): Rename to...
6366 (reduction_fn_for_scalar_code): ...this and return an internal
6368 (vect_model_reduction_cost): Take an internal_fn rather than
6370 (vect_create_epilog_for_reduction): Likewise. Build calls rather
6372 (vectorizable_reduction): Use internal functions rather than tree
6373 codes for the reduction operation. Update calls to the functions
6375 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
6376 Use calls to internal functions rather than REDUC tree codes.
6377 * config/aarch64/aarch64-simd.md: Update comment accordingly.
6379 2017-11-22 Olivier Hainque <hainque@adacore.com>
6381 * config/vxworks.c (vxworks_override_options): Pick default
6382 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
6383 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
6384 DWARF_GNAT_ENCODINGS_DEFAULT.
6385 * config/vxworksae.h: Likewise.
6387 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
6389 PR tree-optimization/83104
6390 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
6393 2017-11-22 Marek Polacek <polacek@redhat.com>
6394 H.J. Lu <hongjiu.lu@intel.com>
6395 Jason Merrill <jason@redhat.com>
6400 * calls.c (initialize_argument_information): Call
6401 warn_parameter_passing_abi target hook.
6402 (store_one_arg): Use 0 for empty record size. Don't push 0 size
6403 argument onto stack.
6404 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
6405 * common.opt: Update -fabi-version description.
6406 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
6407 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
6409 (ix86_is_empty_record): New function.
6410 (ix86_warn_parameter_passing_abi): New function.
6411 (TARGET_EMPTY_RECORD_P): Redefine.
6412 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
6413 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
6414 * doc/tm.texi: Regenerated.
6415 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
6416 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
6417 * dwarf2out.c (get_ultimate_context): Move to tree.c.
6418 * explow.c (hard_function_value): Call arg_int_size_in_bytes
6419 instead of int_size_in_bytes.
6420 * expr.c (copy_blkmode_to_reg): Likewise.
6421 * function.c (aggregate_value_p): Return 0 for empty types.
6422 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
6424 (locate_and_pad_parm): Call arg size_in_bytes instead
6426 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
6427 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
6428 * target.def (empty_record_p, warn_parameter_passing_abi): New target
6430 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
6431 (std_gimplify_va_arg_expr): Skip empty records. Call
6432 arg_size_in_bytes instead size_in_bytes.
6433 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
6434 * tree-core.h (tree_type_common): Add empty_flag.
6435 (tree_decl_common): Update comments.
6436 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
6438 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6439 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
6441 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6442 * tree.c (default_is_empty_type): New function.
6443 (default_is_empty_record): New function.
6444 (arg_int_size_in_bytes): New function.
6445 (arg_size_in_bytes): New function.
6446 (get_ultimate_context): New function.
6447 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
6448 TRANSLATION_UNIT_WARN_EMPTY_P.
6449 (default_is_empty_record, arg_int_size_in_bytes,
6450 arg_size_in_bytes, get_ultimate_context): Declare.
6452 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6454 * config/arm/arm.c (cmse_clear_registers): New function.
6455 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
6456 code by call to cmse_clear_registers.
6457 (cmse_nonsecure_entry_clear_before_return): Likewise.
6459 2017-11-22 Tamar Christina <tamar.christina@arm.com>
6461 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
6462 (vdot_s32, vdotq_s32): New.
6463 (vdot_lane_u32, vdotq_lane_u32): New.
6464 (vdot_lane_s32, vdotq_lane_s32): New.
6467 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6470 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
6471 for unsigned values with fewer HWIs than the precision.
6472 (test_overflow): New function.
6473 (wide_int_cc_tests): Call it.
6475 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6477 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
6478 has the same mode class as Pmode.
6480 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6482 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
6483 padding_bits_to_clear_ptr.
6484 (cmse_nonsecure_entry_clear_before_return): Likewise.
6486 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6488 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
6489 auto_sbitap instead of integer bitfield to control register needing
6492 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6494 PR tree-optimization/83044
6495 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
6496 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
6497 that need the upper bound. Subtract offset from
6498 get_addr_base_and_unit_offset only if positive and subtract it
6499 before division by eltsize rather than after it.
6502 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
6503 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
6504 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
6508 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
6509 expand_binop with *mul_widen_optab, make sure at least one of the
6510 operands doesn't have VOIDmode.
6513 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
6515 PR rtl-optimization/82044
6516 PR tree-optimization/82042
6517 * dse.c (record_store): Check for overflow.
6518 (check_mem_read_rtx): Properly check for overflow if width == -1, call
6519 add_wild_read instead of clear_rhs_from_active_local_stores on
6520 overflow and log it into dump_file.
6522 2017-11-22 Richard Biener <rguenther@suse.de>
6524 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
6525 fake edges to exit when looking for a place to insert.
6526 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
6528 (insert_into_preds_of_block): Commit edge insertion immediately,
6529 assert that doesn't require new BBs.
6530 (fini_pre): Release expressions.
6531 (pass_pre::execute): Shuffle things around a bit, if the fn
6532 is too large do not compute AVAIL either as this is really the
6535 2017-11-22 Richard Biener <rguenther@suse.de>
6537 PR tree-optimization/83089
6538 * tree-if-conv.c (pass_if_conversion::execute): If anything
6539 changed reset SCEV and free the number of iteration estimates.
6541 2017-11-21 Martin Sebor <msebor@redhat.com>
6543 PR tree-optimization/82945
6544 * calls.h (warn_nonstring_bound): Remove unused function.
6546 2017-11-21 Martin Sebor <msebor@redhat.com>
6548 PR tree-optimization/82945
6549 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
6550 * calls.h (maybe_warn_nonstring_arg): Declare new function.
6551 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
6553 (initialize_argument_information): Call maybe_warn_nonstring_arg.
6554 * calls.h (get_attr_nonstring_decl): Declare new function.
6555 * doc/extend.texi (attribute nonstring): Update.
6556 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
6557 get_attr_nonstring_decl and handle it.
6558 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
6559 detection of nul-termination.
6560 (strlen_to_stridx): Change to a pointer.
6561 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
6562 (pass_strlen::execute): Same.
6564 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
6566 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
6567 for the new option -mprefer-vector-width=[none|128|256|512].
6568 * config/i386/i386.c (ix86_target_string): remove old style options
6569 -mprefer-avx256 and make -mprefer-avx128 as alias.
6570 (ix86_option_override_internal): Apply defaults for the
6571 -mprefer-vector-width=[128|256] option.
6572 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
6573 Implement macros to work with -mprefer-vector-width=.
6574 * config/i386/i386.opt: Implemented option
6575 -mprefer-vector-width=[none|128|256|512].
6576 * doc/invoke.texi: Documentation for
6577 -mprefer-vector-width=[none|128|256|512].
6579 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
6581 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
6583 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
6585 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
6587 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
6589 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
6591 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
6592 * expr.c (expand_expr_real_2): Likewise.
6593 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
6594 fold_binary_loc): Likewise.
6595 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
6596 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
6597 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
6598 MINUS_EXPR transformations.
6599 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
6600 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
6601 * tree-inline.c (estimate_operator_cost): Likewise.
6602 * tree-pretty-print.c (dump_generic_node, op_code_prio,
6603 op_symbol_code): Likewise.
6604 * tree-vect-stmts.c (vectorizable_operation): Likewise.
6605 * vr-values.c (extract_range_from_binary_expr): Likewise.
6606 * varasm.c (initializer_constant_valid_p_1): Likewise.
6607 * tree.def: New tree code POINTER_DIFF_EXPR.
6609 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
6611 * config/i386/i386.md (*bswap<mode>2_movbe): Add
6612 integer suffix to movbe mnemonic.
6613 (*bswaphi2_movbe): Ditto.
6614 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
6616 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6619 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
6620 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
6621 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
6623 2017-11-21 Martin Liska <mliska@suse.cz>
6625 * tree-inline.c (expand_call_inline): Remove not needed
6628 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
6629 Jakub Jelinek <jakub@redhat.com>
6632 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
6633 Remove static keyword from f variable.
6635 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6637 PR tree-optimization/83086
6638 * gimple-ssa-store-merging.c
6639 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
6640 rather than n.base_addr.
6642 2017-11-21 Martin Liska <mliska@suse.cz>
6644 PR rtl-optimization/82044
6645 PR tree-optimization/82042
6646 * dse.c (check_mem_read_rtx): Check for overflow.
6648 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6650 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
6653 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6655 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
6656 and its corresponding call arg location note.
6658 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6659 Andrew Burgess <andrew.burgess@embecosm.com>
6661 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
6663 (arc_return_slot_offset): Likewise.
6664 (arc_eh_return_address_location): New declaration.
6665 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
6666 (MUST_SAVE_REGISTER): Add exception handler case.
6667 (MUST_SAVE_RETURN_ADDR): Likewise.
6668 (arc_frame_pointer_required): Likewise.
6669 (arc_frame_pointer_needed): New function.
6670 (arc_compute_frame_size): Changed.
6671 (arc_expand_prologue): Likewise.
6672 (arc_expand_epilogue): Likewise.
6673 (arc_initial_elimination_offset): Likewise.
6674 (arc_return_slot_offset): Delete.
6675 (arc_eh_return_address_location): New function.
6676 (arc_builtin_setjmp_frame_value): Likewise.
6677 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
6678 (EH_RETURN_STACKADJ_RTX): Define.
6679 (EH_RETURN_HANDLER_RTX): Likewise.
6680 * config/arc/arc.md (eh_return): Delete.
6682 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6684 * print-tree.h (debug_vec_tree): Remove prototype.
6685 * gdbinit.in (pvt): Remove macro.
6687 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6689 PR tree-optimization/83047
6690 * gimple-ssa-store-merging.c
6691 (imm_store_chain_info::output_merged_store): If the loads with the
6692 same vuse are in different basic blocks, for load_gsi pick a load
6693 location that is dominated by the other loads.
6696 * config/i386/i386.c (ix86_memmodel_check): Start
6697 -Winvalid-memory-model diagnostics with lowercase letter.
6700 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
6701 set in_cold_section_p to the partition containing loc_list->first.
6702 When seeing loc_list->last_before_switch node, update secname and
6703 perform range_across_switch second partition handling only after that.
6706 * run-rtl-passes.c: Include debug.h.
6707 (run_rtl_passes): Call debug_hooks->assembly_start.
6708 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
6712 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
6713 OPTAB_DIRECT in calls to expand_simple_binop.
6715 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6718 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
6720 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6723 * spellcheck.h (best_match::blithely_get_best_candidate): New
6726 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6728 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
6729 with lower case letter.
6731 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
6733 * config/i386/i386.md (bswaphi2): New expander.
6734 (*bswaphi2_movbe): New insn pattern.
6735 (bswaphi -> rorhi pepehole2): New peephole pattern.
6737 2017-11-20 Jeff Law <law@redhat.com>
6739 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
6740 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
6741 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
6742 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
6743 gimple-ssa-evrp-analyze.h.
6745 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
6746 set BB_VISITED here.
6747 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
6749 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6751 * tree-predcom.c: Add general comment on Store-Store chains.
6752 (split_data_refs_to_components): Postpone clearing eliminate_store_p
6754 (get_chain_last_ref_at): Rename into...
6755 (get_chain_last_write_at): ...this.
6756 (get_chain_last_write_before_load): New function.
6757 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
6758 CT_STORE_STORE when write reference is added.
6759 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
6760 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
6761 (initialize_root_vars_store_elim_1): Ditto.
6762 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
6763 definition is created.
6764 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
6765 chain by replacing it with dominant stored value.
6767 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6769 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
6771 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
6773 * vr-values.c (extract_range_from_binary_expr): Use a full range
6776 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
6778 * config/arm/arm.md (R4_REGNUM): Define constant.
6779 (nonsecure_call_internal): Remove r4 clobber.
6780 (nonsecure_call_value_internal): Likewise.
6781 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
6782 clobber and resequence match_operands.
6783 (nonsecure_call_value_reg_thumb1_v5): Likewise.
6784 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
6785 (nonsecure_call_value_reg_thumb2): Likewise.
6787 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6789 PR tree-optimization/78821
6790 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
6791 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
6792 to the address of the base rather than the base itself.
6793 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
6794 (find_bswap_or_nop_finalize): New function.
6795 (find_bswap_or_nop): Use it.
6796 (bswap_replace): Return a tree rather than bool, change first
6797 argument from gimple * to gimple_stmt_iterator, allow inserting
6798 into an empty sequence, allow ins_stmt to be NULL - then emit
6799 all stmts into gsi. Fix up MEM_REF address gimplification.
6800 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
6801 (struct store_immediate_info): Add N and INS_STMT non-static
6803 (store_immediate_info::store_immediate_info): Initialize them
6804 from newly added ctor args.
6805 (merged_store_group::apply_stores): Formatting fixes. Sort by
6807 (stmts_may_clobber_ref_p): For stores call also
6808 refs_anti_dependent_p.
6809 (gather_bswap_load_refs): New function.
6810 (imm_store_chain_info::try_coalesce_bswap): New method.
6811 (imm_store_chain_info::coalesce_immediate_stores): Use it.
6812 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
6813 (imm_store_chain_info::output_merged_store): Fail if number of
6814 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
6815 and NOP_EXPR rhs_code.
6816 (pass_store_merging::process_store): Compute n and ins_stmt, if
6817 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
6818 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
6820 (pass_store_merging::execute): Calculate dominators.
6822 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
6823 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
6824 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
6825 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6826 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
6827 class pass_optimize_bswap, bswap_replace,
6828 pass_optimize_bswap::execute): Moved to ...
6829 * gimple-ssa-store-merging.c: ... this file.
6830 Include optabs-tree.h.
6831 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
6832 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6833 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
6834 anonymous namespace, remove static keywords.
6835 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
6836 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
6838 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
6841 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6843 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
6845 * vec.h (debug_helper): New function.
6846 (DEFINE_DEBUG_VEC): New macro.
6847 * hash-set.h (debug_helper): New function.
6848 (DEFINE_DEBUG_HASH_SET): New macro.
6849 * cfg.c (debug_slim (edge)): New function.
6850 Call DEFINE_DEBUG_VEC for edges.
6851 Call DEFINE_DEBUG_HASH_SET for edges.
6852 * cfghooks.c (debug_slim (basic_block)): New function.
6853 Call DEFINE_DEBUG_VEC for basic blocks.
6854 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6855 * print-tree.c (debug_slim): New function to handle trees.
6856 Call DEFINE_DEBUG_VEC for trees.
6857 Call DEFINE_DEBUG_HASH_SET for trees.
6858 (debug (vec<tree, va_gc>) &): Remove.
6859 (debug (<vec<tree, va_gc>) *): Remove.
6860 * print-rtl.c (debug_slim): New function to handle const_rtx.
6861 Call DEFINE_DEBUG_VEC for rtx_def.
6862 Call DEFINE_DEBUG_VEC for rtx_insn.
6863 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6864 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6865 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6866 (debug (vec<rtx_insn *> *ptr): Remove.
6867 (debug_insn_vector): Remove.
6868 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6870 2017-11-20 Tom de Vries <tom@codesourcery.com>
6872 PR rtl-optimization/82020
6873 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6874 IF_THEN_ELSE condition.
6876 2017-11-19 Jeff Law <law@redhat.com>
6878 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6879 of degenerates resulting from ignoring an edge.
6881 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6884 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6886 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6889 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6890 for uninitialied counts.
6892 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6895 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6897 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6900 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6902 2017-11-19 Tom de Vries <tom@codesourcery.com>
6904 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6907 2017-11-19 Tom de Vries <tom@codesourcery.com>
6909 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6912 2017-11-19 Tom de Vries <tom@codesourcery.com>
6914 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6917 2017-11-19 Tom de Vries <tom@codesourcery.com>
6919 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6921 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6923 2017-11-19 Tom de Vries <tom@codesourcery.com>
6925 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6927 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6928 semicolon after MCORE_EXPORT_NAME call.
6930 2017-11-19 Tom de Vries <tom@codesourcery.com>
6933 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6934 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6936 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6938 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6939 (cgraph_edge::clone): Cleanup updating of profile.
6940 * ipa-cp.c (update_profiling_info): Likewise.
6941 * ipa-inline-transform.c (inline_transform): Likewise.
6942 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6943 * ipa-split.c (execute_split_functions): Do not split when function
6945 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6946 * profile-count.c (profile_count::dump): Add global0.
6947 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6949 (profile_count::to_sreal_scale): Likewise.
6950 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6951 (profile_count::combine_with_ipa_count): New function.
6952 * profile-count.h (profile_guessed_global0adjusted): New.
6953 (profile_count::adjusted_zero): New.
6954 (profile_count::global0adjusted): New.
6955 (profile_count::combine_with_ipa_count): New.
6956 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6957 correct profile of return block of split functions.
6958 (copy_cfg_body): Remove unused profile_count.
6959 (copy_body): Likewise.
6960 (expand_call_inline): Update.
6961 (tree_function_versioning): Update.
6963 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6965 * hash-set.h (hash_set::empty): New.
6966 * tree-ssa-threadbackward.h: Delete.
6967 * tree-ssa-threadbackward.c (class thread_jumps): New.
6968 Move max_threaded_paths into class.
6969 (fsm_find_thread_path): Remove arguments that are now in class.
6970 (profitable_jump_thread_path): Rename to...
6971 (thread_jumps::profitable_jump_thread_path): ...this.
6972 (convert_and_register_jump_thread_path): Rename to...
6973 (thread_jumps::convert_and_register_current_path): ...this.
6974 (check_subpath_and_update_thread_path): Rename to...
6975 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6976 (register_jump_thread_path_if_profitable): Rename to...
6977 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6978 (handle_phi): Rename to...
6979 (thread_jumps::handle_phi): ...this.
6980 (handle_assignment): Rename to...
6981 (thread_jumps::handle_assignment): ...this.
6982 (fsm_find_control_statement_thread_paths): Rename to...
6983 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6984 (find_jump_threads_backwards): Rename to...
6985 (thread_jumps::find_jump_threads_backwards): ...this.
6986 Initialize path local data.
6987 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6988 from within thread_jumps class.
6989 (pass_early_thread_jumps::execute): Same.
6991 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6993 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6995 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6997 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6998 to canonical location.
7000 2017-11-17 Steve Ellcey <sellcey@cavium.com>
7003 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
7005 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
7007 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
7009 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
7010 rcrt1.o%s/grcrt1.o%s for -static-pie.
7012 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
7014 * i386.c (ix86_multiplication_cost, ix86_division_cost,
7015 ix86_shift_rotate_cost): Break out from ...
7016 (ix86_rtx_costs): ... here.
7017 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
7020 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
7022 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
7023 when entry block was promoted unlikely.
7024 (estimate_bb_frequencies): Increase frequency scale.
7025 * profile-count.h (profile_count): Export precision info.
7027 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
7029 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
7030 disturbin profile of entry block.
7032 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
7034 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
7037 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
7039 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
7040 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
7042 2017-11-17 Jeff Law <law@redhat.com>
7044 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
7045 from evrp_dom_walker class. Various methods moved into new class.
7046 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
7047 (evrp_range_analyzer::enter): New method.
7048 (evrp_range_analyzer::leave): New method.
7049 (evrp_dom_walker): Remove delegators no longer needed by this class.
7050 Replace vr_values data member with evrp_range_analyzer
7052 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
7053 method extracted from evrp_dom_walker::before_dom_children.
7054 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
7055 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
7057 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
7058 Add private copy constructor and move assignment operators.
7059 Privatize methods and class data where trivially possible.
7060 (evrp_dom_walker::cleanup): New function, extracted from
7061 execute_early_vrp. Simplify access to class data.
7063 * vr-values.h (get_output_for_vrp): Prototype.
7064 * vr-values.c (get_output_for_vrp): New function extracted from
7065 vrp_visit_assignment_or_call and extract_range_from_stmt.
7066 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
7068 2017-11-17 Luis Machado <luis.machado@linaro.org>
7070 * config/aarch64/aarch64.c
7071 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
7072 (qdf24xx_tunings) <autoprefetcher_model>: Set to
7073 tune_params::AUTOPREFETCHER_WEAK.
7075 2017-11-17 Tamar Christina <tamar.christina@arm.com>
7078 * config/arm/arm.c (arm_valid_target_attribute_rec):
7079 Parse "arch=" and "+<ext>".
7080 (arm_valid_target_attribute_tree): Re-init global options.
7081 (arm_option_override): Make non-static.
7082 (arm_options_perform_arch_sanity_checks): Make errors fatal.
7083 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
7084 (__ARM_FEATURE_CRC32): Support undef.
7085 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
7086 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
7088 2017-11-17 David Malcolm <dmalcolm@redhat.com>
7090 * gdbinit.in (break-on-diagnostic): New command.
7092 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7094 * config/i386/i386.c (ix86_expand_epilogue): Change simple
7095 return to indirect jump for EH return if control-flow
7096 protection is enabled. Change explicit 'false' argument in
7097 pro_epilogue_adjust_stack with a value of
7099 * config/i386/i386.md (simple_return_indirect_internal):
7100 Remove SImode restriction to support 64-bit.
7102 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
7104 * combine.c (added_notes_insn): New.
7105 (try_combine): Handle added_notes_insn like added_links_insn.
7106 Rewrite return value code.
7107 (distribute_notes): Set added_notes_insn to the earliest insn we added
7110 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
7112 PR rtl-optimization/82621
7113 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
7114 dest of one of those SETs is unused.
7116 2017-11-17 Richard Biener <rguenther@suse.de>
7119 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
7120 * tree-pretty-print.c (dump_generic_node): Handle
7121 annot_expr_parallel_kind.
7122 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
7123 * gimplify.c (gimple_boolify): Likewise.
7125 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7127 * config.gcc (extra_headers): Add cet.h for x86 targets.
7128 * config/i386/cet.h: New file.
7129 * doc/install.texi: Add --enable-cet/--disable-cet.
7131 2017-11-17 Richard Biener <rguenther@suse.de>
7133 PR tree-optimization/83017
7134 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
7135 (gen_parallel_loop): Properly count iterations.
7136 (parallelize_loops): Handle loop->can_be_parallel independent
7137 of flag_loop_parallelize_all. Make static profitability test match
7139 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
7140 * invoke.texi (parloops-min-per-thread): Document.
7142 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
7144 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
7145 upstreaming review comments.
7147 2017-11-17 Tamar Christina <tamar.christina@arm.com>
7149 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
7150 with fast unaligned access.
7151 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
7153 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
7155 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
7156 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
7158 2017-11-17 Richard Biener <rguenther@suse.de>
7160 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
7161 folding of references.
7163 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
7166 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
7167 of replacing call to strncmp with corresponding call to strcmp when
7170 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
7172 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
7173 option prefer-avx256 for skylake-avx512 configuration.
7174 * config/i386/i386.c (ix86_option_override_internal): Ditto.
7175 (get_builtin_code_for_version): Ditto.
7177 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
7178 Monk Chiang <sh.chiang04@gmail.com>
7180 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
7181 (FIXED_REGISTERS): Reserve more register numbers.
7182 (CALL_USED_REGISTERS): Likewise.
7183 (REG_ALLOC_ORDER): Likewise.
7184 (REG_CLASS_CONTENTS): Likewise.
7185 (REGISTER_NAMES): Likewise.
7187 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
7188 Kito Cheng <kito.cheng@gmail.com>
7190 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
7192 * config/nds32/iterators.md: Add vector mode iterators and attributes.
7194 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
7196 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
7197 parameter list for vec_splats.
7199 2017-11-16 Joseph Myers <joseph@codesourcery.com>
7201 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
7203 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
7205 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
7208 * doc/install.texi: Document incompatibility of Perl >=5.6.26
7209 with the required version of automake 1.11.6.
7211 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
7213 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
7215 (DU_C2_3_power9): Correct reservation combinations.
7216 (FP_DIV_power9, VEC_DIV_power9): New.
7217 (power9-alu): Split out rotate/shift...
7218 (power9-rot): ...to here, correct dispatch resource.
7219 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
7221 (power9-fp): Correct latency.
7222 (power9-sdiv): Add div/sqrt resource.
7223 (power9-ddiv): Correct latency, add div/sqrt resource.
7224 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
7225 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
7227 (power9-qpdiv, power9-qpmul): Adjust resource usage.
7229 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
7231 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
7232 switch statement mapping KF built-ins to TF built-ins if we don't
7233 have the proper ISA 3.0 assembler support.
7235 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7237 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
7238 (lower_emutls_function_body): Do not compute it.
7240 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7242 * ipa-split.c (split_bb_info): Turn time to sreal.
7243 (split_point): Likewise.
7244 (dump_split_point): Likewise.
7245 (fine_split_points): Likewise.
7246 (execute_split_functions): Only zero split_bbs; turn time to sreals.
7248 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7250 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
7253 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7255 * predict.c (combine_predictions_for_bb): Preserve zero predicted
7257 (expensive_function_p): Remove useless assert.
7258 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
7260 2017-11-16 Martin Sebor <msebor@redhat.com>
7262 PR tree-optimization/82588
7263 PR tree-optimization/82583
7264 * tree-vrp.c (check_array_ref): Handle flexible array members,
7265 string literals, and inner indices.
7266 (search_for_addr_array): Add detail to diagnostics.
7268 2017-11-16 Nathan Sidwell <nathan@acm.org>
7272 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
7273 (SET_DECL_ASSEMBLER_NAME): Forward to
7274 overwrite_decl_assembler_name.
7275 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
7276 (overwrite_decl_assembler_name): Declare.
7277 * tree.c (overwrite_decl_assembler_name): New.
7278 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
7279 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
7280 (LANG_HOOKS_INITIALIZER): Add it.
7281 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
7282 * langhooks.c (lhd_set_decl_assembler_name): Use
7283 SET_DECL_ASSEMBLER_NAME.
7284 (lhd_overwrite_decl_assembler_name): Default implementation.
7286 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7287 Jackson Woodruff <jackson.woodruff@arm.com>
7289 PR tree-optimization/71026
7290 * match.pd: Canonicalize constant multiplies in division.
7292 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7294 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
7295 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
7296 Remove OPT_fomit_frame_pointer entry.
7297 * common/config/arc/arc-common.c: Likewise.
7298 * common/config/arm/arm-common.c: Likewise.
7299 * common/config/avr/avr-common.c: Likewise.
7300 * common/config/c6x/c6x-common.c: Likewise.
7301 * common/config/cr16/cr16-common.c: Likewise.
7302 * common/config/cris/cris-common.c: Likewise.
7303 * common/config/epiphany/epiphany-common.c: Likewise.
7304 * common/config/fr30/fr30-common.c: Likewise.
7305 * common/config/frv/frv-common.c: Likewise.
7306 * common/config/ia64/ia64-common.c: Likewise.
7307 * common/config/iq2000/iq2000-common.c: Likewise.
7308 * common/config/lm32/lm32-common.c: Likewise.
7309 * common/config/m32r/m32r-common.c: Likewise.
7310 * common/config/mcore/mcore-common.c: Likewise.
7311 * common/config/microblaze/microblaze-common.c: Likewise.
7312 * common/config/mips/mips-common.c: Likewise.
7313 * common/config/mmix/mmix-common.c: Likewise.
7314 * common/config/mn10300/mn10300-common.c: Likewise.
7315 * common/config/nios2/nios2-common.c: Likewise.
7316 * common/config/pa/pa-common.c: Likewise.
7317 * common/config/pdp11/pdp11-common.c: Likewise.
7318 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
7319 * common/config/riscv/riscv-common.c: Likewise.
7320 * common/config/rs6000/rs6000-common.c: Likewise.
7321 * common/config/rx/rx-common.c: Likewise.
7322 * common/config/s390/s390-common.c: Likewise.
7323 * common/config/sh/sh-common.c: Likewise.
7324 * common/config/sparc/sparc-common.c: Likewise.
7325 * common/config/tilegx/tilegx-common.c: Likewise.
7326 * common/config/tilepro/tilepro-common.c: Likewise.
7327 * common/config/v850/v850-common.c: Likewise.
7328 * common/config/visium/visium-common.c: Likewise.
7329 * common/config/xstormy16/xstormy16-common.c: Likewise.
7330 * common/config/xtensa/xtensa-common.c: Likewise.
7331 * invoke.texi (-fomit-frame-pointer): Update documentation.
7333 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7335 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
7337 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7339 * cfg.c (scale_bbs_frequencies_int,
7340 cale_bbs_frequencies_gcov_type): Remove.
7341 * cfg.h (scale_bbs_frequencies_int,
7342 cale_bbs_frequencies_gcov_type): Remove.
7344 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7346 * tree-ssa-loop-manip.c
7347 (scale_dominated_blocks_in_loop): Update to profile counts.
7348 (tree_transform_and_unroll_loop): Likewise.
7350 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7352 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
7353 scale_bbs_frequencies_int.
7355 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7357 * final.c (compute_alignments): Use counts rather than frequencies.
7359 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7361 * cfgloopanal.c: Include sreal.h
7362 (average_num_loop_insns): Use counts and sreal for accounting.
7364 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7366 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
7369 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7370 Kito Cheng <kito.cheng@gmail.com>
7372 * config/nds32/constraints.md: Provide more constraints.
7373 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
7374 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
7375 support constraints usage.
7377 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7379 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
7381 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7382 Kito Cheng <kito.cheng@gmail.com>
7384 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
7385 * config/nds32/nds32.opt: Refine the layout.
7386 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
7387 TARGET_EXT_STRING): Support new options.
7388 * config/nds32/nds32.h: Likewise.
7389 * config/nds32/nds32.md: Likewise.
7390 * config/nds32/nds32-predicates.c: Likewise.
7391 * config/nds32/constraints.md: Likewise.
7392 * common/config/nds32/nds32-common.c: Likewise.
7394 2017-11-16 Julia Koval <julia.koval@intel.com>
7397 * config/i386/gfniintrin.h: Add sse check.
7398 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
7400 2017-11-16 Julia Koval <julia.koval@intel.com>
7402 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
7403 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
7404 (ix86_handle_option): Handle -mavx512vbmi2.
7405 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
7406 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
7407 * config/i386/i386-c.c (__AVX512VBMI2__): New.
7408 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
7409 (ix86_valid_target_attribute_inner_p): Ditto.
7410 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
7411 * config/i386/i386.opt (mavx512vbmi2): New option.
7412 * doc/invoke.texi: Add new option.
7414 2017-11-16 Julia Koval <julia.koval@intel.com>
7416 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
7417 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
7418 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
7419 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
7420 _mm512_gf2p8mul_epi8): New intrinsics.
7421 * config/i386/i386-builtin-types.def
7422 (V64QI_FTYPE_V64QI_V64QI): New type.
7423 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
7424 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
7425 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
7426 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
7427 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
7428 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
7430 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
7432 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
7433 explicitly as a stream of bytes.
7435 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7437 * config/rs6000/altivec.h (vec_xst_be): New #define.
7438 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
7439 and externalize from *altivec_vperm_<mode>_internal.
7440 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
7442 (XL_BE_V8HI): Likewise.
7443 (XL_BE_V4SI): Likewise.
7444 (XL_BE_V4SI): Likewise.
7445 (XL_BE_V2DI): Likewise.
7446 (XL_BE_V4SF): Likewise.
7447 (XL_BE_V2DF): Likewise.
7448 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
7449 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
7450 all array entries with these keys: VSX_BUILTIN_VEC_XL,
7451 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
7452 VSX_BUILTIN_VEC_XST_BE.
7453 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
7454 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
7456 (altivec_init_builtins): Replace conditional calls to def_builtin
7457 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
7458 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
7459 with unconditional calls. Remove calls to def_builtin for
7460 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
7461 __builtin_vec_xst_be.
7462 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
7463 to define_expand, and add alternate RTL generation for P8.
7464 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
7465 vsx_ld_elemrev_v8hi.
7466 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
7467 add alternate RTL generation for P8.
7468 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
7469 vsx_ld_elemrev_v16qi.
7470 (vsx_st_elemrev_v8hi): Convert define_insn
7471 to define_expand, and add alternate RTL generation for P8.
7472 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
7473 vsx_st_elemrev_v8hi.
7474 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
7475 add alternate RTL generation for P8.
7476 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
7477 vsx_st_elemrev_v16qi.
7479 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
7482 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
7483 TARGET_AVX512ER check.
7484 (ix86_option_override_internal): Set MASK_VZEROUPPER if
7485 neither -mzeroupper nor -mno-zeroupper is used and
7486 TARGET_EMIT_VZEROUPPER is set.
7487 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
7488 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
7490 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
7492 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7493 folding of vector compares.
7494 (fold_build_vec_cmp): New helper function.
7495 (fold_compare_helper): New helper function.
7496 (builtin_function_type): Add compare builtins to the list of functions
7497 having unsigned arguments. Cosmetic updates to comment indentation.
7498 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
7499 the not+eq combination.
7501 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7503 PR tree-optimization/82726
7504 PR tree-optimization/70754
7505 * tree-predcom.c (order_drefs_by_pos): New function.
7506 (combine_chains): Move code setting has_max_use_after to...
7507 (try_combine_chains): ...here. New parameter. Sort combined chains
7508 according to position information.
7509 (tree_predictive_commoning_loop): Update call to above function.
7510 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
7512 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7514 PR tree-optimization/82726
7516 2017-01-23 Bin Cheng <bin.cheng@arm.com>
7518 PR tree-optimization/70754
7519 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
7520 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
7521 combined stmt before it if not NULL.
7522 (combine_chains): Process refs reversely and compute dominance point
7526 2017-02-23 Bin Cheng <bin.cheng@arm.com>
7528 PR tree-optimization/79663
7529 * tree-predcom.c (combine_chains): Process refs in reverse order
7530 only for ZERO length chains, and add explaining comment.
7532 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7534 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
7535 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
7536 * doc/invoke.texi (ARM Options): Add armv8.3-a.
7538 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7540 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
7542 2017-11-15 Martin Liska <mliska@suse.cz>
7544 * tree-cfg.c (pass_warn_function_return::execute):
7545 Compare warn_return_type for greater than zero.
7547 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
7551 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
7552 to return true on Xeon and not on Xeon Phi.
7553 (ix86_check_avx256_register): Changed to ...
7554 (ix86_check_avx_upper_register): ... this. Add extra check for
7555 VALID_AVX512F_REG_OR_XI_MODE.
7556 (ix86_avx_u128_mode_needed): Changed
7557 ix86_check_avx256_register to ix86_check_avx_upper_register.
7558 (ix86_check_avx256_stores): Changed to ...
7559 (ix86_check_avx_upper_stores): ... this. Changed
7560 ix86_check_avx256_register to ix86_check_avx_upper_register.
7561 (ix86_avx_u128_mode_after): Changed
7562 avx_reg256_found to avx_upper_reg_found. Changed
7563 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
7564 (ix86_avx_u128_mode_entry): Changed
7565 ix86_check_avx256_register to ix86_check_avx_upper_register.
7566 (ix86_avx_u128_mode_exit): Ditto.
7567 * config/i386/i386.h: (host_detect_local_cpu): New define.
7569 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7571 * config/arm/xgene1.md (xgene1): Split into automatons
7572 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
7573 (xgene1_f_load): Adjust reservations and/or types.
7574 (xgene1_f_store): Likewise.
7575 (xgene1_load_pair): Likewise.
7576 (xgene1_store_pair): Likewise.
7577 (xgene1_fp_load1): Likewise.
7578 (xgene1_load1): Likewise.
7579 (xgene1_store1): Likewise.
7580 (xgene1_move): Likewise.
7581 (xgene1_alu): Likewise.
7582 (xgene1_simd): Likewise.
7583 (xgene1_bfm): Likewise.
7584 (xgene1_neon_load1): Likewise.
7585 (xgene1_neon_store1): Likewise.
7586 (xgene1_neon_logic): Likewise.
7587 (xgene1_neon_st1): Likewise.
7588 (xgene1_neon_ld1r): Likewise.
7589 (xgene1_alu_cond): Added.
7590 (xgene1_shift_reg): Likwise.
7591 (xgene1_bfx): Likewise.
7592 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
7594 2017-11-15 Jakub Jelinek <jakub@redhat.com>
7597 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
7599 (can_widen_mult_without_libcall): New function.
7600 (expand_mul_overflow): If only checking unsigned mul overflow,
7601 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
7602 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
7603 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
7604 (expand_DIVMOD): Formatting fix.
7605 * expmed.h (expand_mult): Add NO_LIBCALL argument.
7606 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
7607 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
7609 PR tree-optimization/82977
7610 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
7611 constructed temporary to strlen_to_stridx.put.
7613 2017-11-15 Martin Liska <mliska@suse.cz>
7615 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
7616 * configure: Regenerate.
7618 2017-11-15 Martin Liska <mliska@suse.cz>
7621 * config/sh/sh-mem.cc: Use proper probability for
7624 2017-11-14 Jeff Law <law@redhat.com>
7626 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
7627 the red zone for stack_clash_protection_final_dynamic_probe targets
7628 when the total dynamic stack size is zero bytes.
7630 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
7631 blocks is post order.
7633 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
7635 * dumpfile.h (TDF_COMPARE_DEBUG): New.
7636 * final.c (rest_of_clean_state): Set it for the
7637 -fcompare-debug dump.
7638 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
7639 class when TDF_COMPARE_DEBUG is set.
7641 * dwarf2out.c (gen_producer_string): Discard
7644 2017-11-15 Joseph Myers <joseph@codesourcery.com>
7647 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
7648 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
7649 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
7650 (__TGMATH_REAL_2_3): Remove macros.
7651 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
7652 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
7654 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
7656 (remquo): Define using __TGMATH_REAL_3.
7658 2017-11-14 Jeff Law <law@redhat.com>
7660 * vr-values.c: New file with contents extracted from tree-vrp.c.
7661 * Makefile.in (OBJS): Add vr-values.o
7662 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
7663 (set_value_range, set_and_canonicalize_value_range): Likewise.
7664 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
7665 (value_range_constant_singleton, symbolic_range_p): Likewise.
7666 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
7667 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
7668 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
7669 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
7670 (find_case_label_range, find_case_label_index): Likewise.
7671 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
7672 (range_int_cst_singleton_p, value_inside_range): Likewise.
7673 (get_single_symbol): Likewise.
7674 (switch_update): Move structure definition here.
7675 (to_remove_edges, to_update_switch_stmts): Provide externs.
7676 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
7677 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
7678 (vrp_val_is_min, set_value_range): Likewise.
7679 (set_and_canonicalize_value_range, copy_value_range): Likewise.
7680 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
7681 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
7682 (range_is_nonnull, range_int_cst_p): Likewwise.
7683 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
7684 (get_single_symbol, operand_less_p): Likewise
7685 (compare_values_warnv, compare_values): Likewise.
7686 (value_inside_range, value_range_constant_singleton): Likewise.
7687 (zero_nonzero_bitgs_from_vr): Likewise.
7688 (extract_range_from_binary_expr_1): Likewise.
7689 (overflow_comparison_p): Likewise.
7690 (to_remove_edges, to_update_switch_stmts): Likewise.
7691 (find_case_label-index, find_case_label_range): Likewise.
7692 (switch_update, set_value_range_to_nonnegative): Remove.
7693 (set_value_range_to_truthvalue): Likewise.
7694 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
7695 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
7696 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
7697 (find_case_label_ranges, test_for_singularity): Likewise.
7698 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
7699 (x_vr_values): Move to its remaining use site.
7701 2017-11-10 Jeff Law <law@redhat.com>
7703 * vr-values.h (VR_INITIALIZER): Move #define here.
7704 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
7705 * Makefile.in (OBJS): Add tree-evrp.o
7706 * tree-vrp.h (assert_info): Move structure definition here.
7707 (set_value_range_to_varying): Prototype.
7708 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7709 (infer_value_range, register_edge_assert_for): Likewise.
7710 (stmt_interesting_for_vrp): Likewise.
7711 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
7712 (set_value_range_to_varying): No longer static.
7713 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7714 (infer_value_range, register_edge_assert_for): Likewise.
7716 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7718 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
7719 generate the XXBRD instruction.
7721 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
7722 (rs6000_builtin_type_compatible): Treat _Float128 and long double
7723 as being compatible if -mabi=ieeelongdouble.
7724 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
7725 to setup float128 built-ins with hardware support.
7726 (BU_FLOAT128_HW_2): Likewise.
7727 (BU_FLOAT128_HW_3): Likewise.
7728 (BU_FLOAT128_HW_VSX_1): Likewise.
7729 (BU_FLOAT128_HW_VSX_2): Likewise.
7730 (scalar_extract_expq): Change float128 built-in functions to
7731 accommodate having both KFmode and TFmode functions. Use the
7732 KFmode variant as the default.
7733 (scalar_extract_sigq): Likewise.
7734 (scalar_test_neg_qp): Likewise.
7735 (scalar_insert_exp_q): Likewise.
7736 (scalar_insert_exp_qp): Likewise.
7737 (scalar_test_data_class_qp): Likewise.
7738 (sqrtf128_round_to_odd): Delete processing the round to odd
7739 built-in functions as special built-in functions, and define them
7740 as float128 built-ins. Use the KFmode variant as the default.
7741 (truncf128_round_to_odd): Likewise.
7742 (addf128_round_to_odd): Likewise.
7743 (subf128_round_to_odd): Likewise.
7744 (mulf128_round_to_odd): Likewise.
7745 (divf128_round_to_odd): Likewise.
7746 (fmaf128_round_to_odd): Likewise.
7747 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7748 support for KFmode and TFmode xststdcqp calls.
7749 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
7750 point, switch the built-in handlers for the get/set float128
7751 exponent, get float128 mantissa, float128 test built-ins, and the
7752 float128 round to odd built-in functions. Eliminate creating the
7753 float128 round to odd built-in functions as special built-ins.
7754 (rs6000_init_builtins): Eliminate special creation of the float128
7755 round to odd built-in functions.
7756 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
7757 function insns to support both TFmode and KFmode variants.
7758 (xsxsigqp_<mode>): Likewise.
7759 (xsiexpqpf_<mode>): Likewise.
7760 (xsiexpqp_<mode>): Likewise.
7761 (xststdcqp_<mode>): Likewise.
7762 (xststdcnegqp_<mode>): Likewise.
7763 (xststdcqp_<mode>): Likewise.
7765 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7767 * tree-ssa-threadupdate.c (compute_path_counts): Remove
7768 unused path_in_freq_ptr parameter.
7769 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
7771 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7773 * ipa-inline.c (edge_badness): Dump sreal frequency.
7774 (compute_inlined_call_time): Match natural implementaiton ...
7775 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
7776 forgotten division by CGRAPH_FREQ_BASE.
7778 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7780 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
7781 Solaris 11. Update comment.
7782 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
7784 * config/sol2.h (STARTFILE_SPEC): Likewise.
7785 * configure: Regenerate.
7787 2017-11-14 Carl Love <cel@us.ibm.com>
7789 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
7790 le_ and be_ prefixes to swap* variables. Remove
7791 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
7794 2017-11-14 Jason Merrill <jason@redhat.com>
7796 Support GTY((cache)) on hash_map.
7797 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
7798 (ggc_cache_remove): Override it instead of ggc_mx.
7799 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
7800 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
7801 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
7802 (simple_cache_map_traits): Override maybe_mx.
7803 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
7804 (hash_map): Friend gt_cleare_cache.
7805 (gt_cleare_cache): New.
7806 * tree.h (tree_cache_traits): New hash_map traits class.
7807 (tree_cache_map): New typedef.
7809 2017-11-14 Richard Biener <rguenther@suse.de>
7811 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
7812 paramter and handling.
7813 (cleanup_control_flow_bb): Likewise.
7814 (cleanup_control_flow_pre): New helper performing a DFS walk
7815 to call cleanup_control_flow_bb in PRE order.
7816 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
7817 via cleanup_control_flow_pre.
7819 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
7821 * config/aarch64/aarch64-simd.md
7822 (aarch64_simd_bsl<mode>_internal): Remove DImode.
7823 (*aarch64_simd_bsl<mode>_alt): Likewise.
7824 (aarch64_simd_bsldi_internal): New.
7825 (aarch64_simd_bsldi_alt): Likewise.
7827 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7829 * tracer.c (better_p): Do not compare frequencies.
7830 * reg-stack.c (better_edge): Likewise.
7831 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
7834 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7836 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
7837 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
7838 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
7839 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7840 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
7841 * predict.c (maybe_hot_frequency_p): Inline to ...
7842 (maybe_hot_count_p): ... here; rewrite to counts.
7843 (counts_to_freqs): Rename to ...
7844 (update_max_bb_count): ... this one.
7845 (expensive_function_p): Use counts.
7846 (estimate_bb_frequencies): Update.
7847 (rebuild_frequencies): Update.
7848 * predict.h (counts_to_freqs): Rename to ...
7849 (update_max_bb_count): ... this one.
7850 * profile.c (compute_branch_probabilities): Add debug info
7851 * tree-inline.c (expand_call_inline): Update debug info.
7852 (optimize_inline_calls): Use update_max_bb_count..
7853 (tree_function_versioning): Use update_max_bb_count..
7854 * value-prof.c (gimple_value_profile_transformations):
7855 Do not use update_max_bb_count.
7857 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7859 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7860 always use frequencies.
7862 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7864 * bb-reorder.c: Remove frequencies from comments.
7865 (better_edge_p): Use profile counts.
7866 (find_traces): Dump profile counts.
7867 (rotate_loop): Use profile counts.
7868 (find_traces_1_round): Likewise.
7869 (connect_better_edge_p): Use counts instead of probabilities for
7871 (copy_bb_p): Drop early check for non-0 frequency.
7872 (sanitize_hot_paths): Update comments.
7874 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7876 * ipa-split.c (struct split_point): Add count.
7877 (consider_split): Do not compute incoming frequency; compute incoming
7878 count and store it to split_point.
7879 (split_function): Set count of the call to split part correctly.
7881 2017-11-13 Carl Love <cel@us.ibm.com>
7883 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7885 2017-11-13 Tom Tromey <tom@tromey.com>
7887 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7889 2017-11-13 Carl Love <cel@us.ibm.com>
7891 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7892 Add support for builtins:
7893 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7894 vector {un,}signed {char,int,short},
7895 vector {un,}signed {char,int,short}) arguments.
7896 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7897 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7898 Add BU_P9V_AV_2 expansions for the builtins.
7899 * config/rs6000/altivec.h (vec_first_match_index,
7900 vec_first_mismatch_index, vec_first_match_or_eos_index,
7901 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7902 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7903 new extern declaration.
7904 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7905 * config/rs6000/vsx.md (first_match_index_<mode>,
7906 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7907 first_mismatch_or_eos_index_<mode>): Add define expand.
7908 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7909 * doc/extend.texi: Update the built-in documenation file for the new
7912 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7914 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7915 into the min/max operations for _Float<N> and _Float<N>X types.
7917 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7920 * dwarf2out.c (do_eh_frame): New static variable.
7921 (dwarf2out_begin_prologue): Set it.
7922 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7924 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7926 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7928 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7930 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7933 * debug.h (dwarf2out_do_eh_frame): Declare.
7934 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7935 (dwarf2out_do_frame): Use it.
7936 (dwarf2out_do_cfi_asm): Likewise.
7937 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7938 (dwarf2out_assembly_start): Likewise.
7939 (dwarf2out_begin_prologue): Fix comment.
7940 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7941 if the target needs either debug or unwind DWARF2 info.
7942 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7943 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7944 -ftrapping-math, -ftrapv and -fwrapv.
7946 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7948 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7949 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7950 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7951 (estimate_edge_size_and_time): Likewise.
7952 (ipa_merge_fn_summary_after_inlining): Likewise.
7953 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7954 (compute_uninlined_call_time): Use sreal_frequency.
7955 (compute_inlined_call_time): Likewise.
7956 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7957 * profile-count.c: Include sreal.h.
7958 (profile_count::to_sreal_scale): New.
7959 * profile-count.h: Forward declare sreal.
7960 (profile_count::to_sreal_scale): Declare.
7962 2017-11-13 Nathan Sidwell <nathan@acm.org>
7964 * diagnostic.c (maybe_line_and_column): New.
7965 (diagnostic_get_location_text): Use it.
7966 (diagnostic_report_current_module): Likewise.
7967 (test_diagnostic_get_location_text): Add tests.
7969 2017-11-13 Luis Machado <luis.machado@linaro.org>
7971 * doc/md.texi (Specifying processor pipeline description): Fix
7972 incorrect latency for the div instruction example.
7974 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7976 PR tree-optimization/78821
7977 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7978 that bit_not_p is the same.
7979 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7980 (split_group): Count precisely bit_not_p bits in each statement.
7981 (invert_op): New function.
7982 (imm_store_chain_info::output_merged_store): Use invert_op to
7983 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7984 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7986 2017-11-13 Martin Liska <mliska@suse.cz>
7988 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7989 (struct source_info): Likewise.
7990 (add_branch_counts): Likewise.
7991 (add_line_counts): Likewise.
7992 (function_summary): Likewise.
7993 (output_intermediate_line): Likewise.
7994 (generate_results): Likewise.
7996 2017-11-13 Martin Liska <mliska@suse.cz>
7998 * gcov.c (struct block_info): Remove typedef for block_t.
7999 (struct line_info): Likewise.
8000 (line_info::has_block): Likewise.
8001 (EXIT_BLOCK): Likewise.
8002 (unblock): Likewise.
8003 (circuit): Likewise.
8004 (get_cycles_count): Likewise.
8005 (process_file): Likewise.
8006 (read_graph_file): Likewise.
8007 (solve_flow_graph): Likewise.
8008 (find_exception_blocks): Likewise.
8009 (add_line_counts): Likewise.
8010 (accumulate_line_info): Likewise.
8011 (output_line_details): Likewise.
8013 2017-11-13 Martin Liska <mliska@suse.cz>
8015 * gcov.c (struct arc_info): Remove typedef for arc_t.
8016 (struct line_info): Likewise.
8017 (add_branch_counts): Likewise.
8018 (output_branch_count): Likewise.
8019 (function_info::~function_info): Likewise.
8020 (circuit): Likewise.
8021 (output_intermediate_line): Likewise.
8022 (read_graph_file): Likewise.
8023 (solve_flow_graph): Likewise.
8024 (find_exception_blocks): Likewise.
8025 (add_line_counts): Likewise.
8026 (accumulate_line_info): Likewise.
8027 (output_line_details): Likewise.
8028 (output_function_details): Likewise.
8030 2017-11-13 Martin Liska <mliska@suse.cz>
8032 * gcov.c (struct function_info): Remove typedef for function_t.
8033 (struct source_info): Likewise.
8034 (source_info::get_functions_at_location): Likewise.
8035 (solve_flow_graph): Likewise.
8036 (find_exception_blocks): Likewise.
8037 (add_line_counts): Likewise.
8038 (output_intermediate_file): Likewise.
8039 (process_file): Likewise.
8040 (generate_results): Likewise.
8041 (release_structures): Likewise.
8042 (read_graph_file): Likewise.
8043 (read_count_file): Likewise.
8044 (accumulate_line_counts): Likewise.
8045 (output_lines): Likewise.
8047 2017-11-13 Martin Liska <mliska@suse.cz>
8049 * gcov.c (function_info::function_info): Remove num_counts
8050 and add vector<gcov_type>.
8051 (function_info::~function_info): Use the vector.
8052 (process_file): Likewise.
8053 (read_graph_file): Likewise.
8054 (read_count_file): Likewise.
8055 (solve_flow_graph): Likewise.
8057 2017-11-13 Martin Liska <mliska@suse.cz>
8059 * gcov.c (function_info::is_artificial): New function.
8060 (process_file): Erase all artificial early.
8061 (generate_results): Skip as all artificial are already
8064 2017-11-13 Martin Liska <mliska@suse.cz>
8066 * gcov.c (read_graph_file): Store to global vector of functions.
8067 (read_count_file): Iterate the vector.
8068 (process_file): Likewise.
8069 (generate_results): Likewise.
8070 (release_structures): Likewise.
8072 2017-11-13 Jakub Jelinek <jakub@redhat.com>
8074 PR tree-optimization/82954
8075 * gimple-ssa-store-merging.c
8076 (imm_store_chain_info::coalesce_immediate_stores): If
8077 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
8079 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
8081 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
8082 Upddate call to ENDIAN_LANE_N.
8083 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
8084 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
8085 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
8086 and use aarch64_endian_lane_rtx.
8087 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
8089 2017-11-12 Tom de Vries <tom@codesourcery.com>
8091 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
8093 2017-11-12 Tom de Vries <tom@codesourcery.com>
8095 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
8097 2017-11-12 Tom de Vries <tom@codesourcery.com>
8099 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
8101 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
8102 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
8103 * defaults.h (ASM_OUTPUT_LABELREF): Same.
8105 2017-11-11 Martin Sebor <msebor@redhat.com>
8108 * doc/extend.texi (attribute nonstring): Remove spurious argument.
8111 * prefic.c (translate_name): Replace strncpy with memcpy to
8112 avoid -Wstringop-truncation.
8114 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8116 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
8118 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8120 * predict.c (maybe_hot_frequency_p): Do not use cfun.
8122 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8124 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
8127 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
8129 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
8130 enable generating XXBRH if the value is in a vector register.
8131 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
8132 value is in a vector register.
8133 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
8134 register to register bswap64's instead of doing the GPR sequence
8135 used on previous machines.
8136 (bswapdi2_xxbrd): New insn.
8137 (bswapdi2_reg): Disallow on ISA 3.0.
8138 (register to register bswap64 splitter): Do not split the insn on
8139 ISA 3.0 systems that use XXBRD.
8141 2017-11-10 Martin Sebor <msebor@redhat.com>
8144 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
8146 (find_subframework_file): Same.
8148 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8150 * auto-profile.c (afdo_indirect_call): Drop frequency.
8151 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
8152 (cgraph_node::create_edge): Drop frequency argument.
8153 (cgraph_node::create_indirect_edge): Drop frequency argument.
8154 (cgraph_edge::make_speculative): Drop frequency arguments.
8155 (cgraph_edge::resolve_speculation): Do not update frequencies
8156 (cgraph_edge::dump_edge_flags): Do not dump frequency.
8157 (cgraph_node::dump): Check consistency in IPA mode.
8158 (cgraph_edge::maybe_hot_p): Use IPA counter.
8159 (cgraph_edge::verify_count_and_frequency): Rename to ...
8160 (cgraph_edge::verify_count): ... this one; drop frequency checking.
8161 (cgraph_node::verify_node): Update.
8162 * cgraph.h (struct cgraph_edge): Drop frequency.
8163 (cgraph_edge::frequency): New function.
8164 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
8166 (cgraph_edge::rebuild_edges): Likewise.
8167 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
8168 (duplicate_thunk_for_node): Do not pass frequency.
8169 (cgraph_node::create_clone): Scale only counts.
8170 (cgraph_node::create_virtual_clone): Do not pass frequency.
8171 (cgraph_node::create_edge_including_clones): Do not pass frequency.
8172 (cgraph_node::create_version_clone): Do not pass frequency.
8173 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
8174 (cgraph_node::expand_thunk): Do not pass frequency.
8175 (cgraph_node::create_wrapper): Do not pass frequency.
8176 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
8178 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
8179 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
8180 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
8181 (gather_caller_stats): Use frequency function.
8182 (ipcp_cloning_candidate_p): Use frequency function.
8183 (ipcp_propagate_stage): Use frequency function.
8184 (get_info_about_necessary_edges): Use frequency function.
8185 (update_profiling_info): Update only IPA profile.
8186 (update_specialized_profile): Use frequency functoin.
8187 (perhaps_add_new_callers): Update only IPA profile.
8188 * ipa-devirt.c (ipa_devirt): Use IPA profile.
8189 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
8190 (dump_ipa_call_summary): Use frequency function.
8191 (estimate_edge_size_and_time): Use frequency function.
8192 (ipa_merge_fn_summary_after_inlining): Use frequency function.
8193 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
8194 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
8195 (update_noncloned_counts): ... ths one; scale counts only.
8196 (clone_inlined_nodes): Do not scale frequency.
8197 (inline_call): Do not pass frequency.
8198 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
8199 (compute_inlined_call_time): Use IPA profile.
8200 (want_inline_small_function_p): Use IPA profile.
8201 (want_inline_self_recursive_call_p): Use IPA profile.
8202 (edge_badness): Use IPA profile.
8203 (lookup_recursive_calls): Use IPA profile.
8204 (recursive_inlining): Do not pass frequency.
8205 (resolve_noninline_speculation): Do not update frequency.
8206 (inline_small_functions): Collect max of IPA profile.
8207 (dump_overall_stats): Dump IPA porfile.
8208 (dump_inline_stats): Dump IPA porfile.
8209 (ipa_inline): Collect IPA stats.
8210 * ipa-inline.h (clone_inlined_nodes): Update prototype.
8211 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
8212 (ipa_propagate_frequency): Use frequency function.
8213 (ipa_profile): Cleanup.
8214 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
8215 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
8216 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
8217 (input_node): Do not stream frequency.
8218 (input_edge): Do not stream frequency.
8219 (merge_profile_summaries): Scale only IPA profiles.
8220 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
8221 * predict.c (drop_profile): Do not recompute frequency.
8222 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
8223 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
8224 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
8225 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
8226 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
8227 * tree-inline.c (copy_bb): Do not scale frequency.
8228 (expand_call_inline): Do not scale frequency.
8229 (tree_function_versioning): Do not scale frequency.
8230 * ubsan.c (ubsan_create_edge): Do not pass frequency.
8232 2017-11-10 Julia Koval <julia.koval@intel.com>
8234 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
8235 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
8236 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
8237 (_mm256_maskz_gf2p8affine_epi64_epi8)
8238 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
8239 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
8240 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
8241 (__builtin_ia32_vgf2p8affineqb_v32qi)
8242 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
8243 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
8245 2017-11-10 Tamar Christina <tamar.christina@arm.com>
8249 (arm_option_override): Refactor.
8250 (arm_option_reconfigure_globals): New.
8251 (arm_options_perform_arch_sanity_checks): New.
8252 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
8254 (arm_options_perform_arch_sanity_checks): Likewise
8256 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
8258 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
8259 (power9-qpmul): New.
8260 * rs6000/rs6000.md ("type" attr): Add qmul.
8261 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
8262 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
8263 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
8265 2017-11-10 Martin Sebor <msebor@redhat.com>
8268 * builtins.c (compute_objsize): Handle arrays that
8269 compute_builtin_object_size likes to fail for. Make extern.
8270 * builtins.h (compute_objsize): Declare.
8271 (check_strncpy_sizes): New function.
8272 (expand_builtin_strncpy): Call check_strncpy_sizes.
8273 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
8274 -Wstringop-truncation.
8275 (gimple_fold_builtin_strncat): Same.
8276 * gimple.c (gimple_build_call_from_tree): Set call location.
8277 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
8278 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
8279 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
8280 (handle_builtin_strlen): Use strlen_to_stridx.
8281 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
8283 Use strlen_to_stridx.
8284 (pass_strlen::execute): Release strlen_to_stridx.
8285 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
8286 (-Wstringop-truncation): Document new option.
8288 2017-11-10 Martin Liska <mliska@suse.cz>
8290 PR gcov-profile/82702
8291 * gcov.c (main): Handle intermediate files in a different
8293 (get_gcov_intermediate_filename): New function.
8294 (output_gcov_file): Remove support of intermediate files.
8295 (generate_results): Allocate intermediate file.
8296 (release_structures): Clean-up properly fn_end.
8297 (output_intermediate_file): Start iterating with line 1.
8299 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8301 PR tree-optimization/82929
8302 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8303 ops_swapped_p non-static data member.
8304 (store_immediate_info::store_immediate_info): Clear it.
8305 (imm_store_chain_info::coalesce_immediate_stores): If swapping
8306 ops set ops_swapped_p.
8307 (count_multiple_uses): Handle ops_swapped_p.
8309 2017-11-10 Martin Liska <mliska@suse.cz>
8311 * coverage.c (coverage_init): Stream information about
8312 support of has_unexecuted_blocks.
8313 * doc/gcov.texi: Document that.
8314 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
8315 * gcov.c (read_graph_file): Likewise.
8316 (output_line_beginning): Fix a small issue with
8319 2017-11-10 Bin Cheng <bin.cheng@arm.com>
8321 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
8322 reference of trivial component.
8324 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8327 * gimple-ssa-store-merging.c
8328 (pass_store_merging::terminate_all_aliasing_chains): For
8329 gimple_store_p stmts also call refs_output_dependent_p.
8331 PR rtl-optimization/82913
8332 * compare-elim.c (try_merge_compare): Punt if def_insn is not
8335 2017-11-09 Jeff Law <law@redhat.com>
8337 * vr-values.h: New file with vr_values class.
8338 * tree-vrp.c: Include vr-values.h
8339 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
8340 data objects into the vr_values class.
8341 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
8342 (get_value_range): Make it a member function within vr_values class.
8343 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
8344 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
8345 (op_with_constant_singleton_value_range): Likewise.
8346 (extract_range_for_var_from_comparison_expr): Likewise.
8347 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
8348 (extract_range_from_binary_expr): Likewise.
8349 (extract_range_from_unary_expr): Likewise.
8350 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
8351 (check_for_binary_op_overflow, extract_range_basic): Likewise.
8352 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
8353 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
8354 (compare_name_with_value, compare_names): Likewise.
8355 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
8356 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
8357 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
8358 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
8359 (extract_range_from_phi_node): Likewise.
8360 (simplify_truth_ops_using_ranges): Likewise.
8361 (simplify_div_or_mod_using_ranges): Likewise.
8362 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
8363 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
8364 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
8365 (simplify_float_conversion_using_ranges): Likewise.
8366 (simplify_internal_call_using_ranges): Likewise.
8367 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
8368 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
8369 poitner via x_vr_values for calls into gimple folder.
8370 (vrp_initialize_lattice): Make this the vr_values ctor.
8371 (vrp_free_lattice): Make this the vr_values dtor.
8372 (set_vr_value): New function.
8373 (class vrp_prop): Add vr_values data member. Add various member
8374 functions as well as member functions that delegate to vr_values.
8375 (check_array_ref): Make a member function within vrp_prop class.
8376 (search_for_addr_array, vrp_initialize): Likewise.
8377 (vrp_finalize): Likewise. Revamp to avoid direct access to
8378 vr_value, values_propagated, etc.
8379 (check_array_bounds): Extract vrp_prop class instance pointer from
8380 walk info structure. Use it to call member functions.
8381 (check_all_array_refs): Make a member function within vrp_prop class.
8382 Smuggle class instance pointer via walk info structure.
8383 (x_vr_values): New local static.
8384 (vrp_valueize): Use x_vr_values to get class instance.
8385 (vr_valueize_1): Likewise.
8386 (class vrp_folder): Add vr_values data member. Add various member
8387 functions as well as member functions that delegate to vr_values.
8388 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
8389 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
8390 class instance from vr_values. Use it to call member functions.
8391 (vrp_dom_walker): Add vr_values data member.
8392 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
8393 instance via x_vr_values.
8394 (identify_jump_threads): Accept vr_values as argument. Store
8395 it into the walker structure.
8396 (evrp_dom_walker): Add vr_values class data member. Add various
8398 (evrp_dom_walker::try_find_new_range): Use vr_values data
8399 member to access the memory allocator.
8400 (evrp_dom_walker::before_dom_children): Store vr_values class
8401 instance into the vrp_folder class.
8402 (evrp_dom_walker::push_value_range): Rework to avoid direct
8403 access to num_vr_values and vr_value.
8404 (evrp_dom_walker::pop_value_range): Likewise.
8405 (execute_early_vrp): Remove call to vrp_initialize_lattice.
8406 Use vr_values to get to dump_all_value_ranges member function.
8407 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
8408 and simplify_cond_using_ranges_2 via vrp_prop class instance.
8409 Pass vr_values class instance down to identify_jump_threads.
8410 Remove call to vrp_free_lattice.
8411 (debug_all_value_ranges): Remove.
8413 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
8414 (vrp_folder): Likewise.
8416 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
8417 Get it from the existing bitmap instead.
8418 (vrp_intersect_ranges_1): Likewise.
8420 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8422 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8424 (store_immediate_info::store_immediate_info): Add bitnotp argument,
8425 set bit_not_p to it.
8426 (imm_store_chain_info::coalesce_immediate_stores): Break group
8427 if bit_not_p is different.
8428 (count_multiple_uses, split_group,
8429 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
8430 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
8431 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
8432 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
8433 to store_immediate_info ctor.
8435 2017-11-09 Jim Wilson <jimw@sifive.com>
8437 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
8438 (scan_prog_file): Likewise.
8440 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8442 * bb-reorder.c (max_entry_frequency): Remove.
8443 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
8444 connect_traces, push_to_next_round_p): Remove prototypes.
8445 (find_traces_1_round): Use counts only.
8446 (push_to_next_round_p): Likewise.
8447 (find_traces): Likewise.
8448 (rotate_loop): Likewise.
8449 (find_traces_1_round): Likewise.
8450 (connect_traces): Likewise.
8451 (edge_order): Likewise.
8453 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
8455 * config/arm/arm.c (output_return_instruction): Add comments to
8456 indicate requirement for cmse_nonsecure_entry return to account
8457 for the size of clearing instruction output here.
8458 (thumb_exit): Likewise.
8459 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
8460 return in hardfloat mode.
8462 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
8464 * config/rs6000/rs6000.c (machine_function): Add a bool,
8465 "toc_is_wrapped_separately".
8466 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
8467 if it wasn't explicitly set or unset, we are optimizing for speed, and
8468 doing separate shrink-wrapping.
8469 (rs6000_get_separate_components): Enable the TOC component if
8470 saving the TOC register in the prologue.
8471 (rs6000_components_for_bb): Handle the TOC component.
8472 (rs6000_emit_prologue_components): Store the TOC register where needed.
8473 (rs6000_set_handled_components): Mark TOC as handled, if handled.
8474 (rs6000_emit_prologue): Don't save the TOC if that is already done.
8476 2017-11-09 Martin Jambor <mjambor@suse.cz>
8478 * ipa-param-manipulation.c: New file.
8479 * ipa-param-manipulation.h: Likewise.
8480 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
8481 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
8482 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
8483 (ipa_parm_adjustment): Likewise.
8484 (ipa_parm_adjustment_vec): Likewise.
8485 (ipa_get_vector_of_formal_parms): Moved declaration to
8486 ipa-param-manipulation.h.
8487 (ipa_get_vector_of_formal_parm_types): Likewise.
8488 (ipa_modify_formal_parameters): Likewise.
8489 (ipa_modify_call_arguments): Likewise.
8490 (ipa_combine_adjustments): Likewise.
8491 (ipa_dump_param_adjustments): Likewise.
8492 (ipa_modify_expr): Likewise.
8493 (ipa_get_adjustment_candidate): Likewise.
8494 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
8495 ipa-param-manipulation.c.
8496 (ipa_get_vector_of_formal_parm_types): Likewise.
8497 (ipa_modify_formal_parameters): Likewise.
8498 (ipa_modify_call_arguments): Likewise.
8499 (ipa_modify_expr): Likewise.
8500 (get_ssa_base_param): Likewise.
8501 (ipa_get_adjustment_candidate): Likewise.
8502 (index_in_adjustments_multiple_times_p): Likewise.
8503 (ipa_combine_adjustments): Likewise.
8504 (ipa_dump_param_adjustments): Likewise.
8505 * tree-sra.c: Also include ipa-param-manipulation.h
8506 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
8509 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8510 Alan Hayward <alan.hayward@arm.com>
8511 David Sherwood <david.sherwood@arm.com>
8513 * doc/sourcebuild.texi (vect_masked_store): Document.
8515 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8516 Alan Hayward <alan.hayward@arm.com>
8517 David Sherwood <david.sherwood@arm.com>
8519 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
8521 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8522 Alan Hayward <alan.hayward@arm.com>
8523 David Sherwood <david.sherwood@arm.com>
8525 * doc/sourcebuild.texi (vect_variable_length): Document.
8527 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8528 Alan Hayward <alan.hayward@arm.com>
8529 David Sherwood <david.sherwood@arm.com>
8531 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
8533 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8534 Alan Hayward <alan.hayward@arm.com>
8535 David Sherwood <david.sherwood@arm.com>
8537 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
8539 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8540 Alan Hayward <alan.hayward@arm.com>
8541 David Sherwood <david.sherwood@arm.com>
8543 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
8544 previously undocumented selectors.
8545 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
8547 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8549 * doc/rtl.texi (const_vector): Say that elements can be
8550 const_wide_ints too.
8551 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
8552 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
8553 (gen_vec_duplicate): Use it instead of CONSTANT_P.
8554 * optabs.c (expand_vector_broadcast): Likewise.
8556 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8557 Alan Hayward <alan.hayward@arm.com>
8558 David Sherwood <david.sherwood@arm.com>
8560 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
8561 scaled index even if the unscaled address was invalid.
8562 Don't increase the complexity of using a scale in that case.
8564 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8565 Alan Hayward <alan.hayward@arm.com>
8566 David Sherwood <david.sherwood@arm.com>
8568 * doc/rtl.texi: Rewrite the subreg rules so that they partition
8569 the inner register into REGMODE_NATURAL_SIZE bytes rather than
8570 UNITS_PER_WORD bytes.
8571 * emit-rtl.c (validate_subreg): Divide subregs into blocks
8572 based on REGMODE_NATURAL_SIZE of the inner mode.
8573 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
8574 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
8575 * expmed.c (lowpart_bit_field_p): Divide the value up into
8576 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
8577 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
8578 whether something is likely to occupy more than one register.
8580 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8583 * ipa-inline-transform.c (update_noncloned_frequencies): Use
8584 profile_count::adjust_for_ipa_scaling.
8585 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
8586 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
8588 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
8590 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8592 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
8593 (split_group): Add total_orig and total_new arguments, estimate the
8594 number of statements related to the store group without store merging
8595 and with store merging.
8596 (imm_store_chain_info::output_merged_store): Adjust split_group
8597 callers, punt if estimated number of statements with store merging
8598 is not smaller than estimated number of statements without it.
8600 (handled_load): Remove has_single_use checks.
8601 (pass_store_merging::process_store): Likewise.
8603 2017-11-09 Richard Biener <rguenther@suse.de>
8605 PR tree-optimization/82902
8606 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
8608 2017-11-09 Martin Liska <mliska@suse.cz>
8611 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
8614 2017-11-09 Martin Liska <mliska@suse.cz>
8616 PR tree-optimization/82669
8617 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
8619 2017-11-09 Martin Liska <mliska@suse.cz>
8621 PR gcov-profile/48463
8622 * coverage.c (coverage_begin_function): Output also end locus
8623 of a function and information whether the function is
8625 * gcov-dump.c (tag_function): Parse and print the information.
8626 * gcov.c (INCLUDE_MAP): Add include.
8627 (INCLUDE_SET): Likewise.
8628 (struct line_info): Move earlier in the source file because
8629 of vector<line_info> in function_info structure.
8630 (line_info::line_info): Likewise.
8631 (line_info::has_block): Likewise.
8632 (struct source_info): Add new member index.
8633 (source_info::get_functions_at_location): New function.
8634 (function_info::group_line_p): New function.
8635 (output_intermediate_line): New function.
8636 (output_intermediate_file): Use the mentioned function.
8637 (struct function_start): New.
8638 (struct function_start_pair_hash): Likewise.
8639 (process_file): Add code that identifies group functions.
8640 Assign lines either to global or function scope.
8641 (generate_results): Skip artificial functions.
8642 (find_source): Assign index for each source file.
8643 (read_graph_file): Read new flag artificial and end_line.
8644 (add_line_counts): Assign it either to global of function scope.
8645 (accumulate_line_counts): Isolate core of the function to
8646 accumulate_line_info and call it for both function and global
8648 (accumulate_line_info): New function.
8649 (output_line_beginning): Fix GNU coding style.
8650 (print_source_line): New function.
8651 (output_line_details): Likewise.
8652 (output_function_details): Likewise.
8653 (output_lines): Iterate both source (global) scope and function
8655 (struct function_line_start_cmp): New class.
8656 * doc/gcov.texi: Reflect changes in documentation.
8658 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8661 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
8662 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
8663 and similarly for not instead of neg.
8665 2017-11-08 Andi Kleen <ak@linux.intel.com>
8667 * config/i386/i386.opt: Add -mforce-indirect-call.
8668 * config/i386/predicates.md: Check for flag_force_indirect_call.
8669 * doc/invoke.texi: Document -mforce-indirect-call
8671 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
8673 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
8675 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
8676 config/riscv/riscv.c (predict.h): New include.
8677 (riscv_slow_unaligned_access_p): No longer static.
8678 (riscv_block_move_straight): Add require.
8679 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8681 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8684 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
8685 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
8686 nonimmediate_operand predicate for operand 1 instead of
8689 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8691 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
8693 * config/aarch64/constraints.md (Uml): New constraint.
8694 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
8697 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8699 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
8700 of two vec_duplicates into a vec_concat.
8702 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8704 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8705 Simplify vec_merge of vec_duplicate and vec_concat.
8706 * config/aarch64/constraints.md (Utq): New constraint.
8707 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
8710 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8712 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8713 Simplify vec_merge of vec_duplicate and const_vector.
8714 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
8716 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
8717 mode iterator. Update predicate on operand 1 to
8718 handle non-const_vec constants. Delete constraints.
8719 (*aarch64_combinez_be<mode>): Likewise for operand 2.
8721 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8723 PR tree-optimization/78821
8724 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
8726 (store_operand_info::store_operand_info): Initialize it to false.
8727 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
8728 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
8729 store in the group, and if chain_info is non-NULL, to ignore altogether
8731 (compatible_load_p): Fail if bit_not_p does not match.
8732 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
8733 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
8734 (pass_store_merging::process_store): Adjust
8735 terminate_all_aliasing_chains calls to pass NULL in all current spots,
8736 call terminate_all_aliasing_chains newly when adding a store into
8737 a chain with non-NULL chain_info.
8739 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8741 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
8743 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8745 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8747 (aarch64_layout_frame): Initialise emit_frame_chain.
8748 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
8749 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
8751 2017-11-08 Martin Liska <mliska@suse.cz>
8753 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
8754 of gimple_call_internal_p.
8756 2017-11-07 Tom de Vries <tom@codesourcery.com>
8758 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
8760 2017-11-07 Tom de Vries <tom@codesourcery.com>
8762 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
8765 2017-11-08 Martin Liska <mliska@suse.cz>
8768 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
8770 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8772 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
8774 2017-11-07 Jakub Jelinek <jakub@redhat.com>
8777 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
8778 (*cmp<mode>_ccz_1): New insn with $k alternative.
8781 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
8782 RTX_COMM_COMPARE as commutative as well.
8783 (ix86_binary_operator_ok): Formatting fix.
8784 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
8785 *<code><mode>3<mask_name><round_saeonly_name>,
8786 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
8787 *<s>mul<mode>3_highpart<mask_name>,
8788 *vec_widen_umult_even_v16si<mask_name>,
8789 *vec_widen_umult_even_v8si<mask_name>,
8790 *vec_widen_umult_even_v4si<mask_name>,
8791 *vec_widen_smult_even_v16si<mask_name>,
8792 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
8793 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
8794 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
8795 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
8796 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
8797 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
8798 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
8799 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
8800 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
8801 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
8802 ix86_binary_operator_ok. Formatting fixes.
8803 (*<plusminus_insn><mode>3<mask_name><round_name>,
8804 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
8807 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
8809 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
8810 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
8811 eq and ne if TARGET_ISEL.
8812 (cmp): New code_iterator.
8813 (UNS, UNSU_, UNSIK): New code_attrs.
8814 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
8815 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
8817 ("eq<mode>3"): ... this.
8818 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
8820 ("ne<mode>3"): ... this.
8822 2017-11-07 Julia Koval <julia.koval@intel.com>
8825 * common/config/i386/i386-common.c
8826 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
8827 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
8828 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
8829 * config/i386/i386.opt: Ditto.
8830 * config/i386/i386.c (ix86_target_string): Ditto.
8831 (ix86_option_override_internal): Ditto.
8832 (ix86_init_mpx_builtins): Move MPX to args2.
8833 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
8834 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8835 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8836 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8837 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8838 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8839 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
8841 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
8844 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
8845 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
8846 (zero-extendsidi peephole2): Remove peephole.
8848 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8851 * stor-layout.c: Include attribs.h.
8852 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8853 explicit lookup of "aligned" attribute.
8855 2017-11-07 Andrew Waterman <andrew@sifive.com>
8857 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8858 (riscv_expand_block_move): Likewise.
8859 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8861 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8862 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8863 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8864 (riscv_adjust_block_mem): Likewise.
8865 (riscv_block_move_loop): Likewise.
8866 (riscv_expand_block_move): Likewise.
8867 * config/riscv/riscv.md (movmemsi): New pattern.
8869 2017-11-07 Michael Clark <michaeljclark@mac.com>
8871 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8872 (MUSL_DYNAMIC_LINKER): Likewise.
8874 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8876 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8877 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8880 2017-11-07 Richard Biener <rguenther@suse.de>
8882 * match.pd: Fix build.
8884 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8885 Jackson Woodruff <jackson.woodruff@arm.com>
8887 PR tree-optimization/71026
8888 * match.pd: Canonicalize negate in division.
8890 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8893 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8895 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8897 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8898 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8900 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8902 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8903 non-scalar integral types.
8904 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8905 (-(A-B), -(~A)): New transformations.
8907 2017-11-07 Tom de Vries <tom@codesourcery.com>
8909 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8910 semicolon after "do {} while (0)".
8911 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8912 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8913 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8914 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8915 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8917 2017-11-07 Tom de Vries <tom@codesourcery.com>
8919 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8920 after "do {} while (0)".
8921 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8922 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8923 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8924 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8925 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8927 2017-11-07 Tom de Vries <tom@codesourcery.com>
8930 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8932 (arm_rtx_costs_internal): Add missing semicolon after
8933 HANDLE_NARROW_SHIFT_ARITH call.
8935 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8937 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8938 disable isel if it was not set explicitly.
8940 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8942 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8943 (add_type_attribute) likewise.
8945 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8947 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8948 of struct ix86_frame.
8949 (ix86_initial_elimination_offset): Likewise.
8950 (ix86_expand_split_stack_prologue): Likewise.
8952 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8954 * tree-vrp.h (enum value_range_type): Update stale comment.
8956 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8957 Alan Hayward <alan.hayward@arm.com>
8958 David Sherwood <david.sherwood@arm.com>
8960 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8961 (aarch64_expand_vec_perm_const): Take the number of units too.
8962 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8963 (aarch64_expand_vec_perm_const): Likewise.
8964 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8965 (vec_perm<mode>): Update accordingly.
8967 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8968 Alan Hayward <alan.hayward@arm.com>
8969 David Sherwood <david.sherwood@arm.com>
8971 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8972 Take the number of units too.
8973 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8974 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8975 but check for a vector mode before rather than after the call.
8976 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8977 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8978 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8979 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8980 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8981 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8982 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8983 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8984 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8985 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8986 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8987 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8988 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8989 (aarch64_sqdmull2_n<mode>): Update accordingly.
8991 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8992 Alan Hayward <alan.hayward@arm.com>
8993 David Sherwood <david.sherwood@arm.com>
8995 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8996 the number of units too.
8997 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8998 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8999 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
9000 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
9001 (vec_store_lanesxi<mode>): Update accordingly.
9003 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
9004 Alan Hayward <alan.hayward@arm.com>
9005 David Sherwood <david.sherwood@arm.com>
9007 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
9008 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
9009 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
9010 of units rather than the mode.
9011 * config/aarch64/iterators.md (nunits): New mode attribute.
9012 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
9013 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
9014 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
9015 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
9016 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
9017 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
9018 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
9019 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
9020 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
9021 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
9022 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
9023 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
9024 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
9025 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
9026 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
9027 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
9028 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
9029 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
9030 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
9031 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
9032 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
9033 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
9034 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
9035 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
9036 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
9037 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
9038 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
9039 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
9040 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
9041 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
9042 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
9043 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
9044 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
9045 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
9046 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
9047 (aarch64_simd_vec_setv2di): Likewise.
9049 2017-11-06 Carl Love <cel@us.ibm.com>
9051 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
9053 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
9054 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
9056 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
9058 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
9059 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
9060 Add power 8 macro expansions.
9061 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
9062 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
9063 power 8 instructions. (VSX_XXBR): Add iterator.
9065 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
9067 * config/arm/arm.md (predicable_short_it): Change default to "no",
9068 improve documentation, remove uses that are identical to the default.
9069 (enabled_for_depr_it): Rename to enabled_for_short_it.
9070 * gcc/config/arm/arm-fixed.md (predicable_short_it):
9071 Remove default uses.
9072 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
9073 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
9074 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
9075 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
9077 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
9080 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
9081 float128 helper macros, which are no longer used after deleting
9082 the old 'q' built-in functions, and moving the round to odd
9083 built-in functions to being special built-in functions.
9084 (BU_FLOAT128_2): Likewise.
9085 (BU_FLOAT128_1_HW): Likewise.
9086 (BU_FLOAT128_2_HW): Likewise.
9087 (BU_FLOAT128_3_HW): Likewise.
9088 (FABSQ): Delete old 'q' built-in functions.
9089 (COPYSIGNQ): Likewise.
9090 (SQRTF128_ODD): Move round to odd built-in functions to be
9091 special built-in functions, so that we can handle
9092 -mabi=ieeelongdouble.
9093 (TRUNCF128_ODD): Likewise.
9094 (ADDF128_ODD): Likewise.
9095 (SUBF128_ODD): Likewise.
9096 (MULF128_ODD): Likewise.
9097 (DIVF128_ODD): Likewise.
9098 (FMAF128_ODD): Likewise.
9099 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
9100 built-in names to 'f128'.
9101 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
9102 old 'q' built-in functions, as the machine independent code for
9103 'f128' built-in functions handles this.
9104 (rs6000_expand_builtin): Add expansion for float128 round to odd
9105 functions, keying off on -mabi=ieeelongdouble of whether to use
9106 the KFmode or TFmode variant.
9107 (rs6000_init_builtins): Initialize the _Float128 round to odd
9109 * doc/extend.texi (PowerPC Built-in Functions): Document the old
9110 _Float128 'q' built-in functions are now mapped into the new
9111 'f128' built-in functions.
9113 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
9115 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
9116 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
9118 2017-11-06 David Malcolm <dmalcolm@redhat.com>
9121 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
9122 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
9123 * toplev.c: Include "ipa-fnsummary.h".
9124 (toplev::finalize): Call ipa_fnsummary_c_finalize.
9126 2017-11-06 Jakub Jelinek <jakub@redhat.com>
9128 PR tree-optimization/82838
9129 * gimple-ssa-store-merging.c
9130 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
9131 on a separate gimple_seq which is then appended to seq.
9133 2017-11-06 Jeff Law <law@redhat.com>
9136 * config/i386/i386.c (PROBE_INTERVAL): Remove.
9137 (get_probe_interval): New functions.
9138 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
9139 (ix86_adjust_stack_and_probe): Likewise.
9140 (output_adjust_stack_and_probe): Likewise.
9141 (ix86_emit_probe_stack_range): Likewise.
9142 (ix86_expand_prologue): Likewise.
9144 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
9146 PR tree-optimization/82816
9147 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
9148 if the modes of the two types are the same.
9149 (convert_plusminus_to_widen): Likewise.
9151 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9153 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
9155 (usadv16qi): New define_expand.
9156 (usadv8hi): New define_expand.
9158 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
9161 * ipa-inline-transform.c (update_noncloned_frequencies): Always
9163 (inline_transform): Likewise.
9164 * predict.c (counts_to_freqs): Remove useless conditional.
9165 * profile-count.h (profile_count::apply_scale): Move sanity check.
9166 * tree-inline.c (copy_bb): Always scale.
9167 (copy_cfg_body): Likewise.
9169 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
9172 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
9174 (*cmp_ite0): Add enabled_for_depr_it attribute.
9175 (*cmp_ite1): Likewise.
9177 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
9179 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
9182 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
9184 * tree-vrp.c (vrp_int_const_binop): Return true on success and
9185 return the value by pointer.
9186 (extract_range_from_multiplicative_op_1): Update accordingly.
9187 Return as soon as an operation fails.
9189 2017-11-05 Tom de Vries <tom@codesourcery.com>
9192 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
9193 (DEF_SANITIZER_BUILTIN): ... here.
9194 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
9195 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
9197 2017-11-05 Tom de Vries <tom@codesourcery.com>
9200 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
9202 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
9203 ASM_OUTPUT_BEFORE_CASE_LABEL call.
9204 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
9206 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9207 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9208 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9210 2017-11-05 Tom de Vries <tom@codesourcery.com>
9213 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
9216 2017-11-04 Michael Clark <michaeljclark@mac.com>
9218 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
9219 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
9221 (*addsi3_extended): Likewise.
9222 (*addsi3_extended2): Likewise.
9223 (<optab>si3): Likewise.
9224 (<optab>di3): Likewise.
9225 (<optab><mode>3): Likewise.
9226 (<*optabe>si3_internal): Likewise.
9227 (zero_extendqi<SUPERQI:mode>2): Likewise.
9228 (*add<mode>hi3): Likewise.
9229 (*xor<mode>hi3): Likewise.
9230 (<optab>di3): Likewise.
9231 (*<optab>si3_extend): Likewise.
9232 (*sge<u>_<X:mode><GPR:mode>): Likewise.
9233 (*slt<u>_<X:mode><GPR:mode>): Likewise.
9234 (*sle<u>_<X:mode><GPR:mode>): Likewise.
9236 2017-11-04 Andrew Waterman <andrew@sifive.com>
9238 * config/riscv/riscv.c (riscv_option_override): Conditionally set
9239 TARGET_STRICT_ALIGN based upon -mtune argument.
9241 2017-11-04 Andrew Waterman <andrew@sifive.com>
9243 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
9245 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
9247 * config/i386/i386.c (choose_basereg): Use optional scratch
9248 register and add assertion.
9249 (x86_emit_outlined_ms2sysv_save): Use scratch register when
9250 needed, and don't allocate stack.
9251 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
9252 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
9253 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
9255 2017-11-03 Jeff Law <law@redhat.com>
9257 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
9258 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
9259 to probe at the start of a noreturn function.
9261 2017-11-03 Jakub Jelinek <jakub@redhat.com>
9263 PR tree-optimization/78821
9264 * gimple-ssa-store-merging.c: Update the file comment.
9265 (MAX_STORE_ALIAS_CHECKS): Define.
9266 (struct store_operand_info): New type.
9267 (store_operand_info::store_operand_info): New constructor.
9268 (struct store_immediate_info): Add rhs_code and ops data members.
9269 (store_immediate_info::store_immediate_info): Add rhscode, op0r
9270 and op1r arguments to the ctor, initialize corresponding data members.
9271 (struct merged_store_group): Add load_align_base and load_align
9273 (merged_store_group::merged_store_group): Initialize them.
9274 (merged_store_group::do_merge): Update them.
9275 (merged_store_group::apply_stores): Pick the constant for
9276 encode_tree_to_bitpos from one of the two operands, or skip
9277 encode_tree_to_bitpos if neither operand is a constant.
9278 (class pass_store_merging): Add process_store method decl. Remove
9279 bool argument from terminate_all_aliasing_chains method decl.
9280 (pass_store_merging::terminate_all_aliasing_chains): Remove
9281 var_offset_p argument and corresponding handling.
9282 (stmts_may_clobber_ref_p): New function.
9283 (compatible_load_p): New function.
9284 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
9285 if there is overlap and rhs_code is not INTEGER_CST. For
9286 non-overlapping stores terminate group if rhs is not mergeable.
9287 (get_alias_type_for_stmts): Change first argument from
9288 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
9289 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
9290 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
9292 (get_location_for_stmts): Change first argument from
9293 auto_vec<gimple *> & to vec<gimple *> &.
9294 (struct split_store): Remove orig_stmts data member, add orig_stores.
9295 (split_store::split_store): Create orig_stores rather than orig_stmts.
9296 (find_constituent_stmts): Renamed to ...
9297 (find_constituent_stores): ... this. Change second argument from
9298 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
9299 to info structures rather than the statements.
9300 (split_group): Rename ALLOW_UNALIGNED argument to
9301 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
9302 it. Adjust find_constituent_stores caller.
9303 (imm_store_chain_info::output_merged_store): Handle rhs_code other
9304 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
9305 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
9306 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
9307 (mem_valid_for_store_merging): New function.
9308 (handled_load): New function.
9309 (pass_store_merging::process_store): New method.
9310 (pass_store_merging::execute): Use process_store method. Adjust
9311 terminate_all_aliasing_chains caller.
9313 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9315 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9316 Return true for more constants, symbols and label references.
9317 (aarch64_valid_floating_const): Remove unused function.
9319 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9322 * config/aarch64/aarch64.c (aarch64_layout_frame):
9323 Undo forcing of LR at bottom of frame.
9325 2017-11-03 Jeff Law <law@redhat.com>
9328 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
9329 for int_registers_saved.
9331 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
9332 extracted from tree-ssa-dom.c.
9333 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
9334 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
9335 (record_equivalences_from_incoming_edge): Add additional argument
9336 to single_pred_edge_ignoring_loop_edges call.
9337 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
9338 (uncprop_dom_walker::before_dom_children): Add additional argument
9339 to single_pred_edge_ignoring_loop_edges call.
9340 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
9341 single_pred_edge_ignoring_loop_edges rather than open coding.
9342 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
9344 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
9346 * match.pd (-(-A)): Rewrite.
9348 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9350 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
9351 (rs6000_emit_int_cmove): New declaration.
9352 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
9353 (rs6000_emit_sISEL): Delete.
9354 (rs6000_emit_int_cmove): Make non-static.
9355 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
9356 instead of rs6000_emit_sISEL.
9358 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
9360 * asan.c (create_cond_insert_point): Maintain profile.
9361 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
9363 * basic-block.h (struct basic_block_def): Remove frequency.
9364 (EDGE_FREQUENCY): Use to_frequency
9365 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
9367 (find_traces): Update to use to_frequency.
9368 (find_traces_1_round): Likewise; use only IPA counts.
9369 (bb_to_key): Likewise.
9370 (connect_traces): Use IPA counts only.
9371 (copy_bb_p): Update to use to_frequency.
9372 (fix_up_crossing_landing_pad): Likewise.
9373 (sanitize_hot_paths): Likewise.
9374 * bt-load.c (basic_block_freq): Likewise.
9375 * cfg.c (init_flow): Set count_max to uninitialized.
9376 (check_bb_profile): Remove frequencies; check counts.
9377 (dump_bb_info): Do not dump frequencies.
9378 (update_bb_profile_for_threading): Update counts only.
9379 (scale_bbs_frequencies_int): Likewise.
9380 (MAX_SAFE_MULTIPLIER): Remove.
9381 (scale_bbs_frequencies_gcov_type): Update counts only.
9382 (scale_bbs_frequencies_profile_count): Update counts only.
9383 (scale_bbs_frequencies): Update counts only.
9384 * cfg.h (struct control_flow_graph): Add count-max.
9385 (update_bb_profile_for_threading): Update prototype.
9386 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
9387 (find_many_sub_basic_blocks): Likewise.
9388 * cfgcleanup.c (try_forward_edges): Likewise.
9389 (try_crossjump_to_edge): Likewise.
9390 * cfgexpand.c (expand_gimple_cond): Likewise.
9391 (expand_gimple_tailcall): Likewise.
9392 (construct_init_block): Likewise.
9393 (construct_exit_block): Likewise.
9394 * cfghooks.c (verify_flow_info): Check consistency of counts.
9395 (dump_bb_for_graph): Do not dump frequencies.
9396 (split_block_1): Do not update frequencies.
9397 (split_edge): Do not update frequencies.
9398 (make_forwarder_block): Do not update frequencies.
9399 (duplicate_block): Do not update frequencies.
9400 (account_profile_record): Do not update frequencies.
9401 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
9402 for global heuristics.
9403 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
9404 (expected_loop_iterations_unbounded): Use counts only.
9405 * cfgloopmanip.c (scale_loop_profile): Simplify.
9406 (create_empty_loop_on_edge): Simplify
9408 (duplicate_loop_to_header_edge): Simplify
9409 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
9410 (update_br_prob_note): Take care of removing note when profile
9412 (relink_block_chain): Do not dump frequency.
9413 (rtl_account_profile_record): Use to_frequency.
9414 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
9415 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
9416 (cgraph_update_edges_for_call_stmt_node): Likewise.
9417 (cgraph_edge::verify_count_and_frequency): Update.
9418 (cgraph_node::verify_node): Temporarily disable frequency verification.
9419 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
9420 to_cgraph_frequency.
9421 (cgraph_edge::rebuild_edges): Convert to ipa counts.
9422 * cgraphunit.c (init_lowered_empty_function): Do not initialize
9424 (cgraph_node::expand_thunk): Update profile.
9425 * except.c (dw2_build_landing_pads): Do not update frequency.
9426 * final.c (compute_alignments): Use to_frequency.
9427 (dump_basic_block_info): Do not dump frequency.
9428 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
9429 (dump_gimple_bb_header): Do not dump frequency.
9430 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
9432 * gimple-streamer-in.c (input_bb): Do not stream frequency.
9433 * gimple-streamer-out.c (output_bb): Do not stream frequency.
9434 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
9435 (init_before_recovery): Do not update frequency.
9436 (sched_create_recovery_edges): Do not update frequency.
9437 * hsa-gen.c (convert_switch_statements): Do not update frequency.
9438 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
9439 (ipa_cp_c_finalize): Set max_count to uninitialized.
9440 * ipa-fnsummary.c (get_minimal_bb): Use counts.
9441 (param_change_prob): Use counts.
9442 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
9444 * ipa-split.c (consider_split): Use to_frequency.
9445 (split_function): Use to_frequency.
9446 * ira-build.c (loop_compare_func): Likewise.
9447 (mark_loops_for_removal): Likewise.
9448 (mark_all_loops_for_removal): Likewise.
9449 * loop-doloop.c (doloop_modify): Do not update frequency.
9450 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
9452 * lto-streamer-in.c (input_function): Update count_max.
9453 * omp-expand.c (expand_omp_taskreg): Update count_max.
9454 * omp-simd-clone.c (simd_clone_adjust): Update profile.
9455 * predict.c (maybe_hot_frequency_p): Use to_frequency.
9456 (maybe_hot_count_p): Use ipa counts only.
9457 (maybe_hot_bb_p): Simplify.
9458 (maybe_hot_edge_p): Simplify.
9459 (probably_never_executed): Do not take frequency argument.
9460 (probably_never_executed_bb_p): Do not pass frequency.
9461 (probably_never_executed_edge_p): Likewise.
9462 (combine_predictions_for_bb): Check that profile is nonzero.
9463 (propagate_freq): Do not set frequency.
9464 (drop_profile): Simplify.
9465 (counts_to_freqs): Simplify.
9466 (expensive_function_p): Use to_frequency.
9467 (propagate_unlikely_bbs_forward): Simplify.
9468 (determine_unlikely_bbs): Simplify.
9469 (estimate_bb_frequencies): Add hack to silence graphite issues.
9470 (compute_function_frequency): Use ipa counts.
9471 (pass_profile::execute): Update.
9472 (rebuild_frequencies): Use counts only.
9473 (force_edge_cold): Use counts only.
9474 * profile-count.c (profile_count::dump): Dump new count types.
9475 (profile_count::differs_from_p): Check compatiblity.
9476 (profile_count::to_frequency): New function.
9477 (profile_count::to_cgraph_frequency): New function.
9478 * profile-count.h (struct function): Declare.
9479 (enum profile_quality): Add profile_guessed_local and
9480 profile_guessed_global0.
9481 (class profile_proability): Decrease number of bits to 29;
9482 update from_reg_br_prob_note and to_reg_br_prob_note.
9483 (class profile_count: Update comment; decrease number of bits
9484 to 61. Check compatibility.
9485 (profile_count::compatible_p): New private member function.
9486 (profile_count::ipa_p): New member function.
9487 (profile_count::operator<): Handle global zero correctly.
9488 (profile_count::operator>): Handle global zero correctly.
9489 (profile_count::operator<=): Handle global zero correctly.
9490 (profile_count::operator>=): Handle global zero correctly.
9491 (profile_count::nonzero_p): New member function.
9492 (profile_count::force_nonzero): New member function.
9493 (profile_count::max): New member function.
9494 (profile_count::apply_scale): Handle IPA scalling.
9495 (profile_count::guessed_local): New member function.
9496 (profile_count::global0): New member function.
9497 (profile_count::ipa): New member function.
9498 (profile_count::to_frequency): Declare.
9499 (profile_count::to_cgraph_frequency): Declare.
9500 * profile.c (OVERLAP_BASE): Delete.
9501 (compute_frequency_overlap): Delete.
9502 (compute_branch_probabilities): Do not use compute_frequency_overlap.
9503 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
9504 * sched-ebb.c (rank): Use counts only.
9505 * shrink-wrap.c (handle_simple_exit): Use counts only.
9506 (try_shrink_wrapping): Use counts only.
9507 (place_prologue_for_one_component): Use counts only.
9508 * tracer.c (find_best_predecessor): Use to_frequency.
9509 (find_trace): Use to_frequency.
9510 (tail_duplicate): Use to_frequency.
9511 * trans-mem.c (expand_transaction): Do not update frequency.
9512 * tree-call-cdce.c: Do not update frequency.
9513 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
9514 (gimple_merge_blocks): Likewise.
9515 (gimple_split_edge): Likewise.
9516 (gimple_duplicate_sese_region): Likewise.
9517 (gimple_duplicate_sese_tail): Likewise.
9518 (move_sese_region_to_fn): Likewise.
9519 (gimple_account_profile_record): Likewise.
9520 (insert_cond_bb): Likewise.
9521 * tree-complex.c (expand_complex_div_wide): Likewise.
9522 * tree-eh.c (lower_resx): Update profile.
9523 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
9525 (initialize_cfun): Do not initialize frequencies
9526 (freqs_to_counts): Delete.
9527 (copy_cfg_body): Ignore count parameter.
9528 (copy_body): Update.
9529 (expand_call_inline): Update count_max.
9530 (optimize_inline_calls): Update count_max.
9531 (tree_function_versioning): Update count_max.
9532 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
9533 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
9535 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
9536 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
9537 (try_peel_loop): Likewise.
9538 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
9540 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
9541 (tree_transform_and_unroll_loop): Do not use frequencies
9542 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
9543 Use reliable prediction only.
9544 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
9545 * tree-ssa-sink.c (select_best_block): Use to_frequency.
9546 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
9547 probability scaling.
9548 * tree-ssa-threadupdate.c (create_block_for_threading): Do
9549 not update frequency
9550 (any_remaining_duplicated_blocks): Likewise.
9551 (update_profile): Likewise.
9552 (estimated_freqs_path): Delete.
9553 (freqs_to_counts_path): Delete.
9554 (clear_counts_path): Delete.
9555 (ssa_fix_duplicate_block_edges): Likewise.
9556 (duplicate_thread_path): Likewise.
9557 * tree-switch-conversion.c (gen_inbound_check): Use counts.
9558 * tree-tailcall.c (decrease_profile): Do not update frequency.
9559 (eliminate_tail_call): Likewise.
9560 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
9561 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
9562 (optimize_mask_stores): Likewise.
9563 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9564 * ubsan.c (ubsan_expand_null_ifn): Update profile.
9565 (ubsan_expand_ptr_ifn): Update profile.
9566 * value-prof.c (gimple_ic): Simplify.
9567 * value-prof.h (gimple_ic): Update prototype.
9568 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
9569 * ipa-inline.c (compute_uninlined_call_time): Be sure that
9571 (want_inline_self_recursive_call_p): Likewise.
9572 (resolve_noninline_speculation): Only cummulate defined counts.
9573 (inline_small_functions): Use nonzero_p.
9574 (ipa_inline): Do not access freed node.
9576 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9578 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9579 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
9581 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
9583 * config/riscv/riscv.c (riscv_legitimize_move): Handle
9584 non-legitimate address.
9586 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9588 * config/rs6000/rs6000.md (*lt0_disi): Delete.
9589 (*lt0_<mode>di, *lt0_<mode>si): New.
9591 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9593 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
9594 TARGET_PAIRED_FLOAT.
9596 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
9597 Jim Wilson <jim.wilson@linaro.org>
9599 * config/aarch64/aarch64-cores.def (saphira): New CPU.
9600 * config/aarch64/aarch64-tune.md: Regenerated.
9601 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
9602 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
9604 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
9606 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
9607 (arc_expand_prologue): Restore blink for millicode.
9608 * config/arc/linux.h (LINK_EH_SPEC): Defined.
9610 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
9613 * config/i386/i386.c (ix86_vector_duplicate_value): Use
9614 gen_vec_duplicate after forcing the scalar into a register.
9616 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
9618 * combine (try_combine): Print the insns input to try_combine to the
9621 2017-11-02 Steve Ellcey <sellcey@cavium.com>
9624 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
9625 Remove second argument from aarch64_process_target_attr call.
9626 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
9628 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
9630 (aarch64_handle_attr_arch): Remove second argument.
9631 (aarch64_handle_attr_cpu): Ditto.
9632 (aarch64_handle_attr_tune): Ditto.
9633 (aarch64_handle_attr_isa_flags): Ditto.
9634 (aarch64_process_one_target_attr): Ditto.
9635 (aarch64_process_target_attr): Ditto.
9636 (aarch64_option_valid_attribute_p): Remove second argument.
9637 on aarch64_process_target_attr call.
9639 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9641 * diagnostic.c: Include "selftest-diagnostic.h".
9642 (selftest::assert_location_text): New function.
9643 (selftest::test_diagnostic_get_location_text): New function.
9644 (selftest::diagnostic_c_tests): Call it.
9646 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9648 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
9649 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
9650 (class selftest::test_diagnostic_context): Move to...
9651 * selftest-diagnostic.c: New file.
9652 * selftest-diagnostic.h: New file.
9654 2017-11-02 James Bowman <james.bowman@ftdichip.com>
9656 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
9657 offset range for FT32B.
9658 * config/ft32/ft32.h: option "mcompress" enables relaxation.
9659 * config/ft32/ft32.md: Add TARGET_NOPM.
9660 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
9661 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
9663 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
9665 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
9667 2017-11-02 Jeff Law <law@redhat.com>
9669 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
9670 virtual keyword on FINAL OVERRIDE members.
9672 * tree-ssa-propagate.h (ssa_propagation_engine): Group
9673 virtuals together. Add virtual destructor.
9674 (substitute_and_fold_engine): Similarly.
9676 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
9678 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
9680 2017-11-02 Richard Biener <rguenther@suse.de>
9682 PR tree-optimization/82795
9683 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
9685 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9687 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
9689 (gcc_GAS_CHECK_FEATURE): Remove.
9690 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
9691 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
9692 * configure: Regenerate.
9694 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
9696 * config/arc/arc.c (hwloop_optimize): Account for empty
9699 2017-11-02 Richard Biener <rguenther@suse.de>
9702 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
9703 Truncate ARRAY_REF index and element size.
9705 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9707 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
9709 2017-11-01 Jeff Law <law@redhat.com>
9711 * tree-ssa-ccp.c (ccp_folder): New class derived from
9712 substitute_and_fold_engine.
9713 (ccp_folder::get_value): New member function.
9714 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
9715 (ccp_fold_stmt): Remove prototype.
9716 (ccp_finalize): Call substitute_and_fold from the ccp_class.
9717 * tree-ssa-copy.c (copy_folder): New class derived from
9718 substitute_and_fold_engine.
9719 (copy_folder::get_value): Renamed from get_value.
9720 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
9721 * tree-vrp.c (vrp_folder): New class derived from
9722 substitute_and_fold_engine.
9723 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
9724 (vrp_folder::get_value): New member function.
9725 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
9726 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
9727 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
9728 provide a class interface to folder/substitute routines.
9729 (ssa_prop_fold_stmt_fn): Remove typedef.
9730 (ssa_prop_get_value_fn): Likewise.
9731 (subsitute_and_fold): Remove prototype.
9732 (replace_uses_in): Likewise.
9733 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
9734 Renamed from replace_uses_in. Call the virtual member function
9735 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
9736 (substitute_and_fold_dom_walker): Remove initialization of
9737 data member entries for calbacks. Add substitute_and_fold_engine
9738 member and initialize it.
9739 (substitute_and_fold_dom_walker::before_dom_children0: Use the
9740 member functions for get_value, replace_phi_args_in c
9741 replace_uses_in, and fold_stmt calls.
9742 (substitute_and_fold_engine::substitute_and_fold): Renamed from
9743 substitute_and_fold. Remove assert. Update ctor call.
9745 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
9746 (ssa_prop_visit_phi_fn): Likewise.
9747 (class ssa_propagation_engine): New class to provide an interface
9749 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
9751 (ssa_prop_visit_phi): Likewise.
9752 (ssa_propagation_engine::simulate_stmt): Moved into class.
9753 Call visit_phi/visit_stmt from the class rather than via
9754 file scoped static variables.
9755 (ssa_propagation_engine::simulate_block): Moved into class.
9756 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
9757 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
9758 set file scoped statics for the visit_stmt/visit_phi callbacks.
9759 * tree-complex.c (complex_propagate): New class derived from
9760 ssa_propagation_engine.
9761 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
9762 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
9763 (tree_lower_complex): Call ssa_propagate via the complex_propagate
9765 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
9766 ssa_propagation_engine.
9767 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
9768 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
9769 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
9770 * tree-ssa-copy.c (copy_prop): New class derived from
9771 ssa_propagation_engine.
9772 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
9773 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
9774 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
9775 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
9776 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
9777 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
9778 (execute_vrp): Call ssa_propagate from the vrp_prop class.
9780 2017-11-01 Jakub Jelinek <jakub@redhat.com>
9782 PR rtl-optimization/82778
9783 PR rtl-optimization/82597
9784 * compare-elim.c (struct comparison): Add in_a_setter field.
9785 (find_comparison_dom_walker::before_dom_children): Remove killed
9786 bitmap and df_simulate_find_defs call, instead walk the defs.
9787 Compute last_setter and initialize in_a_setter. Merge definitions
9788 with first initialization for a few variables.
9789 (try_validate_parallel): Use insn_invalid_p instead of
9790 recog_memoized. Return insn rather than just the pattern.
9791 (try_merge_compare): Fix up comment. Don't uselessly test if
9792 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
9794 (execute_compare_elim_after_reload): Remove df_chain_add_problem
9797 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9798 Alan Hayward <alan.hayward@arm.com>
9799 David Sherwood <david.sherwood@arm.com>
9801 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
9802 aarch64_hard_regno_nregs to get the number of registers
9805 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9806 Alan Hayward <alan.hayward@arm.com>
9807 David Sherwood <david.sherwood@arm.com>
9809 * config/aarch64/constraints.md (Upl): Rename to...
9811 * config/aarch64/aarch64.md
9812 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
9815 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9816 Alan Hayward <alan.hayward@arm.com>
9817 David Sherwood <david.sherwood@arm.com>
9819 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
9820 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
9823 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9824 Alan Hayward <alan.hayward@arm.com>
9825 David Sherwood <david.sherwood@arm.com>
9827 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
9828 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
9829 (aarch64_evpc_dup): Generate rtl direcly, rather than using
9831 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
9832 of a single element.
9833 * config/aarch64/iterators.md: Add a comment above the permute
9834 unspecs to say that they are generated directly by
9835 aarch64_expand_vec_perm_const.
9836 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
9838 2017-11-01 Nathan Sidwell <nathan@acm.org>
9840 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
9842 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9844 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
9845 and medany code models, and describe what they do.
9847 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9849 Revert accidental duplicate:
9851 * combine.c (can_change_dest_mode): Reject changes in
9852 REGMODE_NATURAL_SIZE.
9854 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9856 PR rtl-optimization/64682
9857 PR rtl-optimization/69567
9858 PR rtl-optimization/69737
9859 PR rtl-optimization/82683
9860 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9861 register mentioned in the note, drop the note, unless it came from I3,
9862 in which case it should go to I3 again.
9864 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9866 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9867 and return false if not.
9868 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9870 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9872 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9873 range is known to be empty.
9875 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9876 Alan Hayward <alan.hayward@arm.com>
9877 David Sherwood <david.sherwood@arm.com>
9879 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9880 and CONST_VECTOR_NUNITS instead of computing the number of units from
9881 the byte sizes of the vector and element.
9882 (simplify_binary_operation_1): Likewise.
9883 (simplify_const_binary_operation): Likewise.
9884 (simplify_ternary_operation): Likewise.
9886 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9887 Alan Hayward <alan.hayward@arm.com>
9888 David Sherwood <david.sherwood@arm.com>
9890 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9891 (int_mem_offset): ...this new function.
9892 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9893 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9894 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
9897 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9898 Alan Hayward <alan.hayward@arm.com>
9899 David Sherwood <david.sherwood@arm.com>
9901 * lower-subreg.c (interesting_mode_p): New function.
9902 (compute_costs, find_decomposable_subregs, decompose_register)
9903 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9904 (resolve_clobber, dump_choices): Use it.
9906 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9907 Alan Hayward <alan.hayward@arm.com>
9908 David Sherwood <david.sherwood@arm.com>
9910 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
9912 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9913 Alan Hayward <alan.hayward@arm.com>
9914 David Sherwood <david.sherwood@arm.com>
9916 * alias.c (find_base_value, find_base_term): Only process integer
9917 truncations. Check the precision rather than the size.
9919 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9920 Alan Hayward <alan.hayward@arm.com>
9921 David Sherwood <david.sherwood@arm.com>
9923 * machmode.h (is_narrower_int_mode): New function
9924 * optabs.c (expand_float, expand_fix): Use it.
9925 * dwarf2out.c (rotate_loc_descriptor): Likewise.
9927 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9928 Alan Hayward <alan.hayward@arm.com>
9929 David Sherwood <david.sherwood@arm.com>
9931 * rtl.h (narrower_subreg_mode): New function.
9932 * ira-color.c (update_costs_from_allocno): Use it.
9934 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9935 Alan Hayward <alan.hayward@arm.com>
9936 David Sherwood <david.sherwood@arm.com>
9938 * optabs-query.h (convert_optab_p): New function, split out from...
9939 (convert_optab_handler): ...here.
9940 (widening_optab_handler): Delete.
9941 (find_widening_optab_handler): Remove permit_non_widening parameter.
9942 (find_widening_optab_handler_and_mode): Likewise. Provide an
9943 override that operates on mode class wrappers.
9944 * optabs-query.c (widening_optab_handler): Delete.
9945 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9946 parameter. Assert that the two modes are the same class and that
9947 the "from" mode is narrower than the "to" mode. Use
9948 convert_optab_handler instead of widening_optab_handler.
9949 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9950 instead of widening_optab_handler.
9951 * expr.c (expand_expr_real_2): Update calls to
9952 find_widening_optab_handler.
9953 * optabs.c (expand_widen_pattern_expr): Likewise.
9954 (expand_binop_directly): Take the insn_code as a parameter.
9955 (expand_binop): Only call find_widening_optab_handler for
9956 conversion optabs; use optab_handler otherwise. Update calls
9957 to find_widening_optab_handler and expand_binop_directly.
9958 Use convert_optab_handler instead of widening_optab_handler.
9959 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9960 find_widening_optab_handler and use scalar_mode rather than
9962 (convert_plusminus_to_widen): Likewise.
9964 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9965 Alan Hayward <alan.hayward@arm.com>
9966 David Sherwood <david.sherwood@arm.com>
9968 * machmode.h (fixed_size_mode): New class.
9969 * rtl.h (get_pool_mode): Return fixed_size_mode.
9970 * gengtype.c (main): Add fixed_size_mode.
9971 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9972 (get_raw_arg_mode): Likewise.
9973 * doc/tm.texi: Regenerate.
9974 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9975 * targhooks.c (default_get_reg_raw_mode): Likewise.
9976 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9977 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9978 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9979 (msp430_get_raw_result_mode): Likewise.
9980 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9981 * dbxout.c (dbxout_parms): Require fixed-size modes.
9982 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9983 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9984 * omp-low.c (lower_oacc_reductions): Likewise.
9985 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9986 (simplify_subreg): Update accordingly.
9987 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9988 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9989 that aren't fixed-size.
9990 (get_pool_mode): Return a fixed_size_mode.
9991 (output_constant_pool_2): Take a fixed_size_mode.
9993 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9994 Alan Hayward <alan.hayward@arm.com>
9995 David Sherwood <david.sherwood@arm.com>
9997 * doc/rtl.texi (vec_series): Document.
9998 (const): Say that the operand can be a vec_series.
9999 * rtl.def (VEC_SERIES): New rtx code.
10000 * rtl.h (const_vec_series_p_1): Declare.
10001 (const_vec_series_p): New function.
10002 * emit-rtl.h (gen_const_vec_series): Declare.
10003 (gen_vec_series): Likewise.
10004 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
10005 (gen_vec_series): Likewise.
10006 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
10007 * simplify-rtx.c (simplify_unary_operation): Handle negations
10009 (simplify_binary_operation_series): New function.
10010 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
10011 (test_vector_ops_series): New function.
10012 (test_vector_ops): Call it.
10013 * config/powerpcspe/altivec.md (altivec_lvsl): Use
10014 gen_const_vec_series.
10015 (altivec_lvsr): Likewise.
10016 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
10018 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
10019 Alan Hayward <alan.hayward@arm.com>
10020 David Sherwood <david.sherwood@arm.com>
10022 * doc/rtl.texi (const): Update description of address constants.
10023 Say that vector constants are allowed too.
10024 * common.md (E, F): Use CONSTANT_P instead of checking for
10026 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
10027 checking for CONST_VECTOR.
10028 * expmed.c (make_tree): Use build_vector_from_val for a CONST
10030 * expr.c (expand_expr_real_2): Check for vector modes instead
10031 of checking for CONST_VECTOR.
10032 * rtl.h (const_vec_p): New function.
10033 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
10034 (unwrap_const_vec_duplicate): Handle them here too.
10036 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
10037 David Malcolm <dmalcolm@redhat.com>
10038 Alan Hayward <alan.hayward@arm.com>
10039 David Sherwood <david.sherwood@arm.com>
10041 * rtl.h (vec_duplicate_p): New function.
10042 * selftest-rtl.c (assert_rtx_eq_at): New function.
10043 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
10044 (assert_rtx_eq_at): Declare.
10045 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
10046 * selftest-run-tests.c (selftest::run_tests): Call it.
10047 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
10048 (simplify_unary_operation_1): Recursively handle vector duplicates.
10049 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
10051 (simplify_subreg): Handle subregs of vector duplicates.
10052 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
10053 (selftest::simplify_rtx_c_tests): New functions.
10055 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
10056 Alan Hayward <alan.hayward@arm.com>
10057 David Sherwood <david.sherwood@arm.com>
10059 * emit-rtl.h (gen_const_vec_duplicate): Declare.
10060 (gen_vec_duplicate): Likewise.
10061 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
10063 (gen_const_vector): ...here.
10064 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
10065 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
10066 whose elements are all equal.
10067 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
10068 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
10069 (simplify_relational_operation): Likewise.
10070 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
10072 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
10073 (aarch64_expand_vector_init): Likewise.
10074 * config/arm/arm.c (neon_vdup_constant): Likewise.
10075 (neon_expand_vector_init): Likewise.
10076 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
10077 (arm_block_set_unaligned_vect): Likewise.
10078 (arm_block_set_aligned_vect): Likewise.
10079 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
10080 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
10081 (expand_vec_perm_even_odd_pack): Likewise.
10082 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
10083 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
10084 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
10085 gen_const_vec_duplicate.
10086 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
10087 * config/mips/mips.c (mips_gen_const_int_vector): Use
10088 gen_const_vec_duplicate.
10089 (mips_expand_vector_init): Use CONST0_RTX.
10090 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
10091 (define_split): Use gen_const_vec_duplicate.
10092 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
10093 (define_split): Use gen_const_vec_duplicate.
10094 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
10095 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
10096 * config/spu/spu.c (spu_const): Likewise.
10098 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
10099 Alan Hayward <alan.hayward@arm.com>
10100 David Sherwood <david.sherwood@arm.com>
10102 * combine.c (can_change_dest_mode): Reject changes in
10103 REGMODE_NATURAL_SIZE.
10105 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
10107 * configure.ac (--enable-libssp): New.
10108 (gcc_cv_libc_provides_ssp): Check for explicit setting before
10109 trying to determine target-specific default. Adjust indentation.
10110 * configure: Regenerated.
10111 * doc/install.texi (Configuration): Expand --disable-libssp
10114 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
10116 config/i386/i386.c (ix86_expand_epilogue): Correct stack
10119 2017-10-31 Martin Jambor <mjambor@suse.cz>
10122 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
10124 2017-10-31 David Malcolm <dmalcolm@redhat.com>
10126 * auto-profile.c (autofdo_source_profile::read): Use
10127 UNKNOWN_LOCATION rather than 0.
10128 * diagnostic-core.h (warning_at_rich_loc): Rename to...
10129 (warning_at): ...this overload.
10130 (warning_at_rich_loc_n): Rename to...
10131 (warning_n): ...this overload.
10132 (error_at_rich_loc): Rename to...
10133 (error_at): ...this overload.
10134 (pedwarn_at_rich_loc): Rename to...
10135 (pedwarn): ...this overload.
10136 (permerror_at_rich_loc): Rename to...
10137 (permerror): ...this overload.
10138 (inform_at_rich_loc): Rename to...
10139 (inform): ...this overload.
10140 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
10141 (diagnostic_n_impl_richloc): Rename to...
10142 (diagnostic_n_impl): ...this rich_location *-based decl.
10143 (inform_at_rich_loc): Rename to...
10144 (inform): ...this, and add an assertion.
10145 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
10146 (warning_at_rich_loc): Rename to...
10147 (warning_at): ...this, and add an assertion.
10148 (warning_at_rich_loc_n): Rename to...
10149 (warning_n): ...this, and add an assertion.
10150 (warning_n): Update location_t-based implementation for removal of
10151 location_t-based diagnostic_n_impl.
10152 (pedwarn_at_rich_loc): Rename to...
10153 (pedwarn): ...this, and add an assertion.
10154 (permerror_at_rich_loc): Rename to...
10155 (permerror): ...this, and add an assertion.
10156 (error_n): Update for removal of location_t-based diagnostic_n_impl.
10157 (error_at_rich_loc): Rename to...
10158 (error_at): ...this, and add an assertion.
10159 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
10160 (driver::do_spec_on_infiles): Likewise.
10161 * substring-locations.c (format_warning_va): Update for renaming
10162 of inform_at_rich_loc.
10164 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
10166 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
10167 _Float<N>X built-in functions so that the variant without the
10168 "__builtin_" prefix is only enabled for the GNU C and Objective C
10169 languages when they are in non-strict ANSI/ISO mode.
10170 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
10171 * target.def (floatn_builtin_p): Add a target hook to control
10172 whether _Float<N> and _Float<N>X built-in functions without the
10173 "__builtin_" prefix are enabled, and return true for C and
10174 Objective C in the default hook. Include langhooks.h in
10176 * targhooks.h (default_floatn_builtin_p): Likewise.
10177 * targhooks.c (default_floatn_builtin_p): Likewise.
10178 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
10179 floatn_builtin_p target hook.
10180 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
10182 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
10183 Eric Botcazou <ebotcazou@adacore.com>
10185 PR rtl-optimization/81803
10186 * lra-constraints.c (curr_insn_transform): Also reload the whole
10187 register for a strict subreg no wider than a word if this is for
10188 a WORD_REGISTER_OPERATIONS target.
10190 2017-10-31 Jason Merrill <jason@redhat.com>
10192 * gdbinit.in: Skip over inlines from timevar.h.
10194 2017-10-31 Martin Liska <mliska@suse.cz>
10196 * doc/gcov.texi: Document new option.
10197 * gcov.c (print_usage): Likewise print it.
10198 (process_args): Support the argument.
10199 (format_count): New function.
10200 (format_gcov): Use the function.
10202 2017-10-31 Martin Liska <mliska@suse.cz>
10204 * gcov.c (struct name_map): do not use typedef.
10205 Define operator== and operator<.
10206 (name_search): Remove.
10207 (name_sort): Remove.
10208 (main): Do not allocate names.
10209 (process_file): Add vertical space.
10210 (generate_results): Use std::find.
10211 (release_structures): Do not release memory.
10212 (find_source): Use std::find.
10214 2017-10-31 Martin Liska <mliska@suse.cz>
10216 * gcov.c (struct line_info): Remove it's typedef.
10217 (line_info::line_info): Add proper ctor.
10218 (line_info::has_block): Do not use a typedef.
10219 (struct source_info): Do not use typedef.
10220 (circuit): Likewise.
10221 (get_cycles_count): Likewise.
10222 (output_intermediate_file): Iterate via vector iterator.
10223 (add_line_counts): Use std::vector methods.
10224 (accumulate_line_counts): Likewise.
10225 (output_lines): Likewise.
10227 2017-10-31 Martin Liska <mliska@suse.cz>
10229 * gcov.c (struct source_info): Remove typedef.
10230 (source_info::source_info): Add proper ctor.
10231 (accumulate_line_counts): Use struct, not it's typedef.
10232 (output_gcov_file): Likewise.
10233 (output_lines): Likewise.
10234 (main): Do not allocate an array.
10235 (output_intermediate_file): Use size of vector container.
10236 (process_file): Resize the vector.
10237 (generate_results): Do not preallocate, use newly added vector
10239 (release_structures): Do not release sources.
10240 (find_source): Use vector methods.
10241 (add_line_counts): Do not use typedef.
10243 2017-10-31 Martin Liska <mliska@suse.cz>
10245 * doc/gcov.texi: Document that.
10246 * gcov.c (add_line_counts): Mark lines with a non-executed
10248 (output_line_beginning): Handle such lines.
10249 (output_lines): Pass new argument.
10250 (output_intermediate_file): Print it in intermediate format.
10252 2017-10-31 Martin Liska <mliska@suse.cz>
10254 * color-macros.h: New file.
10255 * diagnostic-color.c: Factor out color related to macros to
10257 * doc/gcov.texi: Document -k option.
10258 * gcov.c (INCLUDE_STRING): Include string.h.
10259 (print_usage): Add -k option.
10260 (process_args): Parse it.
10261 (pad_count_string): New function.
10262 (output_line_beginning): Likewise.
10263 (DEFAULT_LINE_START): New macro.
10264 (output_lines): Support color output.
10266 2017-10-31 Martin Liska <mliska@suse.cz>
10268 PR gcov-profile/82633
10269 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
10270 their interaction with GCOV infrastructure.
10271 * configure.ac: Add -fkeep-{inline,static}-functions to
10273 * configure: Regenerate.
10275 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
10278 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
10280 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
10283 * config/rs6000/rs6000.md (allocate_stack): Force update interval
10284 into a register if it does not fit into an immediate offset field.
10286 2017-10-31 Olivier Hainque <hainque@adacore.com>
10288 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
10290 2017-10-31 Julia Koval <julia.koval@intel.com>
10292 * config.gcc: Add gfniintrin.h.
10293 * config/i386/gfniintrin.h: New.
10294 * config/i386/i386-builtin-types.def
10295 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
10296 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
10297 __builtin_ia32_vgf2p8affineinvqb_v32qi,
10298 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
10299 __builtin_ia32_vgf2p8affineinvqb_v16qi,
10300 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
10301 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
10302 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
10303 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
10304 V64QI_FTYPE_V64QI_V64QI_INT): New types.
10305 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
10306 * config/i386/immintrin.h: Include gfniintrin.h.
10307 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
10309 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
10311 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
10313 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10315 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
10316 (arm_ashldi3_1bit): Remove pattern.
10317 (ashrdi3): Remove shift by 1 expansion.
10318 (arm_ashrdi3_1bit): Remove pattern.
10319 (lshrdi3): Remove shift by 1 expansion.
10320 (arm_lshrdi3_1bit): Remove pattern.
10321 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
10322 cost of ashldi3 by 1.
10323 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
10324 (<shift>di3_neon): Likewise.
10326 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10328 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
10329 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
10330 and (*aarch64_simd_mov<VQ:mode>).
10331 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
10332 pattern alternative.
10333 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
10334 attributes to match pattern alternative.
10336 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
10338 * config.gcc (powerpc*-*-*): Add emmintrin.h.
10339 * config/rs6000/emmintrin.h: New file.
10340 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
10342 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10344 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
10345 * (movdi_vfp_cortexa8): Remove pattern.
10347 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10349 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
10351 (Specific, alpha*-dec-osf5.1): Remove.
10352 (Specific, mips-sgi-irix5): Remove.
10353 (Specific, mips-sgi-irix6): Remove.
10355 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10357 PR middle-end/22141
10358 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
10359 arguments to clear_bit_region_be.
10361 2017-10-30 Jim Wilson <wilson@tuliptree.org>
10363 * gimplify.c: Include memmodel.h.
10365 2017-10-30 Martin Jambor <mjambor@suse.cz>
10367 * omp-grid.c (grid_attempt_target_gridification): Also insert a
10368 condition whether loop should be executed at all.
10370 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
10372 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
10373 gimple folding of vec_madd() intrinsics.
10374 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
10375 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
10376 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
10378 2017-10-30 Richard Biener <rguenther@suse.de>
10380 PR tree-optimization/82762
10382 2017-10-23 Richard Biener <rguenther@suse.de>
10384 PR tree-optimization/82129
10386 2017-08-01 Richard Biener <rguenther@suse.de>
10388 PR tree-optimization/81181
10389 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10390 (compute_antic): ... end of iteration here.
10392 2017-10-30 Joseph Myers <joseph@codesourcery.com>
10394 * doc/invoke.texi (C Dialect Options): Document -std=c17,
10395 -std=iso9899:2017 and -std=gnu17.
10396 * doc/standards.texi (C Language): Document C17 support.
10397 * doc/cpp.texi (Overview): Mention -std=c17.
10398 (Standard Predefined Macros): Document C11 and C17 values of
10399 __STDC_VERSION__. Do not refer to C99 support as incomplete.
10400 * doc/extend.texi (Inline): Do not list individual options for
10401 standards newer than C99.
10402 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
10404 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
10407 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
10409 * asan.c (asan_finish_file): Align asan globals array by shadow
10412 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10414 PR middle-end/22141
10415 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
10416 (struct store_immediate_info): Add bitregion_start and bitregion_end
10418 (store_immediate_info::store_immediate_info): Add brs and bre
10419 arguments and initialize bitregion_{start,end} from those.
10420 (struct merged_store_group): Add bitregion_start, bitregion_end,
10421 align_base and mask fields. Drop unnecessary struct keyword from
10422 struct store_immediate_info. Add do_merge method.
10423 (clear_bit_region_be): Use memset instead of loop storing zeros.
10424 (merged_store_group::do_merge): New method.
10425 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
10426 stores as long as the surrounding bitregions have no gaps.
10427 (merged_store_group::merge_overlapping): Use do_merge.
10428 (merged_store_group::apply_stores): Test that bitregion_{start,end}
10429 is byte aligned, rather than requiring that start and width are
10430 byte aligned. Drop unnecessary struct keyword from
10431 struct store_immediate_info. Allocate and populate also mask array.
10432 Make start of the arrays relative to bitregion_start rather than
10433 start and size them according to bitregion_{end,start} difference.
10434 (struct imm_store_chain_info): Drop unnecessary struct keyword from
10435 struct store_immediate_info.
10436 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
10437 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
10438 struct keyword from struct store_immediate_info.
10439 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
10440 between stores as long as the surrounding bitregions have no gaps.
10442 (struct split_store): Add orig non-static data member.
10443 (split_store::split_store): Initialize orig to false.
10444 (find_constituent_stmts): Return store_immediate_info *, non-NULL
10445 if there is exactly a single original stmt. Change stmts argument
10446 to pointer from reference, if NULL, don't push anything to it. Add
10447 first argument, use it to optimize skipping over orig stmts that
10448 are known to be before bitpos already. Simplify.
10449 (split_group): Return unsigned int count how many stores are or
10450 would be needed rather than a bool. Add allow_unaligned argument.
10451 Change split_stores argument from reference to pointer, if NULL,
10452 only do a dry run computing how many stores would be produced.
10453 Rewritten algorithm to use both alignment and misalign if
10454 !allow_unaligned and handle bitfield stores with gaps.
10455 (imm_store_chain_info::output_merged_store): Set start_byte_pos
10456 from bitregion_start instead of start. Compute allow_unaligned
10457 here, if true, do 2 split_group dry runs to compute which one
10458 produces fewer stores and prefer aligned if equal. Punt if
10459 new count is bigger or equal than original before emitting any
10460 statements, rather than during that. Remove no longer needed
10461 new_ssa_names tracking. Replace num_stmts with
10462 split_stores.length (). Use 32-bit stack allocated entries
10463 in split_stores auto_vec. Try to reuse original store lhs/rhs1
10464 if possible. Handle bitfields with gaps.
10465 (pass_store_merging::execute): Ignore bitsize == 0 stores.
10466 Compute bitregion_{start,end} for the stores and construct
10467 store_immediate_info with that. Formatting fixes.
10469 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
10472 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
10473 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
10475 2017-10-29 Jim Wilson <wilson@tuliptree.org>
10477 * gimplify.c: Include tm_p.h.
10479 * common.opt (gcoff): Re-add as ignored option.
10480 (gcoff1, gcoff2, gcoff3): Likewise.
10482 * Makefile.in (OBJS): Delete sdbout.o.
10483 (GTFILES): Delete $(srcdir)/sdbout.c.
10484 * debug.h: Delete sdb_debug_hooks.
10485 * final.c: Delete sdbout.h include.
10486 (final_scan_insn): Delete SDB_DEBUG check.
10487 (rest_of_clean_state): Likewise.
10488 * output.h: Delete sdb_begin_function_line.
10489 * sdbout.c: Delete.
10490 * sdbout.h: Delete.
10491 * toplev.c: Delete sdbout.h include.
10492 (process_options): Delete SDB_DEBUG check.
10493 * tree-core.h (tree_type_common): Delete pointer field of
10495 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
10496 TYPE_SYMTAB_POINTER.
10497 * tree.h (TYPE_SYMTAB_POINTER): Delete.
10498 (TYPE_SYMTAB_IS_POINTER): Delete.
10499 (TYPE_SYMTAB_IS_DIE): Renumber.
10500 * xcoffout.c: Refer to former sdbout.c file.
10501 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
10503 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
10504 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
10505 (Debugging Options): Delete -gcoff.
10506 (-gstabs): Delete SDB reference.
10508 (-gcoff@var{level}): Delete.
10509 * doc/passes.texi (Debugging information output): Delete SDB and
10510 sdbout.c references.
10511 * doc/tm.texi: Regenerate.
10512 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
10513 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
10515 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
10516 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
10517 (SDB_DEBUGGING_INFO): Delete.
10518 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
10519 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
10520 * target.def (output_source_filename): Delete COFF reference.
10522 * common.opt (gcoff): Delete.
10523 (gxcoff+): Update Negative chain.
10524 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
10526 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
10527 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
10528 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
10529 and SDB references.
10530 (expand_function_start): Change sdb reference to past tense.
10531 (expand_function_end): Change sdb reference to past tense.
10532 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
10533 * opts.c (debug_type_names): Delete coff entry.
10534 (common_handle_option): Delete OPT_gcoff case.
10535 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
10537 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
10538 * config/cris/cris.h: Delete SDB reference in comment.
10539 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
10540 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
10541 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
10542 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
10544 (ix86_expand_prologue): Likewise.
10545 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
10546 * config/ia64/ia64.h: Likewise.
10547 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
10548 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
10550 * config/mmix/mmix.h: Likewise.
10551 * config/nds32/nds32.c: Likewise.
10552 * config/stormy/storym16.h: Likewise.
10553 * config/visium/visium.h: Likewise.
10554 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
10556 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
10558 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
10559 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
10560 FRAME_POINTER_REGNUM point at high end of local var area.
10562 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
10564 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
10565 Move comment around. Do not reset best_edge for a copiable
10566 destination if the copy would cause a partition change.
10567 (better_edge_p): Remove redundant check.
10569 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10571 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
10573 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10575 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
10576 for math functions that have _Float<N> and _Float<N>X variants.
10577 (mathfn_built_in_2): Add support for math functions that have
10578 _Float<N> and _Float<N>X variants.
10579 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
10580 (expand_builtin_mathfn_ternary): Add support for fma with
10581 _Float<N> and _Float<N>X variants.
10582 (expand_builtin): Likewise.
10583 (fold_builtin_3): Likewise.
10584 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
10585 create math function _Float<N> and _Float<N>X variants as external
10587 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
10588 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
10589 the __builtin_ prefix and if not strict ansi, without the prefix.
10590 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
10591 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
10592 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
10593 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
10594 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
10595 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
10596 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
10597 function signatures for fma _Float<N> and _Float<N>X variants.
10598 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
10599 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
10600 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
10601 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
10602 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
10603 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
10604 * gencfn-macros.c (print_case_cfn): Add support for math functions
10605 that have _Float<N> and _Float<N>X variants.
10606 (print_define_operator_list): Likewise.
10607 (fltfn_suffixes): Likewise.
10609 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
10610 for math functions that have _Float<N> and _Float<N>X variants.
10611 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
10612 and _Float<N>X variants.
10613 (COPYSIGN): Likewise.
10616 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
10617 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
10619 (integer_valued_read_call_p): Likewise.
10620 * fold-const-call.c (fold_const_call_ss): Likewise.
10621 (fold_const_call_sss): Add support for copysign, fmin, and fmax
10622 _Float<N> and _Float<N>X variants.
10623 (fold_const_call_ssss): Add support for fma _Float<N> and
10624 _Float<N>X variants.
10625 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
10626 support for copysign and fma _Float<N> and _Float<N>X variants.
10627 (backprop::process_builtin_call_use): Likewise.
10628 * tree-call-cdce.c (can_test_argument_range); Add support for
10629 sqrt _Float<N> and _Float<N>X variants.
10630 (edom_only_function): Likewise.
10631 (get_no_error_domain): Likewise.
10632 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
10633 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
10634 copysign _Float<N> and _Float<N>X variants.
10635 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
10636 handled by machine independent code.
10637 (FMAF128): Likewise.
10638 * doc/cpp.texi (Common Predefined Macros): Document defining
10639 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
10640 fma _Float<N> and _Float<N>X variants.
10642 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10645 * config/i386/i386-modes.def (CCFPU): Remove definition.
10646 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
10647 (ix86_cc_modes_compatible): Ditto.
10648 (ix86_expand_carry_flag_compare): Ditto.
10649 (ix86_expand_int_movcc): Ditto.
10650 (ix86_expand_int_addcc): Ditto.
10651 (ix86_reverse_condition): Ditto.
10652 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
10653 Return true/false for unordered/ordered fp comparisons.
10654 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
10655 (ix86_prepare_fp_compare_args): Update for rename.
10656 (ix86_expand_fp_compare): Update for rename. Generate unordered
10657 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
10658 (ix86_expand_sse_compare_and_jump): Ditto.
10659 * config/i386/predicates.md (fcmov_comparison_operator):
10660 Remove CCFPU mode handling.
10661 (ix86_comparison_operator): Ditto.
10662 (ix86_carry_flag_operator): Ditto.
10663 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
10664 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
10665 (*cmpu<mode>_cc_i387): Ditto.
10666 (FPCMP): Remove mode iterator.
10667 (unord): Remove mode attribute.
10668 (unord_subst): New define_subst transformation
10669 (unord): New define_subst attribute.
10670 (unordered): Ditto.
10671 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
10672 (*cmpi<unord>xf_i387): Ditto.
10673 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
10674 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
10675 using unord_subst transformation.
10676 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
10677 (round_saeonly): Also handle CCFP mode.
10678 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
10679 Remove UNSPEC_SAHF unspec handling.
10681 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10683 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
10685 2017-10-27 Jeff Law <law@redhat.com>
10687 * gimple-ssa-sprintf.c: Include domwalk.h.
10688 (class sprintf_dom_walker): New class, derived from dom_walker.
10689 (sprintf_dom_walker::before_dom_children): New function.
10690 (struct call_info): Moved into sprintf_dom_walker class
10691 (compute_formath_length, handle_gimple_call): Likewise.
10692 (sprintf_length::execute): Call the dominator walker rather
10693 than walking the statements.
10695 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
10696 gimple statement locations.
10697 (check_array_bounds): Corresponding changes. Get the statement's
10698 location directly from wi->stmt.
10700 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
10703 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
10705 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10707 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
10708 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
10710 2017-10-27 Jakub Jelinek <jakub@redhat.com>
10713 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
10714 * config/i386/i386.c (maybe_get_pool_constant): Removed.
10715 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
10716 maybe_get_pool_constant.
10717 * config/i386/predicates.md (zero_extended_scalar_load_operand):
10720 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10722 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
10723 2.26 caveat. Update gas and gld versions.
10724 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
10727 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10729 * cgraph.h (set_malloc_flag): Declare.
10730 * cgraph.c (set_malloc_flag_1): New function.
10731 (set_malloc_flag): Likewise.
10732 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
10733 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
10735 (read_ipa_call_summary): Add support for reading is_return_callee.
10736 (write_ipa_call_summary): Stream is_return_callee.
10737 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
10738 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
10739 ipa-prop.h, ipa-fnsummary.h.
10740 (pure_const_names): Change to static.
10741 (malloc_state_e): Define.
10742 (malloc_state_names): Define.
10743 (funct_state_d): Add field malloc_state.
10744 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
10745 (check_retval_uses): New function.
10746 (malloc_candidate_p): Likewise.
10747 (analyze_function): Add support for malloc attribute.
10748 (pure_const_write_summary): Stream malloc_state.
10749 (pure_const_read_summary): Add support for reading malloc_state.
10750 (dump_malloc_lattice): New function.
10751 (propagate_malloc): New function.
10752 (warn_function_malloc): New function.
10753 (ipa_pure_const::execute): Call propagate_malloc and
10754 ipa_free_fn_summary.
10755 (pass_local_pure_const::execute): Add support for malloc attribute.
10756 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
10757 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
10759 2017-10-27 Martin Liska <mliska@suse.cz>
10761 PR gcov-profile/82457
10762 * doc/invoke.texi: Document that one needs a non-strict ISO mode
10763 for fork-like functions to be properly instrumented.
10765 2017-10-27 Richard Biener <rguenther@suse.de>
10767 PR middle-end/81659
10768 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
10769 info when we redirected EH.
10771 2017-10-26 Michael Collison <michael.collison@arm.com>
10773 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
10775 (<optab>_trunchf<GPI:mode>2: New pattern.
10776 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
10777 * config/aarch64/iterators.md (wv): New mode attribute.
10778 (vf, VF): New mode attributes.
10779 (vgp, VGP): New mode attributes.
10780 (s): Update attribute with SImode and DImode prefixes.
10782 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10784 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
10785 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
10786 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
10787 (nios2_option_overide): Initialize it. Don't allow R0-relative
10788 addressing with PIC.
10789 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
10790 (nios2_symbolic_constant_p): Likewise.
10791 (nios2_legitimate_address_p): Likewise.
10792 (nios2_r0rel_section_name_p): New.
10793 (nios2_symbol_ref_in_r0rel_data_p): New.
10794 (nios2_emit_move_sequence): Handle r0rel_constant_p.
10795 (r0rel_constant_p): New.
10796 (nios2_print_operand_address): Handle r0rel_constant_p.
10797 (nios2_cdx_narrow_form_p): Likewise.
10798 * config/nios2/nios2.opt (mr0rel-sec=): New option.
10799 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
10800 (Nios II Options): Document -mr0rel-sec.
10802 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10804 * config/nios2/nios2.c: Include xregex.h.
10805 (nios2_gprel_sec_regex): New.
10806 (nios2_option_overide): Initialize it. Don't allow GP-relative
10807 addressing with PIC.
10808 (nios2_small_section_name_p): Check for regex match.
10809 * config/nios2/nios2.opt (mgprel-sec=): New option.
10810 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
10811 (Nios II Options): Document -mgprel-sec.
10813 2017-10-26 Jim Wilson <wilson@tuliptree.org>
10815 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
10817 2017-10-26 Tom de Vries <tom@codesourcery.com>
10819 PR tree-optimization/82707
10820 * gimple.c (gimple_copy): Fix unsharing of
10821 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
10823 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
10825 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
10826 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
10827 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
10828 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
10829 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
10830 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
10831 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
10832 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
10833 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
10834 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
10835 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
10836 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
10837 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
10838 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
10839 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
10840 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
10841 _mm512_mask_cmpunord_ps_mask): New intrinsics.
10843 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10845 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
10847 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
10848 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10849 warning to rs6000.c. Remove the Undocumented flag, since it has
10851 (-mabi=ibmlongdouble): Likewise.
10852 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10853 already set, set the default format for long double.
10854 (rs6000_debug_reg_global): Print whether long double is IBM or
10856 (rs6000_option_override_internal): Rework setting long double
10857 format. Only warn if the user is changing the long double default
10858 and they did not use -Wno-psabi.
10859 * doc/invoke.texi (PowerPC options): Update the documentation for
10860 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10862 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10863 Alan Hayward <alan.hayward@arm.com>
10864 David Sherwood <david.sherwood@arm.com>
10866 * rtl.h (wider_subreg_mode): New function.
10867 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10868 rather than an unsigned int *.
10869 * ira-color.c (regno_max_ref_width): Replace with...
10870 (regno_max_ref_mode): ...this new variable.
10871 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10872 Use wider_subreg_mode.
10873 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10874 rather than an unsigned int *.
10875 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10876 (process_alt_operands): Likewise.
10877 (invariant_p): Likewise.
10878 * lra-spills.c (assign_mem_slot): Likewise.
10879 (add_pseudo_to_slot): Likewise.
10880 * lra.c (collect_non_operand_hard_regs): Likewise.
10881 (add_regs_to_insn_regno_info): Likewise.
10882 * reload1.c (regno_max_ref_width): Replace with...
10883 (regno_max_ref_mode): ...this new variable.
10884 (reload): Update accordingly. Update call to
10885 ira_sort_regnos_for_alter_reg.
10886 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10887 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10888 (scan_paradoxical_subregs): Likewise.
10890 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10892 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10893 (aarch64_frame): Add emit_frame_chain boolean.
10894 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10895 Move eh_return case to aarch64_layout_frame.
10896 (aarch64_layout_frame): Initialize emit_frame_chain.
10897 (aarch64_expand_prologue): Use emit_frame_chain.
10899 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10901 * config/aarch64/aarch64.c (aarch64_layout_frame):
10902 Ensure LR is always stored at the bottom of the callee-saves.
10903 Remove rarely used frame layout which saves callee-saves at top of
10904 frame, so the store of LR can be used as a valid probe in all cases.
10906 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10908 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10909 Improve unaligned TImode/TFmode base/offset split.
10911 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10912 Alan Hayward <alan.hayward@arm.com>
10913 David Sherwood <david.sherwood@arm.com>
10915 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10916 * combine.c (find_single_use_1): Likewise.
10917 (expand_field_assignment): Likewise.
10918 (move_deaths): Likewise.
10919 * lra-constraints.c (simplify_operand_subreg): Likewise.
10920 (curr_insn_transform): Likewise.
10921 * lra.c (collect_non_operand_hard_regs): Likewise.
10922 (add_regs_to_insn_regno_info): Likewise.
10923 * rtlanal.c (reg_referenced_p): Likewise.
10924 (covers_regno_no_parallel_p): Likewise.
10926 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10928 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10929 Don't print any bits outside the precision of the value.
10930 * wide-int.cc (test_printing): Add some new tests.
10932 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10934 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10935 supports -xbrace_comment option.
10936 * configure: Regenerate.
10937 * config.in: Regenerate.
10938 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10939 (ASM_CPU_SPEC): Use it.
10941 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10943 * target.def (static_rtx_alignment): New hook.
10944 * targhooks.h (default_static_rtx_alignment): Declare.
10945 * targhooks.c (default_static_rtx_alignment): New function.
10946 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10947 * doc/tm.texi: Regenerate.
10948 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10949 instead of targetm.constant_alignment. Remove call to
10950 set_mem_attributes.
10951 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10952 (cris_preferred_mininum_alignment): New function, split out from...
10953 (cris_constant_alignment): ...here.
10954 (cris_static_rtx_alignment): New function.
10955 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10957 (ix86_constant_alignment): ...here.
10958 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10959 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10960 (mmix_static_rtx_alignment): New function.
10961 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10962 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10964 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10967 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10968 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10970 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10972 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10973 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10974 mark as a sign-extending load.
10975 (local_pic_load_u): Define.
10977 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10979 PR middle-end/82062
10980 * fold-const.c (operand_equal_for_comparison_p): Also return true
10981 if ARG0 is a simple variant of ARG1 with narrower precision.
10982 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10984 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10986 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10988 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10989 scatter_static, scatter_per_elt.
10990 * x86-tune-costs.h: Add new cost entries.
10992 2017-10-25 Richard Biener <rguenther@suse.de>
10994 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10995 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10996 class pass_fre): Move to ...
10997 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10998 class pass_fre): ... here and adjust for statistics.
11000 2017-10-25 Jakub Jelinek <jakub@redhat.com>
11003 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
11004 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
11005 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
11006 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
11009 2017-10-25 Richard Biener <rguenther@suse.de>
11011 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
11012 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
11013 eliminate_push_avail, eliminate_insert): Move inside...
11014 (class eliminate_dom_walker): ... this class in preparation
11016 (fini_eliminate): Remove by merging with ...
11017 (eliminate): ... this function. Adjust for class changes.
11018 (pass_pre::execute): Remove fini_eliminate call.
11019 (pass_fre::execute): Likewise.
11021 2017-10-24 Jakub Jelinek <jakub@redhat.com>
11024 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
11025 (VPERMI2, VPERMI2I): New mode iterators.
11026 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
11027 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
11029 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
11030 mode iterator. Remove 3 old define_insn patterns.
11031 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
11032 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
11033 VPERMI2 mode iterator, remove the other two expanders.
11034 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
11035 to use VPERMI2 mode iterator, add another alternative for vpermi2*
11036 instructions, remove the other two patterns.
11037 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
11038 mode iterator, remove the other two patterns.
11039 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
11040 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
11041 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
11042 and adjust argument order accordingly.
11043 (ix86_expand_vec_perm): Adjust caller.
11044 (expand_vec_perm_1): Likewise.
11045 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
11046 (expand_vec_perm_vpermt2_vpshub2): ... this.
11047 (ix86_expand_vec_perm_const_1): Adjust caller.
11048 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
11051 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
11052 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
11053 (vec_shl_<mode>): Remove unused expander.
11054 (avx512bw_<shift_insn><mode>3): New define_insn.
11055 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
11056 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
11058 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
11061 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
11064 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
11066 PR rtl-optimization/82396
11067 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
11068 (autopref_multipass_init): Simplify initialization.
11069 (autopref_rank_data): Simplify sort order.
11070 * gcc/sched-int.h (autopref_multipass_data_): Remove
11071 multi_mem_insn_p, min_offset and max_offset.
11073 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
11075 PR middle-end/60580
11076 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
11077 Check special value of flag_omit_frame_pointer.
11078 (aarch64_can_eliminate): Likewise.
11079 (aarch64_override_options_after_change_1): Simplify handling of
11080 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
11082 2017-10-24 Richard Biener <rguenther@suse.de>
11084 PR tree-optimization/82697
11085 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
11086 zero for conditional load and unconditional store.
11088 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
11090 * doc/install.texi: Document bootstrap-cet.
11092 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
11095 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
11096 ENDBR instruction at function entrance if function is only
11099 2017-10-24 Jakub Jelinek <jakub@redhat.com>
11102 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
11103 patterns to better describe from which operation the CF is computed.
11104 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
11105 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
11106 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
11107 is 0, use _0 suffixed expanders instead of emitting a comparison
11110 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
11112 * config/i386/i386.md(*movsf_internal, *movdf_internal):
11113 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
11115 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
11117 PR middle-end/82569
11118 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
11119 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
11120 * loop-iv.c (iv_get_reaching_def): Likewise.
11121 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
11122 variable is promoted and the partition contains undefined values.
11124 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11126 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
11128 (nios2_address_cost): Define.
11129 (nios2_legitimize_address): Recognize (exp + constant) directly.
11130 (TARGET_ADDRESS_COST): Define.
11132 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11134 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
11135 (nios2_symbolic_memory_operand_p): Declare.
11136 (nios2_split_large_constant): Declare.
11137 (nios2_split_symbolic_memory_operand): Declare.
11138 * config/nios2/nios2.c: Adjust includes.
11139 (nios2_symbolic_constant_allowed): New.
11140 (nios2_symbolic_constant_p): New.
11141 (nios2_plus_symbolic_constant_p): New.
11142 (nios2_valid_addr_expr_p): Recognize addresses involving
11143 symbolic constants.
11144 (nios2_legitimate_address_p): Likewise, also LO_SUM.
11145 (nios2_symbolic_memory_operand_p): New.
11146 (nios2_large_constant_p): New.
11147 (nios2_split_large_constant): New.
11148 (nios2_split_plus_large_constant): New.
11149 (nios2_split_symbolic_memory_operand): New.
11150 (nios2_legitimize_address): Code refactoring. Handle addresses
11151 involving symbolic constants.
11152 (nios2_emit_move_sequence): Likewise.
11153 (nios2_print_operand): Improve error output.
11154 (nios2_print_operand_address): Handle LO_SUM.
11155 (nios2_cdx_narrow_form_p): Likewise.
11156 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
11157 operands involving symbolic constants.
11158 (movhi_internal, movsi_internal): Likewise.
11159 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
11160 (extendhisi2, extendqi<mode>2): Likewise.
11162 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11164 * tree-pass.h (PROP_rtl_split_insns): Define.
11165 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
11167 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11169 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
11171 2017-10-23 Jakub Jelinek <jakub@redhat.com>
11174 * target.def (const_not_ok_for_debug_p): Default to
11175 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
11176 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
11177 * targhooks.c (default_const_not_ok_for_debug_p): New function.
11178 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
11179 which targetm.const_not_ok_for_debug_p returned true.
11180 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
11182 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
11184 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
11185 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
11186 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
11188 (ix86_const_not_ok_for_debug_p): New function.
11189 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
11190 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
11192 2017-10-23 David Malcolm <dmalcolm@redhat.com>
11195 * system.h: Conditionally include "unique-ptr.h" if
11196 INCLUDE_UNIQUE_PTR is defined.
11197 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
11198 of defining INCLUDE_UNIQUE_PTR before including "system.h".
11200 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
11202 * config/rl78/rl78.md: New define_expand "subdi3".
11204 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
11207 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
11208 DF_REF_INSN if DF_REF_INSN_INFO is false.
11210 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
11212 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
11213 xmm_move instead of sse_move.
11214 (sse_store_index): New function.
11215 (ix86_register_move_cost): Be more sensible about mismatch stall;
11216 model AVX moves correctly; make difference between sse->integer and
11218 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
11219 moves; make difference between SSE and AVX.
11220 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
11221 and zmm_move. Increase size of sse load and store tables;
11222 add unaligned load and store tables; add ssemmx_to_integer.
11223 * x86-tune-costs.h: Update all entries according to real
11224 move latencies from Agner Fog's manual and chip documentation.
11226 2017-10-23 Jakub Jelinek <jakub@redhat.com>
11229 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
11230 * config/i386/constraints.md (Wf): New constraint.
11231 * config/i386/i386.md (UNSPEC_SBB): New unspec.
11232 (cmp<dwi>_doubleword): Removed.
11233 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
11234 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
11235 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
11236 expand with cmp<dwi>_doubleword. For LTU and GEU use
11237 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
11239 * common.opt (gcolumn-info): Enable by default.
11240 * doc/invoke.texi (gcolumn-info): Document new default.
11242 2017-10-23 Richard Biener <rguenther@suse.de>
11244 PR tree-optimization/82672
11245 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
11246 Fold the stmt if we propagated into it.
11248 2017-10-23 Richard Biener <rguenther@suse.de>
11250 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
11251 (bitmap_remove_expr_from_set): ... this. All callers call this
11252 for non-constant values.
11253 (bitmap_set_subtract): Rename to...
11254 (bitmap_set_subtract_expressions): ... this. Adjust and
11256 (bitmap_set_contains_value): Remove superfluous check.
11257 (bitmap_set_replace_value): Inline into single caller ...
11258 (bitmap_value_replace_in_set): ... here and simplify.
11259 (dependent_clean): Merge into ...
11260 (clean): ... this using an overload. Adjust.
11261 (prune_clobbered_mems): Adjust.
11262 (compute_antic_aux): Likewise.
11263 (compute_partial_antic_aux): Likewise.
11265 2017-10-23 Richard Biener <rguenther@suse.de>
11267 PR tree-optimization/82129
11269 2017-08-01 Richard Biener <rguenther@suse.de>
11271 PR tree-optimization/81181
11272 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
11273 (compute_antic): ... end of iteration here.
11275 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11277 * target.def (starting_frame_offset): New hook.
11278 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
11279 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
11280 * doc/tm.texi.in: Regenerate.
11281 * hooks.h (hook_hwi_void_0): Declare.
11282 * hooks.c (hook_hwi_void_0): New function.
11283 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
11284 STARTING_FRAME_OFFSET.
11285 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
11286 * reload1.c (reload): Likewise.
11287 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
11288 instead of STARTING_FRAME_OFFSET.
11289 * function.c (try_fit_stack_local): Likewise.
11290 (assign_stack_local_1): Likewise
11291 (instantiate_virtual_regs): Likewise.
11292 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
11293 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
11294 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
11295 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
11296 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
11297 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
11298 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
11299 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
11300 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
11301 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
11302 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
11303 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
11304 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
11305 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
11306 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
11307 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
11308 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
11309 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
11310 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
11311 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
11312 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
11313 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
11314 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
11315 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
11316 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
11317 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
11318 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
11319 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
11320 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
11321 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
11322 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
11323 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
11324 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
11325 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
11326 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
11327 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
11328 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
11329 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
11330 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
11331 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
11332 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
11333 * config/avr/avr.c (avr_starting_frame_offset): Make static and
11334 return a HOST_WIDE_INT.
11335 (avr_builtin_setjmp_frame_value): Use it instead of
11336 STARTING_FRAME_OFFSET.
11337 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11338 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
11339 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
11341 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11342 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
11343 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
11344 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11345 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
11346 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
11347 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11348 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
11349 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
11350 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11351 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
11352 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
11354 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11355 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
11356 * config/mips/mips.c (mips_compute_frame_info): Refer to
11357 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11358 (mips_starting_frame_offset): New function.
11359 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11360 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
11361 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
11362 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
11363 and return a HOST_WIDE_INT.
11364 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11365 (mmix_initial_elimination_offset): Refer to
11366 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11367 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
11368 * config/pa/pa.c (pa_starting_frame_offset): New function.
11369 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
11370 (pa_expand_prologue): Likewise.
11371 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11372 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
11373 !FRAME_GROWS_DOWNWARD handling to...
11374 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11375 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
11376 !FRAME_GROWS_DOWNWARD handling to...
11377 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11378 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
11379 !FRAME_GROWS_DOWNWARD handling to...
11380 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11381 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
11383 (rs6000_starting_frame_offset): New function.
11384 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
11385 !FRAME_GROWS_DOWNWARD handling to...
11386 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11387 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
11388 !FRAME_GROWS_DOWNWARD handling to...
11389 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11390 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
11391 !FRAME_GROWS_DOWNWARD handling to...
11392 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11393 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
11394 (rs6000_starting_frame_offset): New function.
11395 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
11396 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
11397 * config/vax/vax.c (vax_starting_frame_offset): New function.
11398 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
11399 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11400 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
11401 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
11402 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11403 * system.h (STARTING_FRAME_OFFSET): Poison.
11405 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11407 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11408 SCALAR_TYPE_MODE instead of TYPE_MODE.
11410 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11411 Alan Hayward <alan.hayward@arm.com>
11412 David Sherwood <david.sherwood@arm.com>
11414 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
11416 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11417 Alan Hayward <alan.hayward@arm.com>
11418 David Sherwood <david.sherwood@arm.com>
11420 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
11422 2017-10-23 Richard Biener <rguenther@suse.de>
11424 PR tree-optimization/82129
11425 * tree-ssa-pre.c (bitmap_set_and): Remove.
11426 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
11427 canonicalizing expressions in the set to those with lowest
11428 ID rather than taking that from the first edge.
11430 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11432 * combine.c (rtx_equal_for_field_assignment_p): Use
11433 byte_lowpart_offset.
11435 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11436 Alan Hayward <alan.hayward@arm.com>
11437 David Sherwood <david.sherwood@arm.com>
11439 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
11440 to a SUBREG_PROMOTED_VAR.
11442 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11443 Alan Hayward <alan.hayward@arm.com>
11444 David Sherwood <david.sherwood@arm.com>
11446 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
11447 (expand_debug_source_expr): Likewise.
11448 * combine.c (combine_simplify_rtx): Likewise.
11449 * cse.c (fold_rtx): Likewise.
11450 * optabs.c (expand_float): Likewise.
11451 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11452 (simplify_binary_operation_1): Likewise.
11454 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11455 Alan Hayward <alan.hayward@arm.com>
11456 David Sherwood <david.sherwood@arm.com>
11458 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
11459 (record_promoted_value): Likewise.
11460 * expr.c (expand_expr_real_2): Likewise.
11461 * ree.c (update_reg_equal_equiv_notes): Likewise.
11462 (combine_set_extension): Likewise.
11463 * rtlanal.c (low_bitmask_len): Likewise.
11464 * simplify-rtx.c (neg_const_int): Likewise.
11465 (simplify_binary_operation_1): Likewise.
11467 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11468 Alan Hayward <alan.hayward@arm.com>
11469 David Sherwood <david.sherwood@arm.com>
11471 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
11472 * regcprop.c (maybe_mode_change): Likewise.
11473 * reload1.c (alter_reg): Likewise.
11475 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11477 * inchash.h (inchash::hash::add_wide_int): New function.
11478 * lto-streamer-out.c (hash_tree): Use it.
11480 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11482 * inchash.h (inchash::hash::add_wide_int): Rename to...
11483 (inchash::hash::add_hwi): ...this.
11484 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
11485 (polymorphic_call_target_hasher::hash): Likewise.
11486 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
11487 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
11488 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
11489 * lto-streamer-out.c (hash_tree): Likewise.
11490 * optc-save-gen.awk: Likewise.
11491 * tree.c (add_expr): Likewise.
11493 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11496 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
11497 for ordered inequality comparisons even with TARGET_IEEE_FP.
11499 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11502 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
11503 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11504 Expand with cmp<dwi>_doubleword.
11506 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11508 * extend.texi: Add x86 specific to 'nocf_check' attribute.
11509 List CET intrinsics.
11510 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
11511 specific to -fcf-protection option.
11513 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11515 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
11516 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
11517 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
11518 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
11519 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
11520 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
11521 (extra_objs): Add cet.o for Linux/x86 targets.
11522 (tmake_file): Add i386/t-cet for Linux/x86 targets.
11523 * config/i386/cet.c: New file.
11524 * config/i386/cetintrin.h: Likewise.
11525 * config/i386/t-cet: Likewise.
11526 * config/i386/cpuid.h (bit_SHSTK): New.
11527 (bit_IBT): Likewise.
11528 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
11529 pass IBT and SHSTK bits.
11530 * config/i386/i386-builtin-types.def
11531 (VOID_FTYPE_UNSIGNED_PVOID): New.
11532 (VOID_FTYPE_UINT64_PVOID): Likewise.
11533 * config/i386/i386-builtin.def: Add CET intrinsics.
11534 * config/i386/i386-c.c (ix86_target_macros_internal): Add
11535 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
11536 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
11537 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
11539 * config/i386/i386.c (rest_of_insert_endbranch): New.
11540 (pass_data_insert_endbranch): Likewise.
11541 (pass_insert_endbranch): Likewise.
11542 (make_pass_insert_endbranch): Likewise.
11543 (ix86_notrack_prefixed_insn_p): Likewise.
11544 (ix86_target_string): Add -mibt, -mshstk flags.
11545 (ix86_option_override_internal): Add flag_cf_protection
11547 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
11548 (ix86_print_operand): Add 'notrack' prefix output.
11549 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
11550 (ix86_expand_builtin): Expand CET intrinsics.
11551 (x86_output_mi_thunk): Add 'endbranch' instruction.
11552 * config/i386/i386.h (TARGET_IBT): New.
11553 (TARGET_IBT_P): Likewise.
11554 (TARGET_SHSTK): Likewise.
11555 (TARGET_SHSTK_P): Likewise.
11556 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
11557 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
11558 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
11559 (builtin_setjmp_setup): New pattern.
11560 (builtin_longjmp): Likewise.
11561 (rdssp<mode>): Likewise.
11562 (incssp<mode>): Likewise.
11563 (saveprevssp): Likewise.
11564 (rstorssp): Likewise.
11565 (wrss<mode>): Likewise.
11566 (wruss<mode>): Likewise.
11567 (setssbsy): Likewise.
11568 (clrssbsy): Likewise.
11569 (nop_endbr): Likewise.
11570 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
11572 * config/i386/immintrin.h: Include <cetintrin.h>.
11573 * config/i386/linux-common.h
11574 (file_end_indicate_exec_stack_and_cet): New prototype.
11575 (TARGET_ASM_FILE_END): New.
11577 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11579 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
11580 latencies instead of having separate table; make difference between
11581 integer and float costs.
11582 * i386.h (processor_costs): Remove scalar_stmt_cost,
11583 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
11584 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
11586 * x86-tune-costs.h: Remove entries which has been removed in
11587 procesor_costs from all tables; make cond_taken_branch_cost
11588 and cond_not_taken_branch_cost COST_N_INSNS based.
11590 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11592 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
11594 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11596 * config/i386/i386.md (isa): Remove fma_avx512f.
11597 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
11598 <avx512>_fmadd_<mode>_mask3<round_name>,
11599 <avx512>_fmsub_<mode>_mask<round_name>,
11600 <avx512>_fmsub_<mode>_mask3<round_name>,
11601 <avx512>_fnmadd_<mode>_mask<round_name>,
11602 <avx512>_fnmadd_<mode>_mask3<round_name>,
11603 <avx512>_fnmsub_<mode>_mask<round_name>,
11604 <avx512>_fnmsub_<mode>_mask3<round_name>,
11605 <avx512>_fmaddsub_<mode>_mask<round_name>,
11606 <avx512>_fmaddsub_<mode>_mask3<round_name>,
11607 <avx512>_fmsubadd_<mode>_mask<round_name>,
11608 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
11609 (*vec_widen_umult_even_v16si<mask_name>,
11610 *vec_widen_smult_even_v16si<mask_name>): Likewise.
11611 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
11613 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11615 * extend.texi: Add 'nocf_check' documentation.
11616 * gimple.texi: Add second parameter to
11617 gimple_build_call_from_tree.
11618 * invoke.texi: Add -fcf-protection documentation.
11619 * rtl.texi: Add REG_CALL_NOTRACK documenation.
11621 2017-10-20 Richard Biener <rguenther@suse.de>
11623 PR tree-optimization/82473
11624 * tree-vect-loop.c (vectorizable_reduction): Properly get at
11625 the largest input type.
11627 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11629 * c-attribs.c (handle_nocf_check_attribute): New function.
11630 (c_common_attribute_table): Add 'nocf_check' handling.
11631 * gimple-parser.c: Add second argument NULL to
11632 gimple_build_call_from_tree.
11633 * attrib.c (comp_type_attributes): Check nocf_check attribute.
11634 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
11636 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
11637 * common.opt: Add fcf-protection flag.
11638 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
11639 * flag-types.h: Add enum cf_protection_level.
11640 * gimple.c (gimple_build_call_from_tree): Add second parameter.
11641 Add 'nocf_check' attribute propagation to gimple call.
11642 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
11643 (gimple_build_call_from_tree): Update prototype.
11644 (gimple_call_nocf_check_p): New function.
11645 (gimple_call_set_nocf_check): Likewise.
11646 * gimplify.c: Add second argument to gimple_build_call_from_tree.
11647 * ipa-icf.c: Add nocf_check attribute in statement hash.
11648 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
11649 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
11650 * toplev.c (process_options): Add flag_cf_protection handling.
11652 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11654 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
11656 2017-10-20 Richard Biener <rguenther@suse.de>
11658 PR tree-optimization/82603
11659 * tree-if-conv.c (predicate_mem_writes): Make sure to only
11660 remove false predicated stores.
11662 2017-10-20 Richard Biener <rguenther@suse.de>
11664 * graphite-isl-ast-to-gimple.c
11665 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
11666 Remove return value and simplify, dump copied stmt after lhs
11668 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
11669 Reduce dump verbosity.
11670 (gsi_insert_earliest): Likewise.
11671 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
11672 * graphite.c (print_global_statistics): Adjust dumping.
11673 (print_graphite_scop_statistics): Likewise.
11674 (print_graphite_statistics): Do not dump loops here.
11675 (graphite_transform_loops): But here.
11677 2017-10-20 Nicolas Roche <roche@adacore.com>
11679 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
11680 * configure: Regenerate.
11682 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11685 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
11686 functions when optimizing replace GIMPLE_RETURN stmts with
11687 calls to __builtin_unreachable ().
11690 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
11691 for -fsanitize=thread link of executables.
11692 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
11693 link of executables.
11696 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
11697 New mode iterators.
11698 (<shift_insn><mode>3<mask_name>): Change the last of the 3
11699 define_insns for logical vector shifts to use VI248_AVX512BW
11700 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
11701 condition, useless isa and prefix attributes. Change the first
11702 2 of these define_insns to ...
11703 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
11704 define_insn for avx512vl.
11705 (<shift_insn><mode>3): ... and this, new define_insn without
11706 masking for non-avx512vl.
11709 * config/i386/sse.md (*andnot<mode>3,
11710 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
11711 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
11712 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
11713 not applied use empty suffix even for TARGET_AVX512VL.
11714 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
11715 is applied, supply evex,evex or evex,evex,evex instead of just
11718 2017-10-20 Julia Koval <julia.koval@intel.com>
11720 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
11721 (OPTION_MASK_ISA_GFNI_UNSET): New.
11722 (ix86_handle_option): Handle OPT_mgfni.
11723 * config/i386/cpuid.h (bit_GFNI): New.
11724 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
11725 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
11726 * config/i386/i386.c (ix86_target_string): Add -mgfni.
11727 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
11728 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
11729 * config/i386/i386.opt: Add mgfni.
11731 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
11733 * config/msp430/msp430.c (msp430_option_override): Disable
11734 -fdelete-null-pointer-checks.
11735 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
11737 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11739 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
11740 of x87 and SSE instructions.
11742 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11744 * asan.c (create_cond_insert_point): Do not update edge count.
11745 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
11746 (afdo_propagate_circuit): Likewise.
11747 (afdo_calculate_branch_prob): Likewise.
11748 (afdo_annotate_cfg): Likewise.
11749 * basic-block.h (struct edge_def): Remove count.
11750 (edge_def::count): New accessor.
11751 * bb-reorder.c (rotate_loop): Update.
11752 (find_traces_1_round): Update.
11753 (connect_traces): Update.
11754 (sanitize_hot_paths): Update.
11755 * cfg.c (unchecked_make_edge): Update.
11756 (make_single_succ_edge): Update.
11757 (check_bb_profile): Update.
11758 (dump_edge_info): Update.
11759 (update_bb_profile_for_threading): Update.
11760 (scale_bbs_frequencies_int): Update.
11761 (scale_bbs_frequencies_gcov_type): Update.
11762 (scale_bbs_frequencies_profile_count): Update.
11763 (scale_bbs_frequencies): Update.
11764 * cfganal.c (connect_infinite_loops_to_exit): Update.
11765 * cfgbuild.c (compute_outgoing_frequencies): Update.
11766 (find_many_sub_basic_blocks): Update.
11767 * cfgcleanup.c (try_forward_edges): Update.
11768 (try_crossjump_to_edge): Update
11769 * cfgexpand.c (expand_gimple_cond): Update
11770 (expand_gimple_tailcall): Update
11771 (construct_exit_block): Update
11772 * cfghooks.c (verify_flow_info): Update
11773 (redirect_edge_succ_nodup): Update
11774 (split_edge): Update
11775 (make_forwarder_block): Update
11776 (duplicate_block): Update
11777 (account_profile_record): Update
11778 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11779 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11780 * cfgloopmanip.c (scale_loop_profile): Update.
11782 (lv_adjust_loop_entry_edge): Update.
11783 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11784 (force_nonfallthru_and_redirect): Update.
11785 (purge_dead_edges): Update.
11786 (rtl_flow_call_edges_add): Update.
11787 * cgraphunit.c (init_lowered_empty_function): Update.
11788 (cgraph_node::expand_thunk): Update.
11789 * gimple-pretty-print.c (dump_probability): Update.
11790 (dump_edge_probability): Update.
11791 * gimple-ssa-isolate-paths.c (isolate_path): Update.
11792 * haifa-sched.c (sched_create_recovery_edges): Update.
11793 * hsa-gen.c (convert_switch_statements): Update.
11794 * ifcvt.c (dead_or_predicable): Update.
11795 * ipa-inline-transform.c (inline_transform): Update.
11796 * ipa-split.c (split_function): Update.
11797 * ipa-utils.c (ipa_merge_profiles): Update.
11798 * loop-doloop.c (add_test): Update.
11799 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
11800 * lto-streamer-in.c (input_cfg): Update.
11801 (input_function): Update.
11802 * lto-streamer-out.c (output_cfg): Update.
11803 * modulo-sched.c (sms_schedule): Update.
11804 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11805 * predict.c (maybe_hot_edge_p): Update.
11806 (unlikely_executed_edge_p): Update.
11807 (probably_never_executed_edge_p): Update.
11808 (dump_prediction): Update.
11809 (drop_profile): Update.
11810 (propagate_unlikely_bbs_forward): Update.
11811 (determine_unlikely_bbs): Update.
11812 (force_edge_cold): Update.
11813 * profile.c (compute_branch_probabilities): Update.
11814 * reg-stack.c (better_edge): Update.
11815 * shrink-wrap.c (handle_simple_exit): Update.
11816 * tracer.c (better_p): Update.
11817 * trans-mem.c (expand_transaction): Update.
11818 (split_bb_make_tm_edge): Update.
11819 * tree-call-cdce.c: Update.
11820 * tree-cfg.c (gimple_find_sub_bbs): Update.
11821 (gimple_split_edge): Update.
11822 (gimple_duplicate_sese_region): Update.
11823 (gimple_duplicate_sese_tail): Update.
11824 (gimple_flow_call_edges_add): Update.
11825 (insert_cond_bb): Update.
11826 (execute_fixup_cfg): Update.
11827 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
11828 * tree-complex.c (expand_complex_div_wide): Update.
11829 * tree-eh.c (lower_resx): Update.
11830 (unsplit_eh): Update.
11831 (cleanup_empty_eh_move_lp): Update.
11832 * tree-inline.c (copy_edges_for_bb): Update.
11833 (freqs_to_counts): Update.
11834 (copy_cfg_body): Update.
11835 * tree-ssa-dce.c (remove_dead_stmt): Update.
11836 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11837 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11838 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11839 (unloop_loops): Update.
11840 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11841 * tree-ssa-loop-split.c (connect_loops): Update.
11842 (split_loop): Update.
11843 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11844 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11845 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11846 * tree-ssa-reassoc.c (branch_fixup): Update.
11847 * tree-ssa-tail-merge.c (replace_block_by): Update.
11848 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11849 (compute_path_counts): Update.
11850 (update_profile): Update.
11851 (recompute_probabilities): Update.
11852 (update_joiner_offpath_counts): Update.
11853 (estimated_freqs_path): Update.
11854 (freqs_to_counts_path): Update.
11855 (clear_counts_path): Update.
11856 (ssa_fix_duplicate_block_edges): Update.
11857 (duplicate_thread_path): Update.
11858 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11859 (case_bit_test_cmp): Update.
11860 (collect_switch_conv_info): Update.
11861 (gen_inbound_check): Update.
11862 (do_jump_if_equal): Update.
11863 (emit_cmp_and_jump_insns): Update.
11864 * tree-tailcall.c (decrease_profile): Update.
11865 (eliminate_tail_call): Update.
11866 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11867 (vect_do_peeling): Update.
11868 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11869 * ubsan.c (ubsan_expand_null_ifn): Update.
11870 (ubsan_expand_ptr_ifn): Update.
11871 * value-prof.c (gimple_divmod_fixed_value): Update.
11872 (gimple_mod_pow2): Update.
11873 (gimple_mod_subtract): Update.
11874 (gimple_ic): Update.
11875 (gimple_stringop_fixed_value): Update.
11877 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11880 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11882 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11884 PR rtl-optimization/82395
11885 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11886 based on non_spilled_static_chain_regno_p.
11888 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11890 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11891 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11892 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11894 2017-10-19 Martin Sebor <msebor@redhat.com>
11896 PR tree-optimization/82596
11897 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11899 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11901 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11902 (get_mem_refs_of_builtin_call): Likewise.
11903 * builtins.c (expand_builtin_apply): Adjust call to
11904 allocate_dynamic_stack_space.
11905 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11906 the third argument to allocate_dynamic_stack_space, otherwise -1.
11907 (expand_builtin): Deal with all alloca variants.
11908 (is_inexpensive_builtin): Likewise.
11909 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11910 * calls.c (special_function_p): Deal with all alloca variants.
11911 (initialize_argument_information): Adjust call to
11912 allocate_dynamic_stack_space.
11913 (expand_call): Likewise.
11914 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11915 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11916 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11917 use it for the stack usage computation.
11918 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11919 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11920 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11921 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11922 (in_loop_p): Remove first argument and useless check.
11923 (pass_walloca::execute): Remove useless test and adjust call to above.
11924 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11925 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11926 (gimplify_call_expr): Deal with all alloca variants.
11927 * hsa-gen.c (gen_hsa_alloca): Likewise.
11928 (gen_hsa_insns_for_call): Likewise.
11929 * ipa-pure-const.c (special_builtin_state): Likewise.
11930 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11931 * tree-object-size.c (alloc_object_size): Likewise.
11932 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11933 (call_may_clobber_ref_p_1): Likewise.
11934 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11935 (ccp_fold_stmt): Likewise.
11936 (optimize_stack_restore): Likewise.
11937 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11938 (mark_all_reaching_defs_necessary_1): Likewise.
11939 (propagate_necessity): Likewise.
11940 (eliminate_unnecessary_stmts): Likewise.
11941 * tree.c (build_common_builtin_nodes): Build
11942 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11943 (build_alloca_call_expr): New function.
11944 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11945 (CASE_BUILT_IN_ALLOCA): Likewise.
11946 (build_alloca_call_expr): Declare.
11947 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11949 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11952 * dwarf2out.c (new_die_raw): New static inline function.
11953 (new_die): Use it to create the DIE.
11954 (add_AT_external_die_ref): Likewise.
11955 (clone_die): Likewise.
11956 (clone_as_declaration): Likewise.
11957 (dwarf2out_vms_debug_main_pointer): Likewise.
11958 (base_type_die): Likewise. Remove early return for corner cases.
11959 Do not call add_pubtype on the DIE here.
11960 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11961 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11962 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11963 native order exists for base types, attach the DIE manually and call
11964 add_pubtype on it. Do not equate a reverse order DIE to the type.
11966 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11968 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11969 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11971 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11973 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11975 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11976 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11977 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11978 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11979 * opts.c (sanitizer_opts): Add builtin.
11980 * ubsan.c (instrument_builtin): New function.
11981 (pass_ubsan::execute): Call it.
11982 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11983 * doc/invoke.texi: Document -fsanitize=builtin.
11985 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11986 builtins, store max (log2 (align), 0) into uchar field instead of
11987 align into uptr field.
11988 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11989 store uchar 0 field instead of uptr 0 field.
11990 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11991 instead of passing one address of struct with 2 locations pass
11992 two addresses of structs with 1 location each.
11993 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11994 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11995 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11996 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11997 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11998 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11999 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
12000 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
12002 2017-10-19 Martin Liska <mliska@suse.cz>
12005 * file-find.c (remove_prefix): Remove.
12006 * file-find.h (remove_prefix): Likewise.
12007 * gcc-ar.c: Remove smartness of lookup.
12009 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
12011 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
12012 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
12013 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
12015 2017-10-19 Jakub Jelinek <jakub@redhat.com>
12018 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
12019 (setcc + and to xor + setcc): New peephole2.
12021 2017-10-19 Tom de Vries <tom@codesourcery.com>
12023 * doc/sourcebuild.texi (Test Directives, Variants of
12024 dg-require-support): Add dg-require-stack-size.
12026 2017-10-19 Martin Liska <mliska@suse.cz>
12029 * gimplify.c (gimplify_decl_expr): Do not instrument variables
12030 that have a large alignment.
12031 (gimplify_target_expr): Likewise.
12033 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
12035 PR rtl-optimization/82602
12036 * ira.c (rtx_moveable_p): Return false for volatile asm.
12038 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
12041 * config/i386/i386-modes.def (CCGZ): New CC mode.
12042 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
12043 * config/i386/predicates.md (ix86_comparison_operator):
12045 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
12046 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
12047 with double-word subtraction.
12048 (put_condition_code): Handle CCGZmode.
12050 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
12052 * wide-int.cc (debug (const wide_int &)): New.
12053 (debug (const wide_int *)): New.
12054 (debug (const widest_int &)): New.
12055 (debug (const widest_int *)): New.
12057 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
12059 PR middle-end/82556
12060 * lra-constraints.c (curr_insn_transform): Use non-input operand
12061 instead of output one for matched reload.
12063 2017-10-18 Bin Cheng <bin.cheng@arm.com>
12065 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
12066 (tree-ssa-loop-ivopts.h): New header file.
12067 (struct builtin_info): New fields.
12068 (classify_builtin_1): Compute and record base and offset parts for
12069 memset builtin partition by calling strip_offset.
12070 (offset_cmp, fuse_memset_builtins): New functions.
12071 (finalize_partitions): Fuse adjacent memset partitions by calling
12073 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
12074 Expose the interface.
12075 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
12077 2017-10-18 Bin Cheng <bin.cheng@arm.com>
12079 PR tree-optimization/82574
12080 * tree-loop-distribution.c (find_single_drs): New parameter. Check
12081 that data reference must be executed exactly once per iteration
12082 against the outermost loop in nest.
12083 (classify_partition): Update call to above function.
12085 2017-10-18 Richard Biener <rguenther@suse.de>
12087 PR tree-optimization/82591
12088 * graphite.c (graphite_transform_loops): Move code gen message
12090 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12091 Here. Handle scop_to_isl_ast failing.
12092 (scop_to_isl_ast): Limit the number of ISL operations.
12094 2017-10-18 Richard Biener <rguenther@suse.de>
12096 * graphite-isl-ast-to-gimple.c
12097 (translate_isl_ast_to_gimple::set_rename): Simplify.
12098 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
12099 (graphite_copy_stmts_from_block): ... here.
12100 (copy_bb_and_scalar_dependences): Simplify.
12101 (add_parameters_to_ivs_params): Canonicalize.
12102 (generate_entry_out_of_ssa_copies): Simplify.
12103 * graphite-sese-to-poly.c (extract_affine_name): Simplify
12104 by passing in ISL dimension.
12105 (parameter_index_in_region_1): Rename to ...
12106 (parameter_index_in_region): ... this.
12107 (extract_affine): Adjust assert, pass down parameter index.
12108 (add_param_constraints): Use range-info when available.
12109 (build_scop_context): Adjust.
12110 * sese.c (new_sese_info): Adjust.
12111 (free_sese_info): Likewise.
12112 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
12113 Remove unused typedefs.
12114 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
12116 2017-10-18 Martin Liska <mliska@suse.cz>
12118 * combine.c (simplify_compare_const): Add gcc_fallthrough.
12120 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
12122 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
12123 (s390_sched_init): Do not reset s390_sched_state if we entered the
12124 current basic block via a fallthru edge and all others are unlikely.
12126 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
12128 * config/s390/s390.c (NUM_SIDES): New variable.
12129 (LONGRUNNING_THRESHOLD): New variable.
12130 (LATENCY_FACTOR): New variable.
12131 (s390_sched_score): Decrease score for long-running instructions on
12133 (s390_sched_variable_issue): Perform bookkeeping for long-running
12136 2017-10-18 Richard Biener <rguenther@suse.de>
12138 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
12139 Simplify with removal of the parameter rename map.
12140 (set_rename): Likewise.
12141 (should_copy_to_new_region): Likewise.
12142 (graphite_copy_stmts_from_block): Likewise.
12143 (copy_bb_and_scalar_dependences): Remove initialization of
12144 unused copied_bb_map.
12145 (copy_def): Remove.
12146 (copy_internal_parameters): Likewise.
12147 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
12148 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
12149 Use INTEGRAL_TYPE_P.
12150 (parameter_index_in_region_1): Rename to ...
12151 (assign_parameter_index_in_region): ... this. Assert we have
12152 a parameter we handle.
12153 (scan_tree_for_params): Adjust.
12154 * sese.h (parameter_rename_map_t): Remove.
12155 (struct sese_info_t): Remove unused parameter_rename_map and
12156 copied_bb_map members.
12157 * sese.c (new_sese_info): Adjust.
12158 (free_sese_info): Likewise.
12160 2017-10-18 Martin Liska <mliska@suse.cz>
12163 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
12164 on an abnormal edge.
12166 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
12168 * doc/invoke.texi (ffunction-sections and fdata-sections):
12171 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
12173 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
12174 the use statement can throw internally.
12176 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
12178 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
12179 any RTX present on the RHS of a SET.
12180 * compare-elim.c (try_eliminate_compare): Restore comment.
12182 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12184 * langhooks.h (struct lang_hooks): Document that tree_size langhook
12185 may be also called on tcc_type nodes.
12186 * langhooks.c (lhd_tree_size): Likewise.
12188 2017-10-17 David Malcolm <dmalcolm@redhat.com>
12190 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
12191 format_warning_at_substring.
12192 (maybe_warn): Convert source_range * param to a location_t. Pass
12193 UNKNOWN_LOCATION rather than NULL to fmtwarn.
12194 (format_directive): Remove code to extract source_ranges and
12195 source_range * in favor of just a location_t.
12196 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
12198 * substring-locations.c (format_warning_va): Convert
12199 source_range * param to a location_t.
12200 (format_warning_at_substring): Likewise.
12201 * substring-locations.h (format_warning_va): Likewise.
12202 (format_warning_at_substring): Likewise.
12204 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
12206 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
12208 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
12209 and scatter/gather ops.
12211 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
12212 vec_gather_load and vec_scatter_store.
12213 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
12214 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
12215 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
12216 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
12217 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
12218 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
12220 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
12222 * reg-stack.c (compare_for_stack_reg): Add bool argument.
12223 Detect FTST instruction and handle its register pops. Only pop
12224 second operand if can_pop_second_op is true.
12225 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
12226 set can_pop_second_op to false in the compare_for_stack_reg call.
12228 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
12229 output_fp_compare for stack register operands.
12230 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
12231 instructions here. Do not emit stack register pops here. Assert
12232 that FCOMPP pops next to top stack register. Rewrite function.
12234 2017-10-17 Nathan Sidwell <nathan@acm.org>
12236 PR middle-end/82577
12237 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
12238 use DECL_ASSEMBLER_NAME_RAW.
12240 PR middle-end/82546
12241 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
12244 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
12245 Wilco Dijkstra <wilco.dijkstra@arm.com>
12247 * builtins.c (expand_builtin_update_setjmp_buf): Add a
12248 converstion to Pmode from the buf_addr.
12250 2017-10-17 Richard Biener <rguenther@suse.de>
12252 * graphite-dependences.c (scop_get_reads_and_writes): Change
12253 output parameters to references.
12255 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
12257 PR 71026/tree-optimization
12258 * fold-const.c (distribute_real_division): Removed.
12259 (fold_binary_loc): Remove calls to distribute_real_divison.
12261 2017-10-17 Richard Biener <rguenther@suse.de>
12263 * graphite-scop-detection.c
12264 (scop_detection::stmt_has_simple_data_refs_p): Always use
12265 the full nest as region.
12266 (try_generate_gimple_bb): Likewise.
12267 * sese.c (scalar_evolution_in_region): Simplify now that
12268 SCEV can handle instantiation in regions.
12269 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
12270 in the non-loop part of a function if requested.
12272 2017-10-17 Richard Biener <rguenther@suse.de>
12274 PR tree-optimization/82563
12275 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
12277 (graphite_regenerate_ast_isl): Call it.
12278 * graphite-scop-detection.c (build_scops): Remove entry edge split.
12280 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12282 PR tree-optimization/82549
12283 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
12284 Formatting fixes. Instead of calling make_bit_field_ref with negative
12287 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
12289 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
12290 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
12291 _mm_maskz_reduce_ss): New.
12292 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
12293 __builtin_ia32_reducess_mask): Ditto..
12294 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
12295 * config/i386/sse.md (reduces<mode>): Renamed to ...
12296 (reduces<mode><mask_scalar_name>): ... this.
12297 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
12299 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
12300 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
12302 2017-10-16 David Malcolm <dmalcolm@redhat.com>
12304 * Makefile.in (OBJS): Add unique-ptr-tests.o.
12305 * selftest-run-tests.c (selftest::run_tests): Call
12306 selftest::unique_ptr_tests_cc_tests.
12307 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
12308 * unique-ptr-tests.cc: New file.
12310 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
12313 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12315 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12316 (make_hard_regno_born, make_hard_regno_dead): Update
12317 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
12319 2017-10-16 Jeff Law <law@redhat.com>
12321 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12323 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
12325 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
12327 2017-10-16 Olivier Hainque <hainque@adacore.com>
12329 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
12330 with_cpu if we were configured for an e500v2 target cpu name.
12332 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
12334 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
12335 * doc/invoke.texi: Document +nodsp as a valid extension for
12338 2017-10-16 Martin Liska <mliska@suse.cz>
12340 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
12341 (test_set_range): Likewise.
12342 (test_range_functions): Rename to ...
12343 (test_bit_in_range): ... this.
12344 (sbitmap_c_tests): Add new test.
12346 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12348 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
12350 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
12351 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
12353 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12355 * config/aarch64/aarch64-builtins.c
12356 (aarch64_types_quadopu_lane_qualifiers): New.
12357 (TYPES_QUADOPU_LANE): New.
12358 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
12359 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
12360 (aarch64_<sur>dot_laneq<vsi2qi>): New.
12361 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
12362 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
12363 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
12364 (Vdottype, DOTPROD): New.
12365 (sur): Add SDOT and UDOT.
12367 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12369 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
12370 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
12371 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
12372 Add TARGET_DOTPROD.
12373 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
12374 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
12375 Enable TARGET_DOTPROD.
12376 (cortex-a75.cortex-a55): Likewise.
12377 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
12379 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12381 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
12382 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
12384 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
12386 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
12387 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
12388 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
12389 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
12390 * config/arm/types.md (neon_dot, neon_dot_q): New.
12391 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
12393 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12395 * config/arm/arm.h (TARGET_DOTPROD): New.
12396 * config/arm/arm.c (arm_arch_dotprod): New.
12397 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
12398 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
12399 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
12400 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
12401 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
12402 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
12403 * doc/invoke.texi (armv8.2-a): Document dotprod
12405 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12407 * i386.c (ix86_vec_cost): New function.
12408 (ix86_rtx_costs): Handle vector operations better.
12409 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
12410 * x86-tune-costs.h: Add new costs to all tables.
12412 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12414 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
12416 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
12417 divsd, sqrtss and sqrtsd
12418 * x86-tune-costs.h: Add new entries to all costs.
12419 (znver1_cost): Fix to match real instruction latencies.
12421 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12422 Michael Collison <michael.collison@arm.com>
12424 * compare-elim.c: Include emit-rtl.h.
12425 (can_merge_compare_into_arith): New function.
12426 (try_validate_parallel): Likewise.
12427 (try_merge_compare): Likewise.
12428 (try_eliminate_compare): Call the above when no previous clobber
12430 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
12433 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12435 PR middle-end/62263
12436 PR middle-end/82498
12437 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
12438 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
12440 PR middle-end/62263
12441 PR middle-end/82498
12442 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
12443 to be any operand_equal_p operands. For & (B - 1) require
12444 B to be power of 2. Recognize
12445 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
12447 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12450 * optabs.c (expand_memory_blockage): Fix call of
12451 targetm.have_memory_blockage.
12453 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12456 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
12457 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
12458 objects to extra_objs instead of overwriting it.
12460 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12462 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
12463 Use any_fp_register_operand as operand[3] predicate. Simplify
12464 equality test for operands[2] and operands[4] memory location.
12465 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
12466 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
12467 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
12468 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
12469 any_fp_register_operand as operand[1] predicate. Simplify
12470 equality test for operands[0] and operands[3] memory location.
12471 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
12472 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
12473 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
12475 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12477 * target-insns.def: Add memory_blockage.
12478 * optabs.c (expand_memory_blockage): New function.
12479 (expand_asm_memory_barrier): Rename ...
12480 (expand_asm_memory_blockage): ... to this.
12481 (expand_mem_thread_fence): Call expand_memory_blockage
12482 instead of expand_asm_memory_barrier.
12483 (expand_mem_singnal_fence): Ditto.
12484 (expand_atomic_load): Ditto.
12485 (expand_atomic_store): Ditto.
12486 * doc/md.texi (Standard Pattern Names For Generation):
12487 Document memory_blockage instruction pattern.
12489 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
12491 * config/rl78/rl78.c (rl78_emit_libcall): New function.
12492 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
12493 * config/rl78/rl78.md: New define_expand "adddi3".
12495 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12497 * cfghooks.c (verify_flow_info): Disable check that all probabilities
12500 2017-10-13 Jeff Law <law@redhat.com>
12502 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
12504 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12507 * internal-fn.c (expand_mul_overflow): If both operands have
12508 the same highpart of -1 or 0 and the topmost bit of lowpart
12509 is different, overflow is if res <= 0 rather than res < 0.
12511 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
12513 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
12514 TARGET_P9_VECTOR code for unaligned_load case.
12516 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12518 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
12520 2017-10-13 Nathan Sidwell <nathan@acm.org>
12522 * tree-core.h (tree_contains_struct): Make bool.
12523 * tree.c (tree_contains_struct): Likewise.
12524 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
12525 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
12526 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
12527 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
12528 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
12530 2017-10-13 Richard Biener <rguenther@suse.de>
12532 * graphite-isl-ast-to-gimple.c
12533 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
12534 parameters and dominance check.
12535 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
12536 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
12537 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
12538 Do not update SSA form here or do intermediate IL verification.
12539 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
12540 (graphite_initialize): Remove check on the number of loops in
12541 the function and inline into graphite_transform_loops.
12542 (graphite_finalize): Inline into graphite_transform_loops.
12543 (graphite_transform_loops): Perform SSA update and IL verification
12545 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
12547 2017-10-13 Richard Biener <rguenther@suse.de>
12549 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
12550 graphite_expression_type_precision): Avoid global constructor
12552 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
12553 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
12554 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
12555 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
12556 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
12557 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
12559 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
12562 * config/i386/i386.h (ix86_red_zone_size): New.
12563 * config/i386/i386.md (push peephole2s): Replace
12564 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
12566 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12567 Alan Hayward <alan.hayward@arm.com>
12568 David Sherwood <david.sherwood@arm.com>
12570 * combine.c (can_change_dest_mode): Reject changes in
12571 REGMODE_NATURAL_SIZE.
12573 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12574 Alan Hayward <alan.hayward@arm.com>
12575 David Sherwood <david.sherwood@arm.com>
12577 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
12578 (expand_debug_source_expr): Likewise.
12579 * combine.c (combine_simplify_rtx): Likewise.
12580 * cse.c (fold_rtx): Likewise.
12581 * fwprop.c (canonicalize_address): Likewise.
12582 * targhooks.c (default_shift_truncation_mask): Likewise.
12584 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12585 Alan Hayward <alan.hayward@arm.com>
12586 David Sherwood <david.sherwood@arm.com>
12588 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
12589 (widened_mode): Likewise.
12590 (expand_unop): Likewise.
12591 * ree.c (transform_ifelse): Likewise.
12592 (merge_def_and_ext): Likewise.
12593 (combine_reaching_defs): Likewise.
12594 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12596 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12597 Alan Hayward <alan.hayward@arm.com>
12598 David Sherwood <david.sherwood@arm.com>
12600 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
12601 * combine.c (gen_lowpart_for_combine): Likewise.
12602 * dwarf2out.c (rtl_for_decl_location): Likewise.
12603 * final.c (alter_subreg): Likewise.
12604 * rtlhooks.c (gen_lowpart_general): Likewise.
12605 (gen_lowpart_if_possible): Likewise.
12607 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12608 Alan Hayward <alan.hayward@arm.com>
12609 David Sherwood <david.sherwood@arm.com>
12611 * calls.c (expand_call): Use subreg_lowpart_offset.
12612 * cse.c (cse_insn): Likewise.
12613 * regcprop.c (copy_value): Likewise.
12614 (copyprop_hardreg_forward_1): Likewise.
12616 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12619 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
12620 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
12621 =Q constraints to +Q and into insn condition add check
12622 that operands[0] and operands[1] are equal.
12623 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
12624 =Q constraints to +Q and into insn condition add check
12625 that operands[0] is equal to either operands[1] or operands[2].
12628 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
12629 instead of handling MINUS_EXPR twice (once for each argument),
12630 canonicalize operand order and handle just once, use rtype where
12631 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
12634 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
12635 any values of __C while still being pattern recognizable as a simple
12636 rotate instruction.
12638 2017-10-13 Richard Biener <rguenther@suse.de>
12640 PR tree-optimization/82451
12642 2017-10-02 Richard Biener <rguenther@suse.de>
12644 PR tree-optimization/82355
12645 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12646 a mapping for the enclosing loop but avoid generating one for
12647 the loop tree root.
12648 (copy_bb_and_scalar_dependences): Remove premature codegen
12649 error on PHIs in blocks duplicated into multiple places.
12650 * graphite-scop-detection.c
12651 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12652 in the region use it as loop and nest to analyze the DR in.
12653 (try_generate_gimple_bb): Likewise.
12654 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12655 (add_loop_constraints): For blocks in a loop not in the region
12656 create a dimension with a single iteration.
12657 * sese.h (gbb_loop_at_index): Remove assert.
12659 * cfgloop.c (loop_preheader_edge): For the loop tree root
12660 return the single successor of the entry block.
12661 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12662 Reset the SCEV hashtable and niters.
12663 * graphite-scop-detection.c
12664 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
12665 assert that we only have POLYNOMIAL_CHREC that vary in loops
12666 contained in the region.
12667 (scop_detection::graphite_can_represent_expr): Adjust.
12668 (scop_detection::stmt_has_simple_data_refs_p): For loops
12669 not in the region set loop to NULL. The nest is now the
12670 entry edge to the region.
12671 (try_generate_gimple_bb): Likewise.
12672 * sese.c (scalar_evolution_in_region): Adjust for
12673 instantiate_scev change.
12674 * tree-data-ref.h (graphite_find_data_references_in_stmt):
12675 Make nest parameter the edge into the region.
12676 (create_data_ref): Likewise.
12677 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
12678 entry edge into a region and adjust instantiate_scev calls.
12679 (create_data_ref): Likewise.
12680 (graphite_find_data_references_in_stmt): Likewise.
12681 (find_data_references_in_stmt): Pass the loop preheader edge
12682 from the nest argument.
12683 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
12684 parameter the edge into the region.
12685 (instantiate_parameters): Use the loop preheader edge as entry.
12686 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
12688 (get_instantiated_value_entry): Make instantiate_below parameter
12689 the edge into the region.
12690 (instantiate_scev_name): Likewise. Adjust dominance checks,
12691 when we cannot use loop-based instantiation instantiate by
12692 walking use-def chains.
12693 (instantiate_scev_poly): Adjust.
12694 (instantiate_scev_binary): Likewise.
12695 (instantiate_scev_convert): Likewise.
12696 (instantiate_scev_not): Likewise.
12697 (instantiate_array_ref): Remove.
12698 (instantiate_scev_3): Likewise.
12699 (instantiate_scev_2): Likewise.
12700 (instantiate_scev_1): Likewise.
12701 (instantiate_scev_r): Do not blindly handle N-operand trees.
12702 Do not instantiate array-refs. Handle all constants and invariants.
12703 (instantiate_scev): Make instantiate_below parameter
12704 the edge into the region.
12705 (resolve_mixers): Use the loop preheader edge for the region
12706 parameter to instantiate_scev_r.
12707 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
12709 2017-10-13 Richard Biener <rguenther@suse.de>
12711 PR tree-optimization/82525
12712 * graphite-isl-ast-to-gimple.c
12713 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
12715 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
12716 Fail code generation when we cannot represent the isl integer.
12717 (binary_op_to_tree): Elide modulo operations that are no-ops
12718 in the type we code generate. Remove now superfluous code
12721 2017-10-13 Richard Biener <rguenther@suse.de>
12723 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
12724 (scop_detection::harmful_loop_in_region): Remove premature
12725 IV type restriction.
12726 (scop_detection::graphite_can_represent_scev): We can handle
12727 pointer IVs just fine.
12729 2017-10-13 Alan Modra <amodra@gmail.com>
12731 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
12732 "Clobbers and Scratch Registers". Add paragraph on
12733 alternative to clobbers for scratch registers and OpenBLAS
12736 2017-10-13 Alan Modra <amodra@gmail.com>
12738 * doc/extend.texi (Clobbers): Correct vax example. Delete old
12739 example of a memory input for a string of known length. Move
12740 commentary out of table. Add a number of new examples
12741 covering array memory inputs.
12743 2017-10-12 Martin Liska <mliska@suse.cz>
12745 PR tree-optimization/82493
12746 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
12747 (test_range_functions): New function.
12748 (sbitmap_c_tests): Likewise.
12749 * selftest-run-tests.c (selftest::run_tests): Run new tests.
12750 * selftest.h (sbitmap_c_tests): New function.
12752 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12754 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12756 * config/rs6000/amo.h: Fix spacing issue.
12758 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12761 * config/i386/i386.md (*ashl<mode>3_mask_1,
12762 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
12763 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
12766 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12768 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
12769 (profile_probability): Set max_probability
12770 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
12772 * profile-count.c (profile_probability::differs_from_p): Do not
12773 rely on max_probaiblity == 10000
12775 2017-10-12 Jeff Law <law@redhat.com>
12777 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
12780 2017-10-12 Martin Sebor <msebor@redhat.com>
12784 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
12785 (handle_alias_pairs): Call it.
12786 * common.opt (-Wattribute-alias): New option.
12787 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
12788 * doc/invoke.texi (-Wattribute-alias): Document.
12790 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
12793 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12795 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12797 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12798 (make_hard_regno_born, make_hard_regno_dead): Update
12799 bb_killed_pseudos and bb_gen_pseudos.
12801 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12803 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
12805 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
12807 * config/alpha/alpha.c (alpha_split_conditional_move):
12808 Use std::swap instead of manually swapping.
12809 (alpha_stdarg_optimize_hook): Ditto.
12810 (alpha_canonicalize_comparison): Ditto.
12812 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12814 * tree-loop-distribution.c (struct builtin_info): New struct.
12815 (struct partition): Refactor fields into struct builtin_info.
12816 (partition_free): Free struct builtin_info.
12817 (build_size_arg_loc, build_addr_arg_loc): Delete.
12818 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
12819 information from struct builtin_info.
12820 (find_single_drs): New function refactored from classify_partition.
12821 Also moved builtin validity checks to this function.
12822 (compute_access_range, alloc_builtin): New functions.
12823 (classify_builtin_st, classify_builtin_ldst): New functions.
12824 (classify_partition): Refactor code into functions find_single_drs,
12825 classify_builtin_st and classify_builtin_ldst.
12826 (distribute_loop): Don't do runtime alias check when distributing
12828 (find_seed_stmts_for_distribution): New function.
12829 (pass_loop_distribution::execute): Refactor code finding seed
12830 stmts into above function. Support distribution for the innermost
12831 two-level loop nest. Adjust dump information.
12833 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12835 * tree-loop-distribution.c: Adjust the general comment.
12836 (NUM_PARTITION_THRESHOLD): New macro.
12837 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
12838 (classify_partition): Skip builtin pattern of loop nest's inner loop.
12839 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
12840 in call to build_partition_graph.
12841 (finalize_partitions): New parameter. Make loop distribution more
12842 conservative by fusing more partitions.
12843 (distribute_loop): Don't do runtime alias check in case of loop nest
12845 (find_seed_stmts_for_distribution): New function.
12846 (prepare_perfect_loop_nest): New function.
12847 (pass_loop_distribution::execute): Refactor code finding seed stmts
12848 and loop nest into above functions. Support loop nest distribution.
12849 Adjust dump information accordingly.
12851 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12853 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12854 and set PTYPE_SEQUENTIAL for merged partition.
12856 2017-10-12 Richard Biener <rguenther@suse.de>
12858 PR tree-optimization/69728
12860 2017-09-19 Richard Biener <rguenther@suse.de>
12862 PR tree-optimization/69728
12863 * graphite-sese-to-poly.c (schedule_error): New global.
12864 (add_loop_schedule): Handle empty domain by failing the
12866 (build_original_schedule): Handle schedule_error.
12868 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12869 domain by returning an unchanged schedule.
12871 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12873 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12874 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12876 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12878 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12881 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12884 * expr.c (store_field): Don't optimize away bitsize == 0 store
12885 from CALL_EXPR with addressable return type.
12887 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12889 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12890 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12891 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12892 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12893 TARGET_ISEL instead of TARGET_ISEL<sel>.
12895 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12897 * config/rs6000/rs6000.c
12898 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12900 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12902 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12904 (reg_or_cint_operand): Fix comment.
12905 (reg_or_zero_operand): New predicate.
12906 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12907 * config/rs6000/rs6000.c (output_isel): Delete.
12908 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12909 instead of reg_or_cint_operand. Output instruction directly (not via
12911 (isel_unsigned_<mode>): Ditto.
12912 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12913 gpc_reg_operand. Add an instruction alternative for this. Output
12914 instruction directly.
12915 (*isel_reversed_unsigned_<mode>): Ditto.
12917 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12919 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12920 (TARGET_CANONICALIZE_COMPARISON): Define.
12922 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12925 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12926 Check whether the dest is REG before adding REG_EQUIV note.
12928 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12931 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12933 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12934 (make_hard_regno_born, make_hard_regno_dead): Update
12935 bb_killed_pseudos and bb_gen_pseudos.
12937 2017-10-11 Nathan Sidwell <nathan@acm.org>
12939 * incpath.h (enum incpath_kind): Name enum, prefix values.
12940 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12941 * incpath.c (heads, tails): Use INC_MAX.
12942 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12943 (merge_include_chains, split_quote_chain,
12944 register_include_chains): Update incpath_kind names.
12945 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12946 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12948 (add_framework_path, darwin_register_objc_includes): Likewise.
12949 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12951 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12953 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12954 Do not use float_operator operator predicate.
12955 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12956 * config/i386/predicates.md (float_operator): Remove predicate.
12958 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12960 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12961 (*jccxf_i387): Ditto.
12962 (*jcc<mode>_i387): Ditto.
12963 (*jccu<mode>_i387): Ditto.
12964 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12965 (*jcc_*_i387 splitters): Remove.
12966 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12967 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12968 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12971 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12973 * profile-count.h (slow_safe_scale_64bit): New function.
12974 (safe_scale_64bit): New inline.
12975 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12976 * profile-count.c: Include wide-int.h
12977 (slow_safe_scale_64bit): New.
12979 2017-10-11 Nathan Sidwell <nathan@acm.org>
12981 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12982 HAS_DECL_ASSEMBLER_NAME_P.
12983 * gimple-expr.c (gimple_decl_printable_name: Check
12984 HAS_DECL_ASSEMBLER_NAME_P too.
12985 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12986 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12987 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12988 * passes.c (rest_of_decl_compilation): Check
12989 HAS_DECL_ASSEMBLER_NAME_P too.
12990 * recog.c (verify_changes): Likewise.
12991 * tree-pretty-print.c (dump_decl_name): Likewise.
12992 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12994 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12995 (SET_DECL_ASSEMBLER_NAME): Use it.
12996 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12997 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12998 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
13000 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
13002 * config.gcc (i386, x86_64): Add extra objects.
13003 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
13004 (ix86_min_insn_size): Declare.
13005 (ix86_issue_rate): Declare.
13006 (ix86_adjust_cost): Declare.
13007 (ia32_multipass_dfa_lookahead): Declare.
13008 (ix86_macro_fusion_p): Declare.
13009 (ix86_macro_fusion_pair_p): Declare.
13010 (ix86_bd_has_dispatch): Declare.
13011 (ix86_bd_do_dispatch): Declare.
13012 (ix86_core2i7_init_hooks): Declare.
13013 (ix86_atom_sched_reorder): Declare.
13014 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
13015 (COSTS_N_BYTES): Move to x86-tune-costs.h.
13016 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
13017 (rip_relative_addr_p): Rename to ...
13018 (ix86_rip_relative_addr_p): ... this one; export.
13019 (memory_address_length): Update.
13020 (ix86_issue_rate): Move to x86-tune-sched.c.
13021 (ix86_flags_dependent): Move to x86-tune-sched.c.
13022 (ix86_agi_dependent): Move to x86-tune-sched.c.
13023 (exact_dependency_1): Move to x86-tune-sched.c.
13024 (exact_store_load_dependency): Move to x86-tune-sched.c.
13025 (ix86_adjust_cost): Move to x86-tune-sched.c.
13026 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
13027 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
13028 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
13029 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
13030 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
13031 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
13032 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
13033 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
13034 (min_insn_size): Rename to ...
13035 (ix86_min_insn_size): ... this one; export.
13036 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
13037 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
13038 (core2i7_first_cycle_multipass_backtrack): Move to
13039 x86-tune-sched-core.c.
13040 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
13041 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
13042 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
13043 (ix86_avoid_jump_mispredicts): Update.
13044 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
13045 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
13046 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
13047 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
13048 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
13049 (MAX_INSN): Move to ix86-tune-sched-bd.c.
13050 (MAX_IMM): Move to ix86-tune-sched-bd.c.
13051 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
13052 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
13053 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
13054 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
13055 (MAX_STORE): Move to ix86-tune-sched-bd.c.
13056 (BIG): Move to ix86-tune-sched-bd.c.
13057 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
13058 (enum insn_path): Move to ix86-tune-sched-bd.c.
13059 (get_mem_group): Move to ix86-tune-sched-bd.c.
13060 (is_cmp): Move to ix86-tune-sched-bd.c.
13061 (dispatch_violation): Move to ix86-tune-sched-bd.c.
13062 (is_branch): Move to ix86-tune-sched-bd.c.
13063 (is_prefetch): Move to ix86-tune-sched-bd.c.
13064 (init_window): Move to ix86-tune-sched-bd.c.
13065 (allocate_window): Move to ix86-tune-sched-bd.c.
13066 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
13067 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
13068 (process_end_window): Move to ix86-tune-sched-bd.c.
13069 (allocate_next_window): Move to ix86-tune-sched-bd.c.
13070 (find_constant): Move to ix86-tune-sched-bd.c.
13071 (get_num_immediates): Move to ix86-tune-sched-bd.c.
13072 (has_immediate): Move to ix86-tune-sched-bd.c.
13073 (get_insn_path): Move to ix86-tune-sched-bd.c.
13074 (get_insn_group): Move to ix86-tune-sched-bd.c.
13075 (count_num_restricted): Move to ix86-tune-sched-bd.c.
13076 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
13077 (add_insn_window): Move to ix86-tune-sched-bd.c.
13078 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
13079 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
13080 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
13081 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
13082 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
13083 (do_dispatch): Move to ix86-tune-sched-bd.c.
13084 (has_dispatch): Move to ix86-tune-sched-bd.c.
13085 * config/i386/t-i386: Add new object files.
13086 * config/i386/x86-tune-costs.h: New file.
13087 * config/i386/x86-tune-sched-atom.c: New file.
13088 * config/i386/x86-tune-sched-bd.c: New file.
13089 * config/i386/x86-tune-sched-core.c: New file.
13090 * config/i386/x86-tune-sched.c: New file.
13092 2017-10-11 Liu Hao <lh_mouse@126.com>
13094 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
13095 the generic version below instead.
13096 (should_colorize): Recognize Windows consoles as terminals
13098 * pretty-print.c [__MINGW32__] (write_all): New function.
13099 [__MINGW32__] (find_esc_head): Likewise.
13100 [__MINGW32__] (find_esc_terminator): Likewise.
13101 [__MINGW32__] (eat_esc_sequence): Likewise.
13102 [__MINGW32__] (mingw_ansi_fputs): New function that handles
13104 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
13107 2017-10-11 Richard Biener <rguenther@suse.de>
13109 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
13110 Properly call analyze_scalar_evolution with the loop of the stmt.
13112 2017-10-11 Richard Biener <rguenther@suse.de>
13114 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
13115 * tree-core.h (tree_base): Add chrec_var union member.
13116 * tree.h (CHREC_VAR): Remove.
13117 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
13118 * tree-chrec.h (build_polynomial_chrec): Adjust.
13119 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
13120 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
13122 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
13124 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
13125 * match.pd: ... here.
13126 ((T) X == (T) Y): Relax condition.
13128 2017-10-11 Bin Cheng <bin.cheng@arm.com>
13130 PR tree-optimization/82472
13131 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
13133 (break_alias_scc_partitions): Update postorder number.
13135 2017-10-11 Martin Liska <mliska@suse.cz>
13138 * opts.c (parse_no_sanitize_attribute): Do not use error_value
13140 * opts.h (parse_no_sanitize_attribute): Remove last argument.
13142 2017-10-11 Martin Liska <mliska@suse.cz>
13144 * print-rtl.c (print_insn): Move declaration of idbuf
13145 to same scope as name.
13147 2017-10-11 Martin Liska <mliska@suse.cz>
13152 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
13153 volatile arguments.
13155 2017-10-11 Martin Liska <mliska@suse.cz>
13158 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
13159 volatile arguments.
13161 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
13163 * config.gcc (default_gnu_indirect_function): Default to yes for
13164 arm*-*-linux* with glibc.
13166 2017-10-11 Richard Biener <rguenther@suse.de>
13168 * tree-scalar-evolution.c (get_scalar_evolution): Handle
13169 default-defs and types we do not want to analyze.
13170 (interpret_loop_phi): Replace unreachable code with an assert.
13171 (compute_scalar_evolution_in_loop): Remove and inline ...
13172 (analyze_scalar_evolution_1): ... here, replacing condition with
13173 what makes the intent clearer. Remove handling of cases
13174 get_scalar_evolution now handles.
13176 2017-10-10 Jim Wilson <wilson@tuliptree.org>
13178 PR rtl-optimization/81434
13179 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
13180 comment for main loop. In sched_group_found if, also add checks for
13181 pass and min_cost_group.
13183 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
13185 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
13186 (rs6000_insn_cost): New function.
13187 * config/rs6000/rs6000.md (cost): New attribute.
13189 2017-10-10 Jakub Jelinek <jakub@redhat.com>
13190 H.J. Lu <hongjiu.lu@intel.com>
13194 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
13195 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
13196 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
13197 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
13198 to OPTION_MASK_ISA_AVX512VL - builtins that have both
13199 OPTION_MASK_ISA_MMX and some other bit set require both
13200 mmx and the ISAs without the mmx bit.
13201 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
13202 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
13203 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
13204 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
13205 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
13206 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
13207 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
13208 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
13209 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
13210 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
13211 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
13212 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
13213 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
13214 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
13215 Add OPTION_MASK_ISA_MMX.
13217 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
13219 * config.gcc (armv7*-*-freebsd*): New target.
13220 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
13222 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
13224 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
13225 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
13228 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
13230 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
13231 a template parameter.
13232 (WIDE_INT_REF_FOR): Update accordingly.
13233 * tree.h (wi::int_traits <const_tree>): Delete.
13234 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
13235 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
13236 (wi::tree_to_wide_ref): New typedef.
13237 (wi::to_wide): New function.
13238 * calls.c (get_size_range): Use wi::to_wide when operating on
13239 trees as wide_ints.
13240 * cgraph.c (cgraph_node::create_thunk): Likewise.
13241 * config/i386/i386.c (ix86_data_alignment): Likewise.
13242 (ix86_local_alignment): Likewise.
13243 * dbxout.c (stabstr_O): Likewise.
13244 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
13245 * expr.c (const_vector_from_tree): Likewise.
13246 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
13247 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
13248 (fold_negate_expr_1, int_const_binop_1, const_binop)
13249 (fold_convert_const_int_from_real, optimize_bit_field_compare)
13250 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
13251 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
13252 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
13253 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
13254 (fold_not_const, round_up_loc): Likewise.
13255 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
13256 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
13257 (alloca_call_type): Likewise.
13258 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
13259 * godump.c (go_output_typedef): Likewise.
13260 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
13261 * internal-fn.c (get_min_precision): Likewise.
13262 * ipa-cp.c (ipcp_store_vr_results): Likewise.
13263 * ipa-polymorphic-call.c
13264 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
13265 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
13266 (ipa_modify_call_arguments): Likewise.
13267 * match.pd: Likewise.
13268 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
13269 * print-tree.c (print_node_brief, print_node): Likewise.
13270 * stmt.c (expand_case): Likewise.
13271 * stor-layout.c (layout_type): Likewise.
13272 * tree-affine.c (tree_to_aff_combination): Likewise.
13273 * tree-cfg.c (group_case_labels_stmt): Likewise.
13274 * tree-data-ref.c (dr_analyze_indices): Likewise.
13275 (prune_runtime_alias_test_list): Likewise.
13276 * tree-dump.c (dequeue_and_dump): Likewise.
13277 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
13278 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
13279 * tree-pretty-print.c (dump_generic_node): Likewise.
13280 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
13281 (simple_iv_with_niters): Likewise.
13282 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
13283 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
13284 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
13285 * tree-ssa-loop-niter.c (split_to_var_and_offset)
13286 (refine_value_range_using_guard, number_of_iterations_ne_max)
13287 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
13288 (get_cst_init_from_scev, record_nonwrapping_iv)
13289 (scev_var_range_cant_overflow): Likewise.
13290 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
13291 * tree-ssa-pre.c (compute_avail): Likewise.
13292 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
13293 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
13294 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
13295 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
13296 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
13297 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
13298 (dump_case_nodes, try_switch_expansion): Likewise.
13299 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
13300 (vect_do_peeling): Likewise.
13301 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
13302 * tree-vect-stmts.c (vectorizable_load): Likewise.
13303 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
13304 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
13305 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
13306 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
13307 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
13308 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
13309 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
13310 (evrp_dom_walker::before_dom_children): Likewise.
13311 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
13312 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
13313 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
13314 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
13315 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
13316 (get_range_pos_neg): Likewise.
13317 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
13318 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
13319 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
13320 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
13321 * config/avr/avr.c (avr_fold_builtin): Likewise.
13322 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
13323 * config/msp430/msp430.c (msp430_attr): Likewise.
13324 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
13325 * config/powerpcspe/powerpcspe-c.c
13326 (altivec_resolve_overloaded_builtin): Likewise.
13327 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
13328 (rs6000_expand_ternop_builtin): Likewise.
13329 * config/rs6000/rs6000-c.c
13330 (altivec_resolve_overloaded_builtin): Likewise.
13331 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
13332 (rs6000_expand_ternop_builtin): Likewise.
13333 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
13335 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13337 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
13338 when copying loop nest with only one inner loop.
13340 2017-10-10 Richard Biener <rguenther@suse.de>
13342 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
13343 blocks if SCEV is active.
13344 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
13346 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
13347 (scev_initialize): Assert we are not yet initialized.
13349 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13351 * tree-loop-distribution.c (generate_loops_for_partition): Remove
13352 inner loop's exit stmt by making it always exit the loop, otherwise
13353 we would generate an infinite empty loop.
13355 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13357 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
13358 renaming variables in new preheader if it's deleted.
13360 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13362 * tree-loop-distribution.c (struct partition): Remove unused field
13363 loops of the structure.
13364 (partition_alloc, partition_free): Ditto.
13365 (build_rdg_partition_for_vertex): Ditto.
13367 2017-10-09 Jeff Law <law@redhat.com>
13369 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
13370 return type to match prototype and documentation.
13372 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13374 * config/rs6000/rs6000.c (processor_costs): Move to ...
13375 * config/rs6000/rs6000.h: ... here.
13376 (rs6000_cost): Declare.
13378 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
13380 * except.c (setjmp_fn): New global variable.
13381 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
13382 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
13383 if DONT_USE_BUILTIN_SETJMP is defined.
13385 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13387 * target.def (insn_cost): New hook.
13388 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
13389 * doc/tm.texi: Regenerate.
13390 * rtlanal.c (insn_cost): Use the new hook.
13392 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13394 * combine.c (combine_validate_cost): Compute the new insn_cost,
13395 not just pattern_cost.
13396 (try_combine): Adjust comment.
13398 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13400 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
13402 * combine.c (uid_insn_cost): Adjust comment.
13403 (combine_validate_cost): Adjust comment. Use pattern_cost instead
13405 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
13406 * dse.c (find_shift_sequence): Ditto.
13407 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
13408 (bb_valid_for_noce_process_p): Use pattern_cost.
13409 * rtl.h (insn_rtx_cost): Delete.
13410 (pattern_cost): New prototype.
13411 (insn_cost): New prototype.
13412 * rtlanal.c (insn_rtx_cost): Rename to...
13413 (pattern_cost): ... this.
13416 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
13418 * config/i386/i386.md (*jcc_2): Remove insn pattern.
13419 (*jcc<mode>_0_r_i387): Ditto.
13420 (*jccxf_r_i387): Ditto.
13421 (*jcc<mode>_r_i387): Ditto.
13422 (*jccu<mode>_r_i387): Ditto.
13423 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
13424 (*jcc): Rename from *jcc_1.
13426 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13428 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
13429 deferred rescans after the lvx/stvx recombination pre-pass.
13431 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13433 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
13434 memory operation instruction support.
13435 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
13436 (rs6000-ibm-aix[789]*): Likewise.
13437 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
13438 Document new functions.
13440 2017-10-09 Richard Biener <rguenther@suse.de>
13442 PR tree-optimization/82397
13443 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
13444 equality only for semantically equal trees.
13446 2017-10-09 Richard Biener <rguenther@suse.de>
13448 PR tree-optimization/82449
13449 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
13450 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
13451 allow constant addresses.
13452 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
13455 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13457 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
13460 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13463 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
13466 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13469 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
13471 2017-10-09 Jakub Jelinek <jakub@redhat.com>
13474 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
13475 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
13477 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
13479 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
13480 (WI_BINARY_PREDICATE_RESULT): Likewise.
13481 (wi::binary_traits::operator_result): New type.
13482 (wi::binary_traits::predicate_result): Likewise.
13483 (generic_wide_int::operator~, unary generic_wide_int::operator-)
13484 (generic_wide_int::operator==, generic_wide_int::operator!=)
13485 (generic_wide_int::operator&, generic_wide_int::and_not)
13486 (generic_wide_int::operator|, generic_wide_int::or_not)
13487 (generic_wide_int::operator^, generic_wide_int::operator+
13488 (binary generic_wide_int::operator-, generic_wide_int::operator*):
13490 (operator~, unary operator-, operator==, operator!=, operator&)
13491 (operator|, operator^, operator+, binary operator-, operator*): New
13493 * expr.c (get_inner_reference): Use wi::bit_and_not.
13494 * fold-const.c (fold_binary_loc): Likewise.
13495 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
13496 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
13497 (bit_value_binop): Likewise.
13498 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
13499 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
13500 (extract_range_from_binary_expr_1): Likewise.
13501 (masked_increment): Likewise.
13502 (simplify_bit_ops_using_ranges): Likewise.
13504 2017-10-09 Martin Jambor <mjambor@suse.cz>
13507 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
13508 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
13509 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
13511 (hsa_fixup_mov_insn_type): New function.
13512 (hsa_op_with_type::get_in_type): Use it.
13513 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
13514 immediates in an assert.
13515 (hsa_op_with_type::extend_int_to_32bit): New method.
13516 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
13517 types. Convert to dest type if necessary.
13518 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
13519 (reg_for_gimple_ssa): Pass false as min32int to
13520 hsa_type_for_scalar_tree_type.
13521 (gen_hsa_addr): Fixup type when creating addresable temporary.
13522 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
13523 (gen_hsa_unary_operation): Extend operands and convert to dest type if
13524 necessary. Call hsa_fixup_mov_insn_type.
13525 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
13526 extend operands and convert to dest type if necessary.
13527 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
13528 to dest type if necessary.
13529 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
13530 if conversion nt necessary and size matches.
13531 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
13532 to dest type if necessary.
13533 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
13534 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
13536 (gen_hsa_clrsb): Likewise.
13537 (gen_hsa_ffs): Likewise.
13538 (gen_hsa_divmod): Extend operands and convert to dest type if
13540 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
13542 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13544 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
13545 Remove empty default arguments. Use a brace block as output
13547 (conditional return): Ditto.
13549 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
13550 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
13552 (group_ending_nop): Ditto.
13553 (doloop_end): Ditto.
13554 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
13555 (splitters for those): Ditto.
13557 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13559 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
13560 a conditional jump (and the compare for it) so that pc_rtx is the
13562 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
13563 for the deleted and renamed ctr<mode>_internal[234] patterns.
13564 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
13565 Delete second conditional return pattern.
13566 (ctr<mode>_internal2): Delete this second bdnz pattern.
13567 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
13568 (ctr<mode>_internal4): Delete this second bdz pattern.
13570 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13572 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
13573 (always_initialized_rtx_for_ssa_name_p): New predicate.
13574 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
13575 (finish_out_of_ssa): Free new field of SA.
13576 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
13577 * tree-ssa-coalesce.c: Include tree-ssa.h.
13578 (get_parm_default_def_partitions): Remove extern keyword.
13579 (get_undefined_value_partitions): New function.
13580 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
13581 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
13582 uninitialized bits.
13583 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
13585 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13587 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
13589 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13591 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
13592 for targets that preffer 128bit.
13594 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13596 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
13598 2017-10-08 Olivier Hainque <hainque@adacore.com>
13600 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
13601 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
13603 (thumb_set_return_address): Likewise.
13605 2017-10-08 Olivier Hainque <hainque@adacore.com>
13607 * common/config/arm/arm-common.c (arm_except_unwind_info):
13608 Handle DWARF2_UNWIND_INFO.
13610 2017-10-07 Michael Collison <michael.collison@arm.com>
13612 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
13615 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
13617 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
13618 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
13619 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
13620 defined, force the creation of a new block for a dispatch label.
13622 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
13624 * invoke.texi (Wsuggest-attribute=cold): Document.
13625 * common.opt (Wsuggest-attribute=cold): New
13626 * ipa-pure-const.c (warn_function_cold): New function.
13627 * predict.c (compute_function_frequency): Use it.
13628 * predict.h (warn_function_cold): Declare.
13630 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
13632 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
13635 2017-10-06 Martin Liska <mliska@suse.cz>
13637 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
13638 keyword for member functions.
13639 (struct sanopt_tree_couple): New struct.
13640 (struct sanopt_tree_couple_hash): New function.
13641 (struct sanopt_ctx): Add new hash_map.
13642 (has_dominating_ubsan_ptr_check): New function.
13643 (record_ubsan_ptr_check_stmt): Likewise.
13644 (maybe_optimize_ubsan_ptr_ifn): Likewise.
13645 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
13646 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
13648 2017-10-06 Sudakshina Das <sudi.das@arm.com>
13651 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
13652 aarch64_simd_valid_immediate on CONST_VECTORs.
13653 (aarch64_reg_or_bic_imm): Likewise.
13655 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
13657 PR rtl-optimization/82396
13658 * haifa-sched.c (ready_sort_real): Disable qsort checking.
13660 2017-10-06 Sebastian Pop <sebpop@gmail.com>
13662 * graphite-dependences.c (scop_get_reads): Move code to...
13663 (scop_get_must_writes): Move code to...
13664 (scop_get_may_writes): Move code to...
13665 (scop_get_reads_and_writes): ... here.
13666 (scop_get_dependences): Call scop_get_reads_and_writes.
13668 2017-10-06 Jakub Jelinek <jakub@redhat.com>
13670 PR tree-optimization/82434
13671 * fold-const.h (can_native_encode_type_p,
13672 can_native_encode_string_p): Remove.
13673 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
13674 don't encode anything, just return what would be otherwise returned.
13675 (native_encode_fixed, native_encode_complex, native_encode_vector):
13677 (native_encode_string): Likewise. Inline by hand
13678 can_native_encode_string_p.
13679 (can_native_encode_type_p): Remove.
13680 (can_native_encode_string_p): Remove.
13681 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
13682 STRING_CSTs using can_native_encode_string_p, test all
13683 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
13684 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
13685 argument from native_encode_expr.
13686 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
13687 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
13690 2017-10-06 Richard Biener <rguenther@suse.de>
13692 PR tree-optimization/82397
13693 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
13694 operand_equal_p but rely on data_ref_compare_tree for detecting
13696 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
13697 to match up with dr_group_sort_cmp.
13699 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13702 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
13704 * config/s390/s390-builtin-types.def: Regenerate.
13706 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13709 * config/s390/s390-builtin-types.def: Regenerate.
13710 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
13711 Change flag from B_VXE to B_VX.
13712 (s390_vec_min_dbl): Remove B_VXE flag.
13714 2017-10-06 Richard Biener <rguenther@suse.de>
13716 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
13717 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
13718 translate_isl_ast_to_gimple::is_valid_rename,
13719 translate_isl_ast_to_gimple::get_rename,
13720 translate_isl_ast_to_gimple::get_def_bb_for_const,
13721 translate_isl_ast_to_gimple::get_new_name,
13722 translate_isl_ast_to_gimple::collect_all_ssa_names,
13723 translate_isl_ast_to_gimple::copy_loop_phi_args,
13724 translate_isl_ast_to_gimple::collect_all_ssa_names,
13725 translate_isl_ast_to_gimple::copy_loop_phi_args,
13726 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
13727 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
13728 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
13729 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
13730 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
13731 translate_isl_ast_to_gimple::copy_cond_phi_args,
13732 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
13733 translate_isl_ast_to_gimple::edge_for_new_close_phis,
13734 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
13735 translate_isl_ast_to_gimple::rename_uses,
13736 translate_isl_ast_to_gimple::rename_all_uses): Remove.
13737 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
13738 (set_rename_for_each_def): Likewise.
13739 (graphite_copy_stmts_from_block): Handle debug stmt resetting
13740 here. Handle rewriting SCEV analyzable uses here.
13741 (copy_bb_and_scalar_dependences): Generate code for PHI
13743 (graphite_regenerate_ast_isl): Adjust.
13744 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
13745 (add_write, add_read): New functions.
13746 (build_cross_bb_scalars_def): Use it and simplify.
13747 (build_cross_bb_scalars_use): Likewise.
13748 (graphite_find_cross_bb_scalar_vars): Inline into...
13749 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
13750 simulating out-of-SSA. Compute liveout and add dependencies.
13751 (build_scops): Force an empty entry block.
13752 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
13754 (sese_build_liveouts): Declare.
13755 (sese_trivially_empty_bb_p): Likewise.
13756 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
13757 compute liveout and debug_liveout.
13758 (sese_bad_liveouts_use): Remove.
13759 (sese_reset_debug_liveouts_bb): Likewise.
13760 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
13761 (sese_build_liveouts): Build liveout and debug_liveout and store
13763 (new_sese_info): Adjust.
13764 (free_sese_info): Likewise.
13765 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
13766 do not build liveout here.
13767 (move_sese_in_condition): Adjust region entry.
13768 (scev_analyzable_p): Match up with chrec_apply requirements.
13769 (sese_trivially_empty_bb_p): New.
13770 * tree-into-ssa.c (get_reaching_def): Properly support generating
13771 default-defs for incremental rewrite of anonymous names.
13773 2017-10-06 Richard Biener <rguenther@suse.de>
13775 * graphite-sese-to-poly.c (extract_affine): For casts increasing
13776 precision do not perform modulo reduction.
13778 2017-10-06 Richard Biener <rguenther@suse.de>
13780 PR tree-optimization/82436
13781 * tree-vect-slp.c (vect_supported_load_permutation_p): More
13782 conservatively choose the vectorization factor when checking
13783 whether we can perform the required load permutation.
13784 (vect_transform_slp_perm_load): Assert when we may not fail.
13786 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
13788 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
13789 message for incompatible -msdata=* and -mcall-* options.
13791 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13793 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
13794 rate for post-reload scheduling.
13796 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13798 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
13800 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13802 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
13803 to improve monte carlo in scimark.
13805 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13807 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
13808 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
13809 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
13810 pentium4_cost, nocona_cost): Set reassociation width to 1.
13811 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
13812 width to 2 for fp operations and 1 otherwise.
13813 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
13815 (atom_cost): Set reassociation width to 2.
13816 (slm_cost, generic_cost): Set fp reassociation width
13817 to 2 and 1 otherwise.
13818 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
13819 (core_cost): Set fp reassociation width to 4 and vector to 2.
13820 (ix86_reassociation_width): Rewrite using cost table; special case
13821 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
13822 and TARGET_AVX128_OPTIMAL.
13823 * config/i386/i386.h (processor_costs): Add
13824 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
13825 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
13826 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
13827 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
13828 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
13829 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
13831 2017-10-05 Nathan Sidwell <nathan@acm.org>
13833 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
13835 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13837 * config/arm/arm.c (arm_test_fpu_data): New.
13838 (arm_run_selftests): Call arm_test_fpu_data.
13840 2017-10-04 Nathan Sidwell <nathan@acm.org>
13842 * toplev.c (toplev::main): Remove excess parens on pretty_printer
13844 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
13846 2017-10-04 Sudakshina Das <sudi.das@arm.com>
13848 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13849 check type for aarch64_simd_valid_immediate.
13850 (aarch64_output_simd_mov_immediate): Update prototype.
13851 (aarch64_simd_valid_immediate): Update prototype.
13852 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13853 support for ORR-immediate.
13854 (and<mode>3): modified pattern to add support for BIC-immediate.
13855 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13856 now checks for valid immediate for BIC and ORR based on new enum
13858 (aarch64_output_simd_mov_immediate): Function now used to output
13859 BIC/ORR imm as well based on new enum argument.
13860 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13862 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13863 (aarch64_reg_or_bic_imm): Likewise.
13865 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13867 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13868 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13870 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13874 PR rtl-optimization/82396
13875 * haifa-sched.c (autopref_multipass_init): Simplify
13877 (autopref_rank_data): Simplify sort order.
13878 * sched-int.h (autopref_multipass_data_): Remove
13879 multi_mem_insn_p, min_offset and max_offset.
13881 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13883 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13885 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13887 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13888 vect_doubleint_cvt.
13890 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13892 * doc/sourcebuild.texi: Document vect_long_mult.
13894 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13896 PR tree-optimization/82413
13897 * fold-const.c (build_range_check): Use widest_int when comparing
13898 the maximum ETYPE value with HIGH.
13900 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13902 PR rtl-optimization/82396
13903 * haifa-sched.c (autopref_multipass_init): Simplify
13905 (autopref_rank_data): Simplify sort order.
13906 * sched-int.h (autopref_multipass_data_): Remove
13907 multi_mem_insn_p, min_offset and max_offset.
13909 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13911 PR tree-optimization/82381
13912 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13913 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13916 PR tree-optimization/82374
13917 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13918 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13919 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13920 current_function_decl to the new decl.
13922 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13924 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13925 helper macro for IEEE float128 hardware built-in functions.
13926 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13928 (TRUNCF128_ODD): Likewise.
13929 (ADDF128_ODD): Likewise.
13930 (SUBF128_ODD): Likewise.
13931 (MULF128_ODD): Likewise.
13932 (DIVF128_ODD): Likewise.
13933 (FMAF128_ODD): Likewise.
13934 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13935 UNSPEC_TRUNC_ROUND_TO_ODD.
13936 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13937 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13938 floating point round to odd instructions.
13939 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13940 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13941 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13942 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13943 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13944 (trunc<mode>sf2_hw): Change the truncate with round to odd
13945 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13946 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13947 to odd hardware instructions.
13948 (sub<mode>3_odd): Likewise.
13949 (mul<mode>3_odd): Likewise.
13950 (div<mode>3_odd): Likewise.
13951 (sqrt<mode>2_odd): Likewise.
13952 (fma<mode>4_odd): Likewise.
13953 (fms<mode>4_odd): Likewise.
13954 (nfma<mode>4_odd): Likewise.
13955 (nfms<mode>4_odd): Likewise.
13956 (trunc<mode>df2_odd): Change the truncate with round to odd
13957 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13959 * doc/extend.texi (PowerPC built-in functions): Update documentation
13960 for existing IEEE float128-bit built-in functions. Add built-in
13961 functions that generate the IEEE 128-bit floating point round to
13964 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13966 PR rtl-optimization/77729
13967 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13968 to (X&(C1&~C2))|C2 transformations.
13970 2017-10-03 Martin Jambor <mjambor@suse.cz>
13972 PR tree-optimization/82363
13973 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13974 mismatch, mark lacc written regardless of racc.
13976 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13978 PR tree-optimization/82381
13979 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13980 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13981 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13982 fallthrough into reassoc_stmt_dominates_stmt_p.
13985 * combine.c (combine_instructions): Don't combine in unreachable
13988 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13991 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13992 function to not use the have_cpu variable. Do not set cpu_index,
13993 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13994 or the default cpu.
13995 (rs6000_valid_attribute_p): Remove duplicate initializations of
13996 old_optimize and func_optimize.
13997 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13998 (rs6000_activate_target_options): Make global.
13999 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
14002 2017-10-02 Jakub Jelinek <jakub@redhat.com>
14004 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
14005 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
14006 Set *poffset to 0 and *psize and *pmax_size to -1 if
14007 *poffset + *psize overflows in HOST_WIDE_INT.
14009 PR tree-optimization/82387
14010 PR tree-optimization/82388
14011 PR tree-optimization/82389
14012 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
14013 instead of live_bytes non-NULL.
14015 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
14018 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
14021 2017-10-02 Richard Biener <rguenther@suse.de>
14023 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
14024 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
14025 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
14027 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
14029 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
14030 requested precision matches the type's.
14031 * calls.c (alloc_max_size): Calculate the new candidate size as
14032 a widest_int and use wi::to_widest when comparing it with the
14033 current candidate size.
14034 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
14035 zero rather than integer_zero_node.
14036 * match.pd: Check for a no-op conversion before using wi::add
14037 rather than after. Use tree_to_uhwi when summing small shift
14038 counts into an unsigned int.
14040 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
14041 Alan Hayward <alan.hayward@arm.com>
14042 David Sherwood <david.sherwood@arm.com>
14045 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
14046 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
14047 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
14048 POINTER_AND_FP_REGS.
14050 2017-10-02 Richard Biener <rguenther@suse.de>
14052 PR tree-optimization/82355
14053 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
14054 a mapping for the enclosing loop but avoid generating one for
14055 the loop tree root.
14056 (copy_bb_and_scalar_dependences): Remove premature codegen
14057 error on PHIs in blocks duplicated into multiple places.
14058 * graphite-scop-detection.c
14059 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
14060 in the region use it as loop and nest to analyze the DR in.
14061 (try_generate_gimple_bb): Likewise.
14062 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
14063 (add_loop_constraints): For blocks in a loop not in the region
14064 create a dimension with a single iteration.
14065 * sese.h (gbb_loop_at_index): Remove assert.
14067 2017-10-01 Kevin Buettner <kevinb@redhat.com>
14069 * omp-expand.c (adjust_context_scope): New function.
14070 (expand_parallel_call): Call adjust_context_scope.
14072 2017-10-01 Jeff Law <law@redhat.com>
14074 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
14075 dom_opt_dom_walker class with direct access to private members.
14076 Add comments. Call test_for_singularity.
14077 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
14078 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
14079 m_dummy_cond anymore.
14080 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
14082 (pass_dominator:execute): Build the dummy_cond here and pass it
14083 to the dom_opt_dom_walker ctor.
14084 (test_for_singularity): New function.
14086 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
14087 Maya Rashish <coypu@sdf.org>
14089 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
14090 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
14091 (alpha*-*-netbsd*) Use nbsd_tm_file.
14092 (arm*-*-netbsdelf*) Likewise.
14093 (i[34567]86-*-netbsdelf*) Likewise.
14094 (x86_64-*-netbsd*) Likewise.
14095 (mips*-*-netbsd*) Likewise.
14096 (powerpc-*-netbsd*) Likewise.
14097 (sh*-*-netbsd*) Likewise.
14098 (sparc-*-netbsdelf*) Likewise.
14099 (sparc64-*-netbsd*) Likewise.
14100 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
14102 (vax-*-netbsdelf*) Likewise.
14103 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
14104 (UINT_FAST8_TYPE) Likewise.
14105 (INT_FAST16_TYPE) Check CHAR_FAST16.
14106 (UINT_FAST16_TYPE) Likewise.
14108 2017-09-30 Jakub Jelinek <jakub@redhat.com>
14111 * config/i386/i386.md
14112 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
14113 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
14114 *divmodsi4_zext_2): New define_insn_and_split.
14115 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
14116 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
14117 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
14118 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
14119 New define_insn_and_split.
14120 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
14121 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
14122 operands[1] having DImode when mode is SImode.
14124 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
14125 always SImode for DIV and MOD in REG_EQUAL notes.
14127 2017-09-29 Yury Gribov <tetra2005@gmail.com>
14129 PR middle-end/82319
14130 * match.pd: Fix handling of NaNs in pattern.
14132 2017-09-29 Jeff Law <law@redhat.com>
14134 * sbitmap.c (bitmap_bit_in_range_p): New function.
14135 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
14136 * tree-ssa-dse.c (live_bytes_read): New function.
14137 (dse_classify_store): Ignore reads of dead bytes.
14139 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
14140 typos and whitespace errors.
14141 * config/i386/predicates.md (address_no_seg_operand): Likewise.
14142 * config/s390/s390.c (s390_emit_prologue): Likewise.
14144 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
14147 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
14148 with a symbol for LRA.
14150 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
14152 PR rtl-optimization/82338
14153 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
14155 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
14157 * genmodes.c (calc_wider_mode): Suppress qsort macro.
14158 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
14159 (qsort_chk): Declare.
14160 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
14161 (qsort_chk): New function.
14163 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14165 PR tree-optimization/82337
14166 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
14167 phi definition if the PHI result appears in an abnormal PHI.
14168 (find_basis_for_base_expr): Don't record a basis if the LHS of the
14169 basis appears in an abnormal PHI.
14171 2017-09-29 Richard Biener <rguenther@suse.de>
14173 * graphite-isl-ast-to-gimple.c
14174 (translate_isl_ast_to_gimple::set_codegen_error): New function.
14175 (binary_op_to_tree): Use it.
14176 (get_rename_from_scev): Likewise.
14177 (copy_loop_phi_nodes): Likewise.
14178 (copy_bb_and_scalar_dependences): Likewise.
14179 (translate_pending_phi_nodes): Likewise.
14181 2017-09-29 Jakub Jelinek <jakub@redhat.com>
14184 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
14185 for movabsq $(i32 << shift), r64.
14187 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
14189 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
14190 index when encoding %esp as %rsp to avoid 0x67 prefix.
14192 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
14194 * config/i386/i386.md (*movsf_internal, *movdf_internal):
14195 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
14197 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
14199 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
14200 Extensions with more than 16 double VFP registers.
14201 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
14202 to_clear_mask and all code related to it. Replace the remaining
14203 entry by a sbitmap and adapt code accordingly.
14205 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
14207 * brig-builtins.def: Change pure attributes to const.
14209 2017-09-28 Joseph Myers <joseph@codesourcery.com>
14211 * config.gcc (default_gnu_indirect_function): Default to yes for
14212 sparc*-*-linux* with glibc.
14214 2017-09-28 Joseph Myers <joseph@codesourcery.com>
14216 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
14217 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
14218 when creating .init_array and .fini_array sections with priority
14221 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
14224 * config/aarch64/aarch64.c
14225 (aarch64_builtin_support_vector_misalignment): Always return false
14226 when misalignment is unknown.
14228 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
14230 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
14231 this function to return false if the definition used by the swap
14232 instruction is artificial, or if the memory address from which the
14233 constant value is loaded is not represented by a base address held
14234 in a register or if the base address register is a frame or stack
14235 pointer. Additionally, return false if the base address of the
14236 loaded constant is a SYMBOL_REF but is not considered to be a
14238 (replace_swapped_load_constant): New function.
14239 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
14240 loaded constant vector with a load of a swapped constant vector.
14242 2017-09-27 Carl Love <cel@us.ibm.com>
14244 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
14245 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
14246 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
14249 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
14251 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
14252 first, always call autopref_rank_data otherwise.
14254 2017-09-27 Richard Biener <rguenther@suse.de>
14256 * graphite-scop-detection.c (find_scop_parameters): Move
14257 loop bound handling ...
14258 (gather_bbs::before_dom_children): ... here, avoiding the need
14259 to build scop_info->loop_nest.
14260 (record_loop_in_sese): Remove.
14261 * sese.h (sese_info_t::loop_nest): Remove.
14262 * sese.c (new_sese_info): Do not allocate loop_nest.
14263 (free_sese_info): Do not free loop_nest.
14265 2017-09-27 Jakub Jelinek <jakub@redhat.com>
14268 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
14269 lhs from calls if the lhs has addressable type.
14271 2017-09-27 Richard Biener <rguenther@suse.de>
14273 * graphite.h (scop::max_alias_set): New member.
14274 * graphite-scop-detection.c: Remove references to non-existing
14275 --param in comments.
14276 (build_alias_sets): Record the maximum alias set used for drs.
14277 (build_scops): Support zero as unlimited for
14278 --param graphite-max-arrays-per-scop.
14279 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
14280 and inline into ...
14281 (build_poly_sr_1): ... here. Compute alias set based on the
14282 maximum alias set used for drs rather than
14283 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
14285 2017-09-27 Richard Biener <rguenther@suse.de>
14287 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
14288 --param loop-block-tile-size=0 to disable tiling.
14290 2017-09-27 Richard Biener <rguenther@suse.de>
14292 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
14293 (graphite-max-nb-scop-params): Document special value zero.
14294 * domwalk.h (dom_walker::STOP): New symbolical constant.
14295 (dom_walker::dom_walker): Add optional parameter for bb to
14297 (dom_walker::~dom_walker): Declare.
14298 (dom_walker::before_dom_children): Document STOP return value.
14299 (dom_walker::m_user_bb_to_rpo): New member.
14300 (dom_walker::m_bb_to_rpo): Likewise.
14301 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
14302 mapping here if not provided by the user.
14303 (dom_walker::~dom_walker): Free bb to RPO mapping if not
14304 provided by the user.
14305 (dom_walker::STOP): Define.
14306 (dom_walker::walk): Do not compute bb to RPO mapping here.
14307 Support STOP return value from before_dom_children to stop
14309 * graphite-optimize-isl.c (optimize_isl): If the schedule
14310 is the same still generate code if -fgraphite-identity
14311 or -floop-parallelize-all are given.
14312 * graphite-scop-detection.c: Include cfganal.h.
14313 (gather_bbs::gather_bbs): Get and pass through bb to RPO
14315 (gather_bbs::before_dom_children): Return STOP for BBs
14317 (build_scops): Compute bb to RPO mapping and pass it to
14318 the domwalk. Treat --param graphite-max-nb-scop-params=0
14319 as not limiting the number of params.
14320 * graphite.c (graphite_initialize): Remove limit on the
14321 number of basic-blocks in a function.
14322 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
14323 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
14324 default value of 10.
14326 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14328 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
14329 Adjust code to eliminate needing to do the shift right 32-bits
14330 operation after XSCVDPSPN.
14332 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14334 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
14335 ((X / Y) != 0 -> X >= Y): Likewise.
14337 2017-09-26 Carl Love <cel@us.ibm.com>
14339 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
14340 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
14341 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
14342 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
14343 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
14344 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
14345 definitions and overloading.
14346 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
14347 statement for P9V_BUILTIN_XST_LEN_R.
14348 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
14349 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
14350 define_expand and define_insn for the instructions and builtins.
14351 * doc/extend.texi: Update the built-in documentation file for the new
14352 built-in functions.
14353 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
14354 define_insn for the instructions
14356 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
14359 * gcc/config/netbsd-protos.h: New file.
14360 * gcc/config/netbsd.c: New file.
14361 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
14362 * gcc/config/t-netbsd: New file.
14363 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
14364 (tmake_file) Add t-netbsd.
14365 (extra_objs) Add netbsd.o.
14367 2017-09-26 Janus Weil <janus@gcc.gnu.org>
14371 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
14373 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14375 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
14376 sign extension from a vector register to a GPR by doing a 32-bit
14377 direct move and then an EXTSW.
14378 (extendsi<mode>2 splitter): Likewise.
14379 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
14380 right or vector extract after doing XSCVDPSPN. Use
14381 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
14383 (movdi_from_sf_zero_ext): Likewise.
14384 (reload_gpr_from_vsxsf): Likewise.
14385 (p8_mfvsrd_4_disf): Delete, no longer used.
14386 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
14387 then needing to move the SFmode to a GPR to use the XSCVDPSP
14388 instruction instead of FRSP and XSCVDPSPN.
14389 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
14390 it is adjacent to the other XSCVSPDP insns.
14391 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
14392 SFmode to be in traditional Altivec registers.
14393 (vsx_xscvdpspn): Eliminate useless alternative constraint.
14394 (vsx_xscvspdpn): Likewise.
14395 (vsx_xscvspdpn_scalar): Likewise.
14397 2017-09-26 Martin Jambor <mjambor@suse.cz>
14399 * tree-sra.c (compare_access_positions): Put integral types first,
14400 stabilize sorting of integral types, remove conditions putting
14401 non-full-precision integers last.
14402 (sort_and_splice_var_accesses): Disable scalarization if a
14403 non-integert would be represented by a non-full-precision integer.
14405 2017-09-26 Joseph Myers <joseph@codesourcery.com>
14407 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
14408 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
14409 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
14410 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
14411 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
14412 conditionals inside the function instead of around it. Call
14413 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
14414 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
14416 2017-09-26 Richard Biener <rguenther@suse.de>
14418 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
14420 (scop_detection::build_scop_breadth): ... this. Removed.
14421 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
14422 (scop_detection::harmful_stmt_in_bb): Likewise.
14423 (scop_detection::graphite_can_represent_stmt): Likewise.
14424 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
14425 (scop_detection::can_represent_loop): Remove recursion, fold in ...
14426 (scop_detection::can_represent_loop_1): ... this. Removed.
14427 (scop_detection::harmful_loop_in_region): Simplify after inlining
14428 the above and remove more quadraticness.
14429 (build_scops): Adjust.
14430 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
14433 2017-09-26 Jakub Jelinek <jakub@redhat.com>
14436 * config/i386/i386.c (ix86_print_operand_address_as): Only test
14437 REGNO (base) == SP_REG if base is a REG.
14439 PR middle-end/35691
14440 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
14441 if it is different SSA_NAME.
14442 (optimize_range_tests_cmp_bitwise): New function.
14443 (optimize_range_tests): Call it.
14445 2017-09-26 Richard Biener <rguenther@suse.de>
14447 PR tree-optimization/82321
14448 * graphite.c (canonicalize_loop_closed_ssa): Properly check
14449 for the def being inside the loop.
14451 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14453 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
14455 * config/s390/s390-builtins.def: Fix constraint on op4.
14457 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14459 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
14460 independent expanders.
14461 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
14462 ("vec_ordered", "vec_unordered"): New expanders.
14464 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14466 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
14469 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14471 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
14472 vec_unpacks_lo_v16qi.
14473 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
14475 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14477 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
14478 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
14479 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
14481 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14483 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
14485 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
14487 ("*vec_slb<mode>"): New insn pattern.
14488 ("vec_shr_<mode>"): New expander.
14489 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
14490 and force the shift count operand to V16QImode.
14491 ("vec_srb<mode>"): Set shift count mode to V16QI.
14493 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14495 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
14496 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
14497 ("vec_widen_smult_hi_<mode>"): New expander definitions.
14499 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
14502 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
14504 2017-09-26 Richard Biener <rguenther@suse.de>
14506 PR tree-optimization/82320
14507 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
14510 2017-09-25 Jeff Law <law@redhat.com>
14512 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
14513 prototype for new argument.
14514 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
14515 mostly extracted from rs6000_emit_allocate_stack.
14516 (rs6000_emit_probe_stack_range_stack_clash): New function.
14517 (rs6000_emit_allocate_stack): Call
14518 rs6000_emit_probe_stack_range_stack_clash as needed.
14519 (rs6000_emit_probe_stack_range): Add additional argument
14520 to call to gen_probe_stack_range{si,di}.
14521 (output_probe_stack_range): New.
14522 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
14523 (output_probe_stack_range_stack_clash): New.
14524 (rs6000_emit_prologue): Emit notes into dump file as requested.
14525 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
14526 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
14527 Add additional operand and pass it to output_probe_stack_range.
14529 2017-09-25 Bin Cheng <bin.cheng@arm.com>
14531 PR tree-optimization/82163
14532 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
14533 (checking_verify_loop_closed_ssa): New parameter.
14534 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
14535 (check_loop_closed_ssa_stmt): Delete.
14536 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
14537 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
14538 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
14541 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
14543 * brig-builtins.def: Treat HSAIL barrier builtins as
14544 setjmp/longjump style functions.
14546 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14548 * target.def (constant_alignment): New hook.
14549 * defaults.h (CONSTANT_ALIGNMENT): Delete.
14550 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
14551 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
14552 * doc/tm.texi: Regenerate.
14553 * targhooks.h (default_constant_alignment): Declare.
14554 (constant_alignment_word_strings): Likewise.
14555 * targhooks.c (default_constant_alignment): New function.
14556 (constant_alignment_word_strings): Likewise.
14557 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
14558 instead of CONSTANT_ALIGNMENT.
14559 * varasm.c (align_variable, get_variable_align, build_constant_desc)
14560 (force_const_mem): Likewise.
14561 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
14562 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
14563 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
14564 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14565 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
14567 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
14568 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14569 constant_alignment_word_strings.
14570 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
14571 (CONSTANT_ALIGNMENT): Likewise.
14572 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14573 (arm_constant_alignment): New function.
14574 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
14575 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14576 constant_alignment_word_strings.
14577 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
14578 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14579 constant_alignment_word_strings.
14580 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
14581 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14582 (cris_constant_alignment): New function.
14583 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
14584 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14585 (epiphany_constant_alignment): New function.
14586 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
14587 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14588 constant_alignment_word_strings.
14589 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
14590 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14591 constant_alignment_word_strings.
14592 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
14593 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14594 constant_alignment_word_strings.
14595 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
14596 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
14597 * config/i386/i386.c (ix86_constant_alignment): Make static.
14598 Use the same interface as the target hook.
14599 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14600 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
14601 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14602 constant_alignment_word_strings.
14603 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
14604 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
14605 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14606 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
14607 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14608 constant_alignment_word_strings.
14609 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
14610 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14611 constant_alignment_word_strings.
14612 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
14613 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14614 constant_alignment_word_strings.
14615 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
14616 * config/microblaze/microblaze.c (microblaze_constant_alignment):
14618 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14619 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
14620 * config/mips/mips.c (mips_constant_alignment): New function.
14621 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14622 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
14623 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
14624 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14625 (mmix_constant_alignment): Make static. Use the same interface
14626 as the target hook.
14627 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
14628 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14629 constant_alignment_word_strings.
14630 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
14631 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14632 constant_alignment_word_strings.
14633 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
14634 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14635 constant_alignment_word_strings.
14636 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
14637 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14638 (rs6000_constant_alignment): New function.
14639 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
14640 * config/riscv/riscv.c (riscv_constant_alignment): New function.
14641 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14642 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
14643 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14644 (rs6000_constant_alignment): New function.
14645 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
14646 * config/s390/s390.c (s390_constant_alignment): New function.
14647 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14648 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
14649 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14650 constant_alignment_word_strings.
14651 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
14652 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14653 (sparc_constant_alignment): New function.
14654 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
14655 * config/spu/spu.c (spu_constant_alignment): New function.
14656 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14657 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
14658 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14659 constant_alignment_word_strings.
14660 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
14661 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14662 constant_alignment_word_strings.
14663 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
14664 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14665 constant_alignment_word_strings.
14666 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
14667 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14668 (visium_constant_alignment): New function.
14669 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
14670 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14671 (xtensa_constant_alignment): New function.
14672 * system.h (CONSTANT_ALIGNMENT): Poison.
14674 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
14676 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14677 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
14678 (rs6000_builtin_valid_without_lhs): New helper function.
14679 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14680 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
14682 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14684 * target.h (vec_perm_indices): Use unsigned short rather than
14686 (auto_vec_perm_indices): Likewise.
14687 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
14688 Use unsigned int rather than unsigned char.
14689 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
14691 2017-09-25 Richard Biener <rguenther@suse.de>
14693 * cfgloop.h (sort_sibling_loops): Declare.
14694 * cfgloop.c (sort_sibling_loops_cmp): New helper.
14695 (sort_sibling_loops): New function sorting the sibling loop list
14697 * graphite.c (graphite_transform_loops): Sort sibling loops.
14699 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
14701 * target.def (vec_perm_const_ok): Change sel parameter to
14703 * optabs-query.c (can_vec_perm_p): Update accordingly.
14704 * doc/tm.texi: Regenerate.
14705 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
14706 auto_vec_perm_indices and remove separate nelt field.
14707 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
14708 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
14709 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
14710 (aarch64_expand_vec_perm_const): Update accordingly.
14711 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
14712 to vec_perm_indices.
14713 * config/arm/arm.c (expand_vec_perm_d): Change perm to
14714 auto_vec_perm_indices and remove separate nelt field.
14715 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
14716 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
14717 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
14719 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
14720 to vec_perm_indices.
14721 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
14722 sel to vec_perm_indices.
14723 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
14724 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
14725 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
14727 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
14730 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
14733 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
14734 on the FUNCTION_DECL function context if it has a DIE that is a
14737 2017-09-25 Richard Biener <rguenther@suse.de>
14739 PR tree-optimization/82285
14740 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
14743 2017-09-25 Tom de Vries <tom@codesourcery.com>
14747 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
14750 2017-09-25 Richard Biener <rguenther@suse.de>
14752 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
14753 ISL errors other than isl_error_quota happen. Dump if the
14754 schedule is the same.
14755 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
14756 errors instead of aborting inside ISL.
14758 2017-09-25 Iain Sandoe <iain@codesourcery.com>
14761 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
14762 of libgcc_eh for m64.
14763 * config/i386/darwin64.h: Likewise.
14765 2017-09-25 Richard Biener <rguenther@suse.de>
14767 PR middle-end/82144
14768 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
14769 attribute for incomplete types nor twice for complete ones.
14771 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
14774 * config/i386/i386.c (ix86_print_operand_address_as): Encode
14775 %esp as %rsp to avoid 0x67 prefix if there is no index or base
14778 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
14781 * config/i386/i386.opt (mprefer-avx256): Use
14782 ix86_target_flags variable.
14783 * config/i386/i386.c (ix86_target_string): Move
14784 -mprefer-avx256 to flag2_opts.
14786 2017-09-22 Jakub Jelinek <jakub@redhat.com>
14788 PR middle-end/35691
14789 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
14790 and x != -1 | y != -1 into (x & y) != -1.
14792 2017-09-22 Steve Ellcey <sellcey@cavium.com>
14794 * config.gcc: Add new case statement to set
14795 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
14796 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
14797 s390x-*-linux* case statements. Added aarch64 to the list of
14798 supported architectures.
14800 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14802 PR tree-optimization/82289
14803 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
14804 STMT_VINFO_RELEVANT_P.
14806 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14807 Alan Hayward <alan.hayward@arm.com>
14808 David Sherwood <david.sherwood@arm.com>
14810 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
14811 for VR_RANGE only; don't allow VR_ANTI_RANGE.
14812 (extract_range_from_binary_expr_1): Don't call
14813 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
14815 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14816 Alan Hayward <alan.hayward@arm.com>
14817 David Sherwood <david.sherwood@arm.com>
14819 * target.def (preferred_vector_alignment): New hook.
14820 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
14822 * doc/tm.texi: Regenerate.
14823 * targhooks.h (default_preferred_vector_alignment): Declare.
14824 * targhooks.c (default_preferred_vector_alignment): New function.
14825 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
14827 (DR_TARGET_ALIGNMENT): New macro.
14828 (aligned_access_p): Update commentary.
14829 (vect_known_alignment_in_bytes): New function.
14830 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
14832 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
14833 Calculate the misalignment based on the target alignment rather than
14835 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
14836 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
14837 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
14838 the target alignment, rather than masking the element misalignment
14839 with the number of elements in a vector. Also use the target
14840 alignment when calculating the maximum number of peels.
14841 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
14842 instead of TYPE_ALIGN_UNIT.
14843 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
14844 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
14845 (vect_create_addr_base_for_vector_ref): Update call accordingly.
14846 (vect_create_data_ref_ptr): Likewise.
14847 (vect_setup_realignment): Realign by ANDing with
14848 -DR_TARGET_MISALIGNMENT.
14849 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14850 the number of peels based on DR_TARGET_ALIGNMENT.
14851 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14852 with the guaranteed alignment boundary when deciding whether
14854 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14855 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14856 (ensure_base_align): Remove stmt_info parameter. Get the
14857 target base alignment from DR_TARGET_ALIGNMENT.
14858 (vectorizable_store): Update call accordingly. Interpret
14859 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14861 (vectorizable_load): Likewise.
14863 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14864 Alan Hayward <alan.hayward@arm.com>
14865 David Sherwood <david.sherwood@arm.com>
14867 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14868 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14869 (vect_enhance_data_refs_alignment): Likewise.
14871 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14873 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14874 error. Only quit immediately if parsing is complete.
14875 (BEGIN): Initialize fatal_err and parse_done.
14876 (begin fpu, end fpu): Check number of arguments.
14877 (begin arch, end arch): Likewise.
14878 (begin cpu, end cpu): Likewise.
14879 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14880 (optalias): Likewise.
14882 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14884 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14885 * config/arm/arm-isa.h: Delete. Move definitions to ...
14886 * arm-cpus.in: ... here. Use new feature and fgroup values.
14887 * config/arm/arm.c (arm_option_override): Use lower case for feature
14889 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14890 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14891 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14893 (print_isa_bits_for): New function.
14894 (gen_isa): New function.
14895 (gen_comm_data): Use print_isa_bits_for.
14896 (define feature): New keyword.
14897 (define fgroup): New keyword.
14898 * config/arm/t-arm (TM_H): Remove.
14899 (GTM_H): Add arm-isa.h.
14900 (arm-isa.h): Add rule to generate file.
14901 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14902 case for feature bit names.
14904 2017-09-22 Richard Biener <rguenther@suse.de>
14906 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14908 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14909 print of no dependency loops ...
14910 * graphite.c (graphite_transform_loops): ... here.
14911 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14913 (same_close_phi_node, remove_duplicate_close_phi,
14914 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14915 (canonicalize_loop_closed_ssa): ... here and simplify.
14916 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14917 (optimize_isl): Use dump_printf_loc to tell when we stopped
14918 optimizing because of an ISL timeout.
14920 2017-09-22 Richard Biener <rguenther@suse.de>
14922 PR tree-optimization/82291
14923 * tree-if-conv.c (predicate_mem_writes): Make sure to
14924 remove writes in blocks predicated with false.
14926 2017-09-22 Richard Biener <rguenther@suse.de>
14928 * sese.c: Include cfganal.h.
14929 (if_region_set_false_region): Remove.
14930 (create_if_region_on_edge): Likewise.
14931 (move_sese_in_condition): Re-implement without destroying
14934 2017-09-22 Richard Biener <rguenther@suse.de>
14936 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14937 Verify both BBs contain loop PHI nodes before dispatching to
14938 copy_loop_phi_args.
14939 (graphite_regenerate_ast_isl): Do not recompute dominators,
14940 do not verify three times. Restructure for clarity.
14941 * graphite-scop-detection.c (same_close_phi_node,
14942 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14943 defined_in_loop_p, canonicalize_loop_closed_ssa,
14944 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14945 checking and SSA rewrite, move to ...
14946 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14947 (graphite_initialize): Do not pass in ctx, do not reset the
14948 SCEV cache, compute only dominators.
14949 (graphite_transform_loops): Allocate ISL ctx after
14950 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14951 Maintain post-dominators only around build_scops.
14952 * sese.c (if_region_set_false_region): Make static. Free
14953 and recompute dominators.
14954 (move_sese_in_condition): Assert we don't get called with
14955 post-dominators computed.
14956 * sese.h (if_region_set_false_region): Remove.
14958 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14960 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14961 mode attribute for TARGET_AVX512VL.
14963 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14965 * config/i386/i386.opt (mprefer-avx256): New option.
14966 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14968 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14969 for TARGET_PREFER_AVX256.
14970 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14972 2017-09-21 Jeff Law <law@redhat.com>
14974 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14975 Fix dump output if the only stack space is for pushed registers.
14977 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14979 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14982 2017-09-21 Martin Sebor <msebor@redhat.com>
14985 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14986 code (in C++) or code that triggers warnings.
14988 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14990 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14992 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14994 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14995 * sched-rgn.c: Ditto.
14996 * sel-sched-ir.c: Ditto.
14998 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
15000 * toplev.h (set_random_seed): Adjust return type.
15001 * toplev.c (init_local_tick): Move eager initialization of random_seed
15002 to get_random_seed. Adjust comment.
15003 (init_random_seed): Inline to get_random_seed, delete.
15004 (get_random_seed): Initialize random_seed lazily.
15005 (set_random_seed): Do not return previous value.
15006 (print_switch_value): Do not call get_random_seed.
15008 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
15010 * cgraph.c (delete_function_version): New, broken out from...
15011 (cgraph_node::delete_function_version): ...here. Rename to
15012 cgraph_node::delete_function_version_by_decl. Update all uses.
15013 (cgraph_node::remove): Call delete_function_version.
15015 2017-09-21 Jakub Jelinek <jakub@redhat.com>
15018 * tree-inline.c (expand_call_inline): Emit clobber stmts for
15019 VAR_DECLs to which addressable non-volatile parameters are mapped
15020 and for id->retvar after the return value assignment. Clear
15021 id->retval and id->retbnd after inlining.
15023 2017-09-21 Richard Biener <rguenther@suse.de>
15025 PR tree-optimization/82276
15026 PR tree-optimization/82244
15027 * tree-vrp.c (build_assert_expr_for): Set
15028 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
15030 (remove_range_assertions): Revert earlier change.
15032 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
15035 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
15037 2017-09-21 Richard Biener <rguenther@suse.de>
15039 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
15040 Restore valid IL after code generation errors.
15041 * graphite.c (graphite_transform_loops): Diagnose code
15042 generation issues as MSG_MISSED_OPTIMIZATION and continue
15043 with processing SCOPs.
15045 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
15046 Alan Hayward <alan.hayward@arm.com>
15047 David Sherwood <david.sherwood@arm.com>
15049 * calls.c (compute_argument_addresses): Use simplify_gen_binary
15050 rather than choosing between plus_constant and gen_rtx_<CODE>.
15051 * expr.c (emit_push_insn): Likewise.
15052 (expand_expr_real_2): Likewise.
15054 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
15055 Alan Hayward <alan.hayward@arm.com>
15056 David Sherwood <david.sherwood@arm.com>
15058 * loop-unroll.c (split_iv): Call copy_rtx on the step.
15060 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
15061 Alan Hayward <alan.hayward@arm.com>
15062 David Sherwood <david.sherwood@arm.com>
15064 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
15065 calling tree_to_uhwi.
15067 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
15068 Alan Hayward <alan.hayward@arm.com>
15069 David Sherwood <david.sherwood@arm.com>
15071 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
15072 INTEGER_CST rather than a negative test for ADDR_EXPR.
15074 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
15075 Alan Hayward <alan.hayward@arm.com>
15076 David Sherwood <david.sherwood@arm.com>
15078 * tree-vrp.c (extract_range_from_binary_expr_1): Check
15079 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
15081 2017-09-21 Richard Biener <rguenther@suse.de>
15083 PR tree-optimization/71351
15084 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
15085 graphite_create_new_loop_guard): Remove, fold remaining parts
15087 (translate_isl_ast_node_for): ... here and simplify.
15089 2017-09-21 Jakub Jelinek <jakub@redhat.com>
15092 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
15093 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
15094 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
15095 alternative always use QI mode, for -Os imov (=R,R) alternative
15096 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
15099 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15100 Jeff Law <law@redhat.com>
15102 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
15103 (allocate_stack_space): New function, partially extracted from
15104 s390_emit_prologue.
15105 (s390_emit_prologue): Track offset to most recent stack probe.
15106 Code to allocate space moved into allocate_stack_space.
15107 Dump actions when no stack is allocated.
15108 (s390_prologue_plus_offset): New function.
15109 (s390_emit_stack_probe): Likewise.
15111 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
15113 * common.opt (Wa, Wl, Wp, g, gz=): Add
15115 (gno-column-info): Remove.
15116 (gcolumn-info): Drop RejectNegative.
15117 (gno-): New prefix.
15118 (gno-record-gcc-switches): Remove.
15119 (grecord-gcc-switches): Drop RejectNegative.
15120 (gno-split-dwarf): Remove.
15121 (gsplit-dwarf): Drop RejectNegative.
15122 (gno-strict-dwarf): Remove.
15123 (gstrict-dwarf): Drop RejectNegative.
15124 * config/darwin.opt (gfull, gused): Add RejectNegative.
15125 * dwarf2out.c (gen_producer_string): Drop
15126 gno-record-gcc-switches handler.
15127 * optc-gen.awk: Add g to prefixes with negative forms.
15128 * opts-common.c (remapping_prefix_p): New.
15129 (find_opt): Check it.
15130 (generate_canonical_option): Test g prefix.
15131 (option_map): Add -gno- mapping.
15132 (add_misspelling_candidates): Check remapping_prefix_p.
15134 2017-09-20 Jeff Law <law@redhat.com>
15136 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
15137 thinko in stack clash protection support.
15139 * explow.c (compute_stack_clash_protection_loop_data): Use
15140 CONST_INT_P instead of explicit test. Verify object is a
15141 CONST_INT_P before looking at INTVAL.
15142 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
15143 instead of explicit test.
15145 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
15148 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
15149 address instead of to r1 and r11.
15151 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
15153 * config.gcc: Support "knm".
15154 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
15155 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
15157 * config/i386/i386.c (m_KNM): Define.
15158 (processor_target_table): Add "knm".
15160 (ix86_option_override_internal): Add "knm".
15161 (ix86_issue_rate): Add PROCESSOR_KNM.
15162 (ix86_adjust_cost): Ditto.
15163 (ia32_multipass_dfa_lookahead): Ditto.
15164 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
15165 (fold_builtin_cpu): Add M_INTEL_KNM.
15166 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
15167 (processor_type): Add PROCESSOR_KNM.
15168 * config/i386/x86-tune.def: Add m_KNM.
15169 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
15171 2017-09-20 Richard Biener <rguenther@suse.de>
15173 PR tree-optimization/80213
15174 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
15175 are allowed in empty BBs as well.
15176 (canonicalize_loop_closed_ssa): Also look for other complex
15178 (scop_detection::get_sese): Include the loop-closed PHI block
15180 (scop_detection::merge_sese): Remove code adding extra blocks.
15181 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
15182 (build_scops): Assert the final returned scop is invalid.
15184 2017-09-20 Richard Biener <rguenther@suse.de>
15186 PR tree-optimization/82264
15187 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
15189 (vn_phi_lookup): Likewise.
15190 (vn_phi_insert): Likewise.
15192 2017-09-20 Jakub Jelinek <jakub@redhat.com>
15194 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
15195 that fits into uhwi or shwi, add DW_AT_const_value regardless
15196 of early_dwarf without going through RTL, using add_AT_unsigned
15199 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
15200 (DEBUG_LTO_ABBREV_SECTION): Likewise.
15201 (DEBUG_LTO_MACINFO_SECTION): Likewise.
15202 (DEBUG_MACRO_SECTION): Likewise.
15203 (DEBUG_LTO_MACRO_SECTION): Likewise.
15204 (DEBUG_STR_DWO_SECTION): Likewise.
15205 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
15206 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
15207 (DEBUG_LTO_DWO_LINE_SECTION): Define.
15208 (DEBUG_LTO_LINE_STR_SECTION): Define.
15209 (init_sections_and_labels): Initialize debug_line_str_section
15210 variable. Initialize debug_loc_section for -gdwarf-5 to
15211 DEBUG_LOCLISTS_SECTION. Formatting fixes.
15213 2017-09-20 Richard Biener <rguenther@suse.de>
15215 * graphite-sese-to-poly.c (extract_affine): Properly handle
15216 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
15218 2017-09-20 Richard Biener <rguenther@suse.de>
15220 PR tree-optimization/81373
15221 * graphite-scop-detection.c (build_cross_bb_scalars_def):
15222 Force SESE live-out defs to be handled even if they are
15225 2017-09-19 Jeff Law <law@redhat.com>
15227 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
15228 nothing for stack adjustments with REG_STACK_CHECK.
15229 * sched-deps.c (parse_add_or_inc): Reject insns with
15230 REG_STACK_CHECK from dependency breaking.
15231 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
15232 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
15233 * reg-notes.def (STACK_CHECK): New note.
15235 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
15236 (ix86_expand_prologue): Dump stack clash info as needed.
15237 Call ix86_adjust_stack_and_probe_stack_clash as needed.
15239 * function.c (dump_stack_clash_frame_info): New function.
15240 * function.h (dump_stack_clash_frame_info): Prototype.
15241 (enum stack_clash_probes): New enum.
15243 * config/alpha/alpha.c (alpha_expand_prologue): Also check
15244 flag_stack_clash_protection.
15245 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
15246 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
15247 (arm_frame_pointer_required): Likewise.
15248 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
15249 (ia64_expand_prologue): Likewise.
15250 * config/mips/mips.c (mips_expand_prologue): Likewise.
15251 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
15252 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15253 (sparc_flat_expand_prologue): Likewise.
15254 * config/spu/spu.c (spu_expand_prologue): Likewise.
15256 * explow.c: Include "params.h".
15257 (anti_adjust_stack_and_probe_stack_clash): New function.
15258 (get_stack_check_protect): Likewise.
15259 (compute_stack_clash_protection_loop_data): Likewise.
15260 (emit_stack_clash_protection_loop_start): Likewise.
15261 (emit_stack_clash_protection_loop_end): Likewise.
15262 (allocate_dynamic_stack_space): Use get_stack_check_protect.
15263 Use anti_adjust_stack_and_probe_stack_clash.
15264 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
15265 (emit_stack_clash_protection_loop_start): Likewise.
15266 (emit_stack_clash_protection_loop_end): Likewise.
15267 * rtl.h (get_stack_check_protect): Prototype.
15268 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
15269 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
15270 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
15272 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
15274 * doc/tm.texi: Rebuilt.
15275 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
15276 get_stack_check_protect.
15277 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
15278 * config/arm/arm.c (arm_expand_prologue): Likewise.
15279 (arm_frame_pointer_required): Likewise.
15280 * config/i386/i386.c (ix86_expand_prologue): Likewise.
15281 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
15282 * config/mips/mips.c (mips_expand_prologue): Likewise.
15283 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
15284 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
15285 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15286 (sparc_flat_expand_prologue): Likewise.
15288 * common.opt (-fstack-clash-protection): New option.
15289 * flag-types.h (enum stack_check_type): Note difference between
15290 -fstack-check= and -fstack-clash-protection.
15291 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
15292 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
15293 * toplev.c (process_options): Issue warnings/errors for cases
15294 not handled with -fstack-clash-protection.
15295 * doc/invoke.texi (-fstack-clash-protection): Document new option.
15296 (-fstack-check): Note additional problem with -fstack-check=generic.
15297 Note that -fstack-check is primarily for Ada and refer users
15298 to -fstack-clash-protection for stack-clash-protection.
15299 Document new params for stack clash protection.
15301 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
15303 * config/i386/i386.c (ix86_split_long_move): Do not handle
15304 address used for LEA in a special way.
15306 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
15308 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
15310 2017-09-19 Martin Sebor <msebor@redhat.com>
15313 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
15314 of incompatible types.
15316 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
15318 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15319 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
15320 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15321 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
15323 2017-09-19 Richard Biener <rguenther@suse.de>
15325 PR tree-optimization/82244
15326 * tree-vrp.c (remove_range_assertions): Do not propagate
15327 a constant to abnormals but replace the assert with a copy.
15329 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
15331 PR rtl-optimization/57878
15332 PR rtl-optimization/68988
15333 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
15334 avoidance test involving non_reload_pseudos. Move frequency test
15335 below the general fragmentation avoidance test.
15337 2017-09-19 Richard Biener <rguenther@suse.de>
15339 PR tree-optimization/69728
15340 * graphite-sese-to-poly.c (schedule_error): New global.
15341 (add_loop_schedule): Handle empty domain by failing the
15343 (build_original_schedule): Handle schedule_error.
15345 2017-09-19 Richard Biener <rguenther@suse.de>
15347 * graphite-scop-detection.c (scop_detection::can_represent_loop):
15348 Do not iterate to sibling loops but only to siblings of inner
15351 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
15354 * config/m68k/m68k.md (moveq feeding equality comparison): Check
15355 that the registers are different.
15357 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
15359 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
15360 to processor_model and "amdfam17h" to arch_names_table.
15361 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
15363 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15366 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
15368 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15369 Alan Hayward <alan.hayward@arm.com>
15370 David Sherwood <david.sherwood@arm.com>
15372 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
15374 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
15376 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
15377 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
15378 vect_schedule_slp_instance.
15379 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
15380 Update call to vect_slp_analyze_node_operations. Simplify return
15382 (vect_slp_analyze_bb_1): Update call accordingly.
15383 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
15384 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
15385 (vect_schedule_slp): Update call accordingly.
15387 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15388 Alan Hayward <alan.hayward@arm.com>
15389 David Sherwood <david.sherwood@arm.com>
15391 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
15392 with types that aren't in fact scalar.
15394 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15396 * tree-vect-slp.c (vect_record_max_nunits): New function,
15398 (vect_build_slp_tree_1): ...here.
15399 (vect_build_slp_tree_2): Call it for phis too.
15401 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15403 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
15404 to vect_get_vec_def_for_operand when getting the mask operand.
15406 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15407 Alan Hayward <alan.hayward@arm.com>
15408 David Sherwood <david.sherwood@arm.com>
15410 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
15413 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15414 Alan Hayward <alan.hayward@arm.com>
15415 David Sherwood <david.sherwood@arm.com>
15417 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
15418 calculation for vector booleans.
15420 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15421 Alan Hayward <alan.hayward@arm.com>
15422 David Sherwood <david.sherwood@arm.com>
15424 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
15426 (vect_transform_stmt): ...here.
15427 (vect_analyze_stmt): Use it instead of calling
15428 vectorizable_live_operation directly.
15430 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
15432 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
15433 non-SIMT targets in acc vector loops.
15435 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
15437 * configure.ac: Add arc and check if assembler supports gdwarf2.
15438 * configure: Regenerate.
15440 2017-09-18 Richard Biener <rguenther@suse.de>
15442 PR tree-optimization/82220
15443 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
15444 epilogue niters from the min_profitable_iters compute.
15446 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15449 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
15450 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
15451 (ix86_init_pic_reg): Revert 2017-09-01 changes.
15453 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
15456 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
15457 switching to a new text section.
15459 2017-09-18 Richard Biener <rguenther@suse.de>
15461 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
15463 (build_alias_set): Reject aliases with no access function.
15465 2017-09-18 Richard Biener <rguenther@suse.de>
15467 PR tree-optimization/79622
15468 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
15470 (build_cross_bb_scalars_use): Likewise.
15472 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
15474 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
15476 2017-09-18 Alan Modra <amodra@gmail.com>
15479 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
15480 stack_pointer_rtx for count 0. Update comments. Break up
15481 large rtl expression.
15483 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
15485 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
15486 Increase to 20 bytes.
15487 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
15488 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
15489 or avx version of the stub.
15491 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
15494 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
15495 compute the minimum stack alignment. Also update preferred stack
15496 boundary for leaf functions.
15498 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
15500 PR tree-optimization/82228
15501 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
15504 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15506 * common/config/nds32/nds32-common.c
15507 (nds32_option_optimization_table): Refine formatting.
15508 (nds32_option_optimization_table): Use -fsched-pressure and
15509 -fomit-frame-pointer for specific optimization level.
15511 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15513 * config/nds32/nds32.c: Refine formatting and comments.
15514 * config/nds32/nds32.h: Likewise.
15515 * config/nds32/nds32.md: Likewise.
15516 * config/nds32/nds32-cost.c: Likewise.
15517 * config/nds32/nds32-isr.c: Likewise.
15518 * config/nds32/nds32-md-auxiliary.c: Likewise.
15519 * config/nds32/nds32-multiple.md: Likewise.
15520 * config/nds32/nds32-predicates.c: Likewise.
15522 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
15523 Jakub Jelinek <jakub@redhat.com>
15525 Add support for -std=c++2a.
15526 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
15528 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
15530 2017-09-15 Steve Ellcey <sellcey@cavium.com>
15533 * doc/extend.texi (Common Function Attributes): Add
15534 references to ARM, AArch64, and S/390 specific attributes.
15535 (Function Specific Option Pragmas): Add AArch64 and S/390
15536 to list of back ends that support the target pragma.
15538 2017-09-15 Nathan Sidwell <nathan@acm.org>
15540 * doc/standards.texi: Fix C++17 description. Update URLs for
15543 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
15545 * common.opt (Wcast-align=strict): New warning option.
15546 * doc/invoke.texi: Document -Wcast-align=strict.
15548 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
15550 * cgraph.h (cgraph_thunk_info): Add comments.
15551 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
15552 assert for VIRTUAL_* arguments stricter.
15554 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
15556 PR tree-optimization/71026
15557 * match.pd: Move RDIV patterns from fold-const.c
15558 * fold-const.c (distribute_real_division): Removed.
15559 (fold_binary_loc): Remove calls to distribute_real_divison.
15561 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15563 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
15564 c++1z and gnu++1z as deprecated. Change other references to
15565 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
15566 Change -Wc++1z-compat to -Wc++17-compat.
15567 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
15568 * dwarf2out.c (highest_c_language): Handle C++17.
15569 (gen_compile_unit_die): Likewise.
15571 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15573 PR rtl-optimization/82192
15574 * combine.c (make_extraction): Don't look through non-paradoxical
15575 SUBREGs or TRUNCATE if pos + len is or might be bigger than
15578 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15579 Alan Hayward <alan.hayward@arm.com>
15580 David Sherwood <david.sherwood@arm.com>
15582 * target.def (function_arg_offset): New hook.
15583 * targhooks.h (default_function_arg_offset): Declare.
15584 * targhooks.c (default_function_arg_offset): New function.
15585 * function.c (locate_and_pad_parm): Use
15586 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
15587 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
15588 (TARGET_FUNCTION_ARG_OFFSET): ...this.
15589 * doc/tm.texi: Regenerate.
15590 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
15591 * config/spu/spu.c (spu_function_arg_offset): New function.
15592 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
15593 * system.h (FUNCTION_ARG_OFFSET): Poison.
15595 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15596 Alan Hayard <alan.hayward@arm.com>
15597 David Sherwood <david.sherwood@arm.com>
15599 * target.def (truly_noop_truncation): New hook.
15600 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
15601 than TRULY_NOOP_TRUNCATION.
15602 * hooks.h (hook_bool_uint_uint_true): Declare.
15603 * hooks.c (hook_bool_uint_uint_true): New function.
15604 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
15605 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
15606 * doc/tm.texi: Regenerate.
15607 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
15608 rather than TRULY_NOOP_TRUNCATION in comments.
15609 (simplify_comparison): Likewise.
15610 (record_truncated_value): Likewise.
15611 * expmed.c (extract_bit_field_1): Likewise.
15612 (extract_split_bit_field): Likewise.
15613 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
15614 instead of TRULY_NOOP_TRUNCATION.
15615 * function.c (assign_parm_setup_block): Likewise.
15616 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
15617 * rtlhooks.c: Include target.h.
15618 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
15619 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
15620 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
15621 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
15622 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
15623 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
15624 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
15625 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
15626 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
15627 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
15628 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
15629 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
15630 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
15631 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
15632 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
15633 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
15634 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
15635 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
15636 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
15637 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
15638 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
15639 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
15640 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
15641 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
15642 * config/mips/mips.c (mips_truly_noop_truncation): New function.
15643 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15644 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15645 rather than TRULY_NOOP_TRUNCATION in comments.
15646 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
15647 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
15648 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
15649 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
15650 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
15651 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
15652 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
15653 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
15654 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
15655 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
15656 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
15657 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15658 rather than TRULY_NOOP_TRUNCATION in comments.
15659 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
15660 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
15661 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
15662 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
15663 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
15664 TRULY_NOOP_TRUNCATION condition.
15665 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
15666 (TRULY_NOOP_TRUNCATION): Delete.
15667 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
15668 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
15669 * config/spu/spu.c (spu_truly_noop_truncation): New function.
15670 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15671 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
15672 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
15673 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
15674 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15675 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15676 rather than TRULY_NOOP_TRUNCATION in comments.
15677 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
15678 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
15679 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
15680 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
15681 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
15682 * system.h (TRULY_NOOP_TRUNCATION): Poison.
15684 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
15687 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
15688 (*cmp_ior): Likewise.
15689 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
15690 (*ior_scc_scc_cmp): Likewise.
15691 (*and_scc_scc): Likewise.
15692 (*and_scc_scc_cmp): Likewise.
15694 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15695 Alan Hayard <alan.hayward@arm.com>
15696 David Sherwood <david.sherwood@arm.com>
15698 * target.def (can_change_mode_class): New hook.
15699 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
15700 (hard_regno_nregs): Likewise.
15701 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
15702 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
15703 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
15704 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
15705 (LOAD_EXTEND_OP): Update accordingly.
15706 * doc/tm.texi: Regenerate.
15707 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
15708 CANNOT_CHANGE_MODE_CLASS.
15709 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
15710 (REG_CAN_CHANGE_MODE_P): ...this new macro.
15711 * combine.c (simplify_set): Update accordingly.
15712 * emit-rtl.c (validate_subreg): Likewise.
15713 * recog.c (general_operand): Likewise.
15714 * regcprop.c (mode_change_ok): Likewise.
15715 * reload1.c (choose_reload_regs): Likewise.
15716 (inherit_piecemeal_p): Likewise.
15717 * rtlanal.c (simplify_subreg_regno): Likewise.
15718 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
15719 instead of CANNOT_CHANGE_MODE_CLASS.
15720 (reload_cse_simplify_operands): Likewise.
15721 * reload.c (push_reload): Use targetm.can_change_mode_class
15722 instead of CANNOT_CHANGE_MODE_CLASS.
15723 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
15724 REG_CANNOT_CHANGE_MODE_P.
15725 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15726 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
15727 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15728 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15729 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15730 (arm_can_change_mode_class): New function.
15731 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15732 than CANNOT_CHANGE_MODE_CLASS in comments.
15733 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15734 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
15735 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
15736 (ix86_can_change_mode_class): ...this new function, inverting the
15737 sense of the return value.
15738 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15739 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15740 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15741 (ia64_can_change_mode_class): New function.
15742 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15743 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
15744 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
15745 (m32c_can_change_mode_class): ...this new function, inverting the
15746 sense of the return value.
15747 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15748 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15749 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
15750 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
15751 (mips_can_change_mode_class): ...this new function, inverting the
15752 sense of the return value.
15753 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15754 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15755 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15756 (msp430_can_change_mode_class): New function.
15757 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15758 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
15759 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15760 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15761 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15762 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
15763 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15764 (pa_cannot_change_mode_class): Replace with...
15765 (pa_can_change_mode_class): ...this new function, inverting the
15766 sense of the return value.
15767 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15768 than CANNOT_CHANGE_MODE_CLASS in comments.
15769 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15770 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
15771 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15772 (pdp11_cannot_change_mode_class): Replace with...
15773 (pdp11_can_change_mode_class): ...this new function, inverting the
15774 sense of the return value.
15775 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15776 * config/powerpcspe/powerpcspe-protos.h
15777 (rs6000_cannot_change_mode_class_ptr): Delete.
15778 * config/powerpcspe/powerpcspe.c
15779 (rs6000_cannot_change_mode_class_ptr): Delete.
15780 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15781 (rs6000_option_override_internal): Assign to
15782 targetm.can_change_mode_class instead of
15783 rs6000_cannot_change_mode_class_ptr.
15784 (rs6000_cannot_change_mode_class): Replace with...
15785 (rs6000_can_change_mode_class): ...this new function, inverting the
15786 sense of the return value.
15787 (rs6000_debug_cannot_change_mode_class): Replace with...
15788 (rs6000_debug_can_change_mode_class): ...this new function.
15789 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15790 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
15791 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15792 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15793 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
15795 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
15796 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15797 (rs6000_option_override_internal): Assign to
15798 targetm.can_change_mode_class instead of
15799 rs6000_cannot_change_mode_class_ptr.
15800 (rs6000_cannot_change_mode_class): Replace with...
15801 (rs6000_can_change_mode_class): ...this new function, inverting the
15802 sense of the return value.
15803 (rs6000_debug_cannot_change_mode_class): Replace with...
15804 (rs6000_debug_can_change_mode_class): ...this new function.
15805 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15806 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
15807 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
15808 (s390_can_change_mode_class): ...this new function, inverting the
15809 sense of the return value.
15810 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15811 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15812 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
15813 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15814 (sh_cannot_change_mode_class): Replace with...
15815 (sh_can_change_mode_class): ...this new function, inverting the
15816 sense of the return value.
15817 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15818 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15819 (sparc_can_change_mode_class): New function.
15820 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15821 * config/spu/spu.c (spu_can_change_mode_class): New function.
15822 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15823 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15824 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15825 (visium_can_change_mode_class): New function.
15826 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
15828 2017-09-15 Richard Biener <rguenther@suse.de>
15830 PR tree-optimization/82217
15831 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
15832 but not undefined case.
15834 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15837 * postreload.c (reload_cse_simplify_operands): Skip
15838 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
15840 2017-09-15 Richard Biener <rguenther@suse.de>
15842 PR tree-optimization/68823
15843 * graphite-scop-detection.c (build_alias_set): If we have a
15844 possible dependence check whether we can handle them by just
15845 looking at the DRs DR_ACCESS_FNs.
15846 (build_scops): If build_alias_set fails, fail the SCOP.
15848 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15850 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15851 to support float128 built-in functions that require the ISA 3.0
15853 (BU_FLOAT128_3_HW): Likewise.
15854 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15855 built-in functions.
15856 (FMAF128): Likewise.
15858 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15859 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15860 floating point instructions.
15861 (rs6000_invalid_builtin): Likewise.
15862 (rs6000_builtin_mask_names): Likewise.
15863 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15864 (RS6000_BTM_FLOAT128_HW): Likewise.
15865 (RS6000_BTM_COMMON): Likewise.
15866 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15868 * doc/extend.texi (RS/6000 built-in functions): Document the
15869 IEEE 128-bit floating point square root and fused multiply-add
15870 built-in functions.
15872 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15874 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15875 reg (r2) isn't in the set of registers defined in the prologue.
15877 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15878 Alan Hayward <alan.hayward@arm.com>
15879 David Sherwood <david.sherwood@arm.com>
15881 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15882 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15883 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15884 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15885 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15887 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15888 max_vectorization_factor.
15889 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15891 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15892 Alan Hayward <alan.hayward@arm.com>
15893 David Sherwood <david.sherwood@arm.com>
15895 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15896 (vect_worthwhile_without_simd_p): Declare.
15897 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15898 (vectorizable_reduction): Use it.
15899 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15900 (vectorizable_operation): Likewise.
15902 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15903 Alan Hayward <alan.hayward@arm.com>
15904 David Sherwood <david.sherwood@arm.com>
15906 * tree-vectorizer.h (vect_get_num_copies): New function.
15907 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15908 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15909 (vectorizable_induction): Likewise.
15910 (vectorizable_live_operation): Likewise.
15911 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15912 (vectorizable_bswap): Likewise.
15913 (vectorizable_call): Likewise.
15914 (vectorizable_conversion): Likewise.
15915 (vectorizable_assignment): Likewise.
15916 (vectorizable_shift): Likewise.
15917 (vectorizable_operation): Likewise.
15918 (vectorizable_store): Likewise.
15919 (vectorizable_load): Likewise.
15920 (vectorizable_condition): Likewise.
15921 (vectorizable_comparison): Likewise.
15922 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15924 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15925 Alan Hayward <alan.hayward@arm.com>
15926 David Sherwood <david.sherwood@arm.com>
15928 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15929 of vect_init_vector.
15931 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15932 Alan Hayward <alan.hayward@arm.com>
15933 David Sherwood <david.sherwood@arm.com>
15935 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15936 an inline wrapper that provides a location.
15937 (gimple_build_vector): Likewise.
15938 * gimple-fold.c (gimple_build_vector_from_val): New function.
15939 (gimple_build_vector): Likewise.
15940 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15941 functions to build the initial value. Always return a gimple value.
15942 (get_initial_defs_for_reduction): Likewise. Only compute
15944 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15945 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15946 (vectorizable_induction): Use gimple_build_vector rather than
15949 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15950 Alan Hayward <alan.hayward@arm.com>
15951 David Sherwood <david.sherwood@arm.com>
15953 * target.h (vec_perm_indices): New typedef.
15954 (auto_vec_perm_indices): Likewise.
15955 * optabs-query.h: Include target.h
15956 (can_vec_perm_p): Take a vec_perm_indices *.
15957 * optabs-query.c (can_vec_perm_p): Likewise.
15958 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15959 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15960 * tree-vect-generic.c (lower_vec_perm): Likewise.
15961 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15962 (vect_grouped_load_supported): Likewise.
15963 (vect_shift_permute_load_chain): Likewise.
15964 (vect_permute_store_chain): Use auto_vec_perm_indices.
15965 (vect_permute_load_chain): Likewise.
15966 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15967 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15968 Update uses of can_vec_perm_p.
15969 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15970 mode with a number of elements. Take a vec_perm_indices *.
15971 (vect_create_epilog_for_reduction): Update accordingly.
15972 Use auto_vec_perm_indices.
15973 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15974 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15975 (vect_transform_slp_perm_load): Likewise.
15976 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15977 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15978 (vect_gen_perm_mask_checked): Likewise.
15979 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15980 (vect_gen_perm_mask_checked): Likewise.
15981 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15982 (vectorizable_store): Likewise.
15983 (vectorizable_load): Likewise.
15984 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15985 (vectorizable_bswap): Likewise.
15987 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15988 Alan Hayward <alan.hayward@arm.com>
15989 David Sherwood <david.sherwood@arm.com>
15991 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15992 * tree.c (build_vector): Likewise.
15993 (build_vector_from_ctor): Update accordingly.
15994 (build_vector_from_val): Likewise.
15995 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15996 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15997 * tree-vect-generic.c (add_rshift): Likewise.
15998 (expand_vector_divmod): Likewise.
15999 (optimize_vector_constructor): Likewise.
16000 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
16001 (vect_transform_slp_perm_load): Likewise.
16002 (vect_schedule_slp_instance): Likewise.
16003 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
16004 (vectorizable_call): Likewise.
16005 (vect_gen_perm_mask_any): Likewise. Add elements in order.
16006 * expmed.c (make_tree): Likewise.
16007 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
16008 a vector passed to build_vector.
16009 (fold_convert_const): Likewise.
16010 (exact_inverse): Likewise.
16011 (fold_ternary_loc): Likewise.
16012 (fold_relational_const): Likewise.
16013 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
16014 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
16015 (const_unop): Likewise. Store the reduction accumulator in a
16016 variable rather than an array.
16017 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
16018 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
16019 the new vector, rather than constructing it after the input arrays.
16020 (native_interpret_vector): Use auto_vec<tree> when building
16021 a vector passed to build_vector. Add elements in order.
16022 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
16023 auto_vec<tree> when building a vector passed to build_vector.
16024 (vect_create_epilog_for_reduction): Likewise.
16025 (vectorizable_induction): Likewise.
16026 (get_initial_def_for_reduction): Likewise. Fix indentation of
16028 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
16030 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
16031 passed to build_vector.
16033 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
16034 Alan Hayward <alan.hayward@arm.com>
16035 David Sherwood <david.sherwood@arm.com>
16037 * tree-core.h (tree_base::u): Add an "nelts" field.
16038 (tree_vector): Use VECTOR_CST_NELTS as the length.
16039 * tree.c (tree_size): Likewise.
16040 (make_vector): Initialize VECTOR_CST_NELTS.
16041 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
16042 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
16043 TYPE_VECTOR_SUBPARTS.
16044 * expr.c (const_vector_mask_from_tree): Consistently use "units"
16045 as the number of units, setting it from VECTOR_CST_NELTS.
16046 (const_vector_from_tree): Likewise.
16047 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
16048 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
16049 (fold_negate_expr_1): Likewise.
16050 (fold_convert_const): Likewise.
16051 (const_binop): Likewise. Differentiate the number of output and
16053 (const_unop): Likewise.
16054 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
16055 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
16056 in cases that did the opposite.
16058 2017-09-14 Richard Biener <rguenther@suse.de>
16060 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
16063 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
16065 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
16067 2017-09-14 Jakub Jelinek <jakub@redhat.com>
16070 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
16071 if and where to split a bb, except for splitting before debug insn
16072 sequences followed by non-label real insn. Delete debug insns
16073 in between basic blocks.
16075 * combine.c (make_compound_operation_int): Formatting fixes.
16077 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
16078 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16079 * config/netbsd.h (LINK_EH_SPEC): Likewise.
16080 * config/sol2.h (LINK_EH_SPEC): Likewise.
16081 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16082 * config/s390/linux.h (LINK_SPEC): Likewise.
16083 * config/freebsd.h (LINK_EH_SPEC): Likewise.
16084 * config/openbsd.h (LINK_EH_SPEC): Likewise.
16085 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16086 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
16087 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
16088 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16089 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
16090 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
16092 2017-09-13 Jakub Jelinek <jakub@redhat.com>
16094 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
16096 (ENDFILE_LINUX_SPEC): Likewise.
16097 (LINK_EH_SPEC): Likewise.
16098 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
16099 (LINK_OS_LINUX_SPEC32): Likewise.
16100 (LINK_OS_LINUX_SPEC64): Likewise.
16101 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
16102 (LINK_OS_LINUX_SPEC): Likewise.
16104 2017-09-13 Martin Liska <mliska@suse.cz>
16106 PR middle-end/82154
16107 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
16108 CASE_HIGH is NULL_TREE.
16110 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
16111 Alan Hayward <alan.hayward@arm.com>
16112 David Sherwood <david.sherwood@arm.com>
16114 * target.def (secondary_memory_needed): New hook.
16115 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
16116 instead of SECONDARY_MEMORY_NEEDED.
16117 (secondary_memory_needed_mode): Likewise.
16118 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
16119 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
16120 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
16121 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
16122 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
16123 * doc/tm.texi: Regenerate.
16124 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
16125 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
16126 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16127 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
16128 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
16129 * config/i386/i386.c (inline_secondary_memory_needed): Put the
16130 mode argument first and change the reg_class arguments to reg_class_t.
16131 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
16132 Make static. Update the call to inline_secondary_memory_needed.
16133 (ix86_register_move_cost): Update the call to
16134 inline_secondary_memory_needed.
16135 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16136 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
16138 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
16139 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16141 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
16142 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
16143 * config/mips/mips.c (mips_secondary_memory_needed): Make static
16144 and match hook interface. Add comment from mips.h.
16145 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16146 * config/mmix/mmix.md (truncdfsf2): Refer to
16147 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16149 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
16150 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
16151 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16152 (pa_secondary_memory_needed): New function.
16153 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
16154 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
16155 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16156 (pdp11_secondary_memory_needed): Make static and match hook interface.
16157 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
16158 * config/powerpcspe/powerpcspe-protos.h
16159 (rs6000_secondary_memory_needed_ptr): Delete.
16160 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
16162 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16163 (rs6000_option_override_internal): Assign to
16164 targetm.secondary_memory_needed rather than
16165 rs6000_secondary_memory_needed_ptr.
16166 (rs6000_secondary_memory_needed): Match hook interface.
16167 (rs6000_debug_secondary_memory_needed): Likewise.
16168 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
16169 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
16170 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
16171 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16172 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
16173 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
16175 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
16176 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16177 (rs6000_option_override_internal): Assign to
16178 targetm.secondary_memory_needed rather than
16179 rs6000_secondary_memory_needed_ptr.
16180 (rs6000_secondary_memory_needed): Match hook interface.
16181 (rs6000_debug_secondary_memory_needed): Likewise.
16182 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
16183 * config/s390/s390.c (s390_secondary_memory_needed): New function.
16184 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16185 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
16186 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16187 (sparc_secondary_memory_needed): New function.
16188 * lra-constraints.c (check_and_process_move): Refer to
16189 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16191 (curr_insn_transform): Likewise.
16192 (process_alt_operands): Use targetm.secondary_memory_needed
16193 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16194 (check_secondary_memory_needed_p): Likewise.
16195 (choose_split_class): Likewise.
16196 * reload.c: Unconditionally include code that was previously
16197 conditional on SECONDARY_MEMORY_NEEDED.
16198 (push_secondary_reload): Use targetm.secondary_memory_needed
16199 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16200 (push_reload): Likewise.
16201 * reload1.c: Unconditionally include code that was previously
16202 conditional on SECONDARY_MEMORY_NEEDED.
16203 (choose_reload_regs): Use targetm.secondary_memory_needed
16204 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16205 (gen_reload): Likewise.
16206 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
16208 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
16209 Alan Hayward <alan.hayward@arm.com>
16210 David Sherwood <david.sherwood@arm.com>
16212 * target.def (secondary_memory_needed_mode): New hook:
16213 * targhooks.c (default_secondary_memory_needed_mode): Declare.
16214 * targhooks.h (default_secondary_memory_needed_mode): New function.
16215 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
16216 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
16217 * doc/tm.texi: Regenerate.
16218 * lra-constraints.c (check_and_process_move): Use
16219 targetm.secondary_memory_needed_mode instead of
16220 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
16221 (curr_insn_transform): Likewise.
16222 * reload.c (get_secondary_mem): Likewise.
16223 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16224 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
16226 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16227 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16228 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
16229 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16230 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
16232 * config/powerpcspe/powerpcspe-protos.h
16233 (rs6000_secondary_memory_needed_mode): Delete.
16234 * config/powerpcspe/powerpcspe.c
16235 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16236 (rs6000_secondary_memory_needed_mode): Make static.
16237 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16238 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
16240 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16242 (rs6000_secondary_memory_needed_mode): Make static.
16243 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16244 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
16245 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16246 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16247 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16249 (sparc_secondary_memory_needed_mode): New function.
16250 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
16252 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
16254 * config/aarch64/constraints.md (Umq): New constraint.
16255 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
16257 (mov<mode>): Update condition.
16259 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16261 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
16262 when bitposition is the same.
16264 2017-09-13 Richard Biener <rguenther@suse.de>
16266 * dwarf2out.c (output_die_symbol): Remove.
16267 (output_die): Do not output a DIEs symbol.
16269 2017-09-13 Richard Biener <rguenther@suse.de>
16271 PR middle-end/82128
16272 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
16273 default-def to avoid breaking iterator update with the weird
16274 interaction with cgraph_update_edges_for_call_stmt_node.
16276 2017-09-13 Richard Biener <rguenther@suse.de>
16278 * tree-cfg.c (verify_gimple_assign_binary): Add verification
16279 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
16280 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
16281 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
16283 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
16285 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
16286 Disable pc relative literal load irrespective of
16287 TARGET_FIX_ERR_A53_84341 for default.
16289 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
16291 * config/sparc/sparc.c (output_return): Output the source location of
16292 the insn in the delay slot, if any.
16293 (output_sibcall): Likewise.
16295 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
16298 * common.opt (-static-pie): New alias.
16299 (shared): Negate static-pie.
16300 (-no-pie): Update help text.
16302 (static-pie): New option.
16303 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
16304 -static-pie support.
16305 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
16306 (LINK_EH_SPEC): Likewise.
16307 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16308 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16309 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16310 * gcc.c (LINK_COMMAND_SPEC): Likewise.
16311 (init_gcc_specs): Likewise.
16312 (init_spec): Likewise.
16313 (display_help): Update help message for -pie.
16314 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
16317 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
16319 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
16320 (movdi_aarch64): Likewise.
16321 (movti_aarch64): Likewise.
16323 2017-09-12 Simon Wright <simon@pushface.org>
16326 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
16327 calculation of the minor version, always output as 0.
16329 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16332 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16333 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
16334 on it early, rather than manual conversion late. For
16335 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
16336 instead of performing manual conversion.
16338 2017-09-12 Carl Love <cel@us.ibm.com>
16340 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
16341 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
16343 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16344 VMULOSW): Add definitions.
16345 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16346 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16347 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16348 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16349 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16351 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16353 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
16355 (movti_aarch64): Likewise.
16356 (movdf_aarch64): Likewise.
16357 (movtf_aarch64): Likewise.
16358 (load_pairdi): Likewise.
16359 (store_pairdi): Likewise.
16360 (load_pairdf): Likewise.
16361 (store_pairdf): Likewise.
16362 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
16363 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
16364 (ldr_got_small_<mode>): Likewise.
16365 (ldr_got_small_28k_<mode>): Likewise.
16366 (ldr_got_tiny): Likewise.
16367 * config/aarch64/iterators.md (ldst_sz): New.
16368 (ldpstp_sz): Likewise.
16369 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
16371 (thunderx_load): Split load_8 to load_16.
16372 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
16374 (thunderx2t99_storepair_basic): Split store_8 to store_16.
16375 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
16376 (xgene1_store_pair): Split store_8 to store_16.
16377 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
16378 (falkor_st_0_st_sd): Split store_8 to store_16.
16380 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16382 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
16383 and store1/2/3/4 to store_4/8/12/16.
16384 * config/aarch64/aarch64.md: Update for rename.
16385 * config/arm/arm.md: Likewise.: Likewise.
16386 * config/arm/arm.c: Likewise.
16387 * config/arm/thumb1.md: Likewise.
16388 * config/arm/thumb2.md: Likewise.
16389 * config/arm/vfp.md: Likewise.
16390 * config/arm/arm-generic.md: Likewise.
16391 * config/arm/arm1020e.md: Likewise.
16392 * config/arm/arm1026ejs.md: Likewise.
16393 * config/arm/arm1136jfs.md: Likewise.
16394 * config/arm/arm926ejs.md: Likewise.
16395 * config/arm/cortex-a15.md: Likewise.
16396 * config/arm/cortex-a17.md: Likewise.
16397 * config/arm/cortex-a5.md: Likewise.
16398 * config/arm/cortex-a53.md: Likewise.
16399 * config/arm/cortex-a57.md: Likewise.
16400 * config/arm/cortex-a7.md: Likewise.
16401 * config/arm/cortex-a8.md: Likewise.
16402 * config/arm/cortex-a9.md: Likewise.
16403 * config/arm/cortex-m4.md: Likewise.
16404 * config/arm/cortex-m7.md: Likewise.
16405 * config/arm/cortex-r4.md: Likewise.
16406 * config/arm/exynos-m1.md: Likewise.
16407 * config/arm/fa526.md: Likewise.
16408 * config/arm/fa606te.md: Likewise.
16409 * config/arm/fa626te.md: Likewise.
16410 * config/arm/fa726te.md: Likewise.
16411 * config/arm/fmp626.md: Likewise.
16412 * config/arm/iwmmxt.md: Likewise.
16413 * config/arm/ldmstm.md: Likewise.
16414 * config/arm/marvell-pj4.md: Likewise.
16415 * config/arm/xgene1.md: Likewise.
16416 * config/aarch64/thunderx.md: Likewise.
16417 * config/aarch64/thunderx2t99.md: Likewise.
16418 * config/aarch64/falkor.md: Likewise.
16420 2017-09-12 Martin Liska <mliska@suse.cz>
16422 * attribs.c (private_lookup_attribute): New function.
16423 * attribs.h (private_lookup_attribute): Declared here.
16424 (lookup_attribute): Called from this place.
16426 2017-09-12 Richard Biener <rguenther@suse.de>
16428 PR tree-optimization/82157
16429 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
16430 stmts with side-effects.
16432 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16433 Alan Hayward <alan.hayward@arm.com>
16434 David Sherwood <david.sherwood@arm.com>
16436 * target.def (hard_regno_nregs): New hook.
16437 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
16438 * targhooks.h (default_hard_regno_nregs): Declare.
16439 * targhooks.c (default_hard_regno_nregs): New function.
16440 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
16441 (TARGET_HARD_REGNO_NREGS): ...this hook.
16442 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
16443 (CLASS_MAX_NREGS): Likewise.
16444 * doc/tm.texi: Regenerate.
16445 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
16446 instead of HARD_REGNO_NREGS.
16447 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
16448 HARD_REGNO_NREGS in the comment.
16449 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
16450 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
16451 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
16452 Return an unsigned int.
16453 (TARGET_HARD_REGNO_NREGS): Redefine.
16454 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
16455 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
16456 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16457 (arc_hard_regno_nregs): New function.
16458 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
16459 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
16460 (arm_hard_regno_nregs): New function.
16461 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
16462 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
16463 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
16464 (TARGET_HARD_REGNO_NREGS): Redefine.
16465 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
16466 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
16467 (HARD_REGNO_NREGS): Delete.
16468 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
16469 (cr16_hard_regno_nregs): New function.
16470 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
16471 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
16472 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
16473 (cris_hard_regno_nregs): New function.
16474 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
16475 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
16476 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16477 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
16478 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
16479 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
16480 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
16481 (frv_hard_regno_nregs): Make static. Take and return an
16483 (frv_class_max_nregs): Remove outdated copy of documentation.
16484 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
16485 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
16486 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
16487 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
16488 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
16489 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
16490 (TARGET_HARD_REGNO_NREGS): Redefine.
16491 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
16492 (CLASS_MAX_NREGS): Update comment.
16493 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
16494 (ia64_hard_regno_nregs): New function.
16495 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
16496 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
16497 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
16498 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
16499 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
16501 (m32c_hard_regno_nregs): Likewise. Make static.
16502 (TARGET_HARD_REGNO_NREGS): Redefine.
16503 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
16504 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
16505 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
16506 (m68k_hard_regno_nregs): New function.
16507 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
16508 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
16509 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
16510 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
16511 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
16512 Take and return an unsigned int.
16513 (TARGET_HARD_REGNO_NREGS): Redefine.
16514 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
16515 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16516 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
16517 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
16518 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
16519 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
16520 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
16521 (msp430_hard_regno_nregs): Make static. Take and return an
16523 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
16524 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
16525 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
16526 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
16527 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
16528 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
16529 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
16530 (TARGET_HARD_REGNO_NREGS): Redefine.
16531 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
16532 (PA_HARD_REGNO_NREGS): ...this.
16533 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
16534 (PA_HARD_REGNO_NREGS): ...this.
16535 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16536 (pa_hard_regno_nregs): New function.
16537 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
16538 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
16539 (pdp11_hard_regno_nregs): New function.
16540 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
16541 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
16542 (rs6000_hard_regno_nregs_hook): New function.
16543 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
16544 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
16545 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
16546 Take and return an unsigned int. Move earlier in file.
16547 (TARGET_HARD_REGNO_NREGS): Redefine.
16548 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
16549 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
16550 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
16551 (rl78_hard_regno_nregs): Make static. Take and return an
16553 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
16554 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
16555 (rs6000_hard_regno_nregs_hook): New function.
16556 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
16557 * config/rx/rx.c (rx_hard_regno_nregs): New function.
16558 (TARGET_HARD_REGNO_NREGS): Redefine.
16559 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
16560 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
16561 instead of HARD_REGNO_NREGS.
16562 (s390_hard_regno_nregs): New function.
16563 (s390_hard_regno_mode_ok): Add comment from s390.h.
16564 (TARGET_HARD_REGNO_NREGS): Redefine.
16565 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
16566 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
16567 (sh_hard_regno_nregs): New function.
16568 (sh_pass_in_reg_p): Use it.
16569 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
16570 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16571 (sparc_hard_regno_nregs): New function.
16572 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
16573 * config/spu/spu.c (spu_hard_regno_nregs): New function.
16574 (spu_function_arg_advance): Use it, supplying a valid register number.
16575 (TARGET_HARD_REGNO_NREGS): Redefine.
16576 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
16577 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
16578 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
16579 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
16580 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
16581 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
16582 (CLASS_MAX_NREGS): Remove copy of old documentation.
16583 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
16584 (visium_hard_regno_nregs): New function.
16585 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
16586 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
16587 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16588 xtensa_hard_regno_nregs): New function.
16589 * system.h (HARD_REGNO_NREGS): Poison.
16591 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16593 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
16594 hard_regno_nregs instead of HARD_REGNO_NREGS.
16595 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
16596 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
16597 (c6x_expand_epilogue): Likewise.
16598 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
16599 (frv_read_iacc_argument): Likewise.
16600 * config/sh/sh.c: Include regs.h.
16601 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16602 (regs_used): Likewise.
16603 (output_stack_adjust): Likewise.
16604 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
16605 * expmed.c: Include regs.h.
16606 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16607 * ree.c: Include regs.h.
16608 (combine_reaching_defs): Use hard_regno_nregs instead of
16610 (add_removable_extension): Likewise.
16612 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16614 * regs.h (hard_regno_nregs): Turn into a function.
16615 (end_hard_regno): Update accordingly.
16616 * caller-save.c (setup_save_areas): Likewise.
16617 (save_call_clobbered_regs): Likewise.
16618 (replace_reg_with_saved_mem): Likewise.
16619 (insert_restore): Likewise.
16620 (insert_save): Likewise.
16621 * combine.c (can_change_dest_mode): Likewise.
16622 (move_deaths): Likewise.
16623 (distribute_notes): Likewise.
16624 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
16625 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
16626 (rs6000_split_multireg_move): Likewise.
16627 (rs6000_register_move_cost): Likewise.
16628 (rs6000_memory_move_cost): Likewise.
16629 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
16630 (rs6000_split_multireg_move): Likewise.
16631 (rs6000_register_move_cost): Likewise.
16632 (rs6000_memory_move_cost): Likewise.
16633 * cselib.c (cselib_reset_table): Likewise.
16634 (cselib_lookup_1): Likewise.
16635 * emit-rtl.c (set_mode_and_regno): Likewise.
16636 * function.c (aggregate_value_p): Likewise.
16637 * ira-color.c (setup_profitable_hard_regs): Likewise.
16638 (check_hard_reg_p): Likewise.
16639 (calculate_saved_nregs): Likewise.
16640 (assign_hard_reg): Likewise.
16641 (improve_allocation): Likewise.
16642 (calculate_spill_cost): Likewise.
16643 * ira-emit.c (modify_move_list): Likewise.
16644 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
16645 (ira_hard_reg_in_set_p): Likewise.
16646 * ira.c (setup_reg_mode_hard_regset): Likewise.
16647 (clarify_prohibited_class_mode_regs): Likewise.
16648 (check_allocation): Likewise.
16649 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16650 (lra_setup_reg_renumber): Likewise.
16651 (setup_try_hard_regno_pseudos): Likewise.
16652 (spill_for): Likewise.
16653 (assign_hard_regno): Likewise.
16654 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
16655 * lra-constraints.c (in_class_p): Likewise.
16656 (lra_constraint_offset): Likewise.
16657 (simplify_operand_subreg): Likewise.
16658 (lra_constraints): Likewise.
16659 (split_reg): Likewise.
16660 (split_if_necessary): Likewise.
16661 (invariant_p): Likewise.
16662 (inherit_in_ebb): Likewise.
16663 * lra-lives.c (process_bb_lives): Likewise.
16664 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
16665 (get_hard_regs): Likewise.
16666 (do_remat): Likewise.
16667 * lra-spills.c (assign_spill_hard_regs): Likewise.
16668 * mode-switching.c (create_pre_exit): Likewise.
16669 * postreload.c (reload_combine_recognize_pattern): Likewise.
16670 * recog.c (peep2_find_free_register): Likewise.
16671 * regcprop.c (kill_value_regno): Likewise.
16672 (set_value_regno): Likewise.
16673 (copy_value): Likewise.
16674 (maybe_mode_change): Likewise.
16675 (find_oldest_value_reg): Likewise.
16676 (copyprop_hardreg_forward_1): Likewise.
16677 * regrename.c (check_new_reg_p): Likewise.
16678 (regrename_do_replace): Likewise.
16679 * reload.c (push_reload): Likewise.
16680 (combine_reloads): Likewise.
16681 (find_dummy_reload): Likewise.
16682 (operands_match_p): Likewise.
16683 (find_reloads): Likewise.
16684 (find_equiv_reg): Likewise.
16685 (reload_adjust_reg_for_mode): Likewise.
16686 * reload1.c (count_pseudo): Likewise.
16687 (count_spilled_pseudo): Likewise.
16688 (find_reg): Likewise.
16689 (clear_reload_reg_in_use): Likewise.
16690 (free_for_value_p): Likewise.
16691 (allocate_reload_reg): Likewise.
16692 (choose_reload_regs): Likewise.
16693 (reload_adjust_reg_for_temp): Likewise.
16694 (emit_reload_insns): Likewise.
16695 (delete_output_reload): Likewise.
16696 * rtlanal.c (subreg_get_info): Likewise.
16697 * sched-deps.c (sched_analyze_reg): Likewise.
16698 * sel-sched.c (init_regs_for_mode): Likewise.
16699 (mark_unavailable_hard_regs): Likewise.
16700 (choose_best_reg_1): Likewise.
16701 (verify_target_availability): Likewise.
16702 * valtrack.c (dead_debug_insert_temp): Likewise.
16703 * var-tracking.c (track_loc_p): Likewise.
16704 (emit_note_insn_var_location): Likewise.
16705 * varasm.c (make_decl_rtl): Likewise.
16706 * reginfo.c (choose_hard_reg_mode): Likewise.
16707 (init_reg_modes_target): Refer directly to
16708 this_target_regs->x_hard_regno_nregs.
16710 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16712 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
16713 instead of hard_regno_nregs.
16715 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16717 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
16718 end_hard_regno instead of hard_regno_nregs.
16719 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
16720 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
16721 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
16722 * ira-color.c (improve_allocation): Likewise.
16723 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16724 * lra-lives.c (mark_regno_live): Likewise.
16725 (mark_regno_dead): Likewise.
16726 * lra-remat.c (operand_to_remat): Likewise.
16727 * lra.c (collect_non_operand_hard_regs): Likewise.
16728 * postreload.c (reload_combine_note_store): Likewise.
16729 (move2add_valid_value_p): Likewise.
16730 * reload.c (regno_clobbered_p): Likewise.
16732 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16734 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
16736 * config/v850/v850.c (v850_reorg): Likewise.
16737 * reload.c (refers_to_regno_for_reload_p): Likewise.
16738 (find_equiv_reg): Likewise.
16739 * reload1.c (reload_reg_reaches_end_p): Likewise.
16741 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16743 * caller-save.c (add_used_regs): Use REG_NREGS instead of
16745 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
16746 * config/arm/arm.c (output_move_neon): Likewise.
16747 (arm_attr_length_move_neon): Likewise.
16748 (neon_split_vcombine): Likewise.
16749 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
16750 (c6x_mark_reg_written): Likewise.
16751 (c6x_dwarf_register_span): Likewise.
16752 * config/i386/i386.c (ix86_save_reg): Likewise.
16753 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
16754 (rws_access_reg): Likewise.
16755 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
16756 * mode-switching.c (create_pre_exit): Likewise.
16757 * ree.c (combine_reaching_defs): Likewise.
16758 (add_removable_extension): Likewise.
16759 * regcprop.c (find_oldest_value_reg): Likewise.
16760 (copyprop_hardreg_forward_1): Likewise.
16761 * reload.c (reload_inner_reg_of_subreg): Likewise.
16762 (push_reload): Likewise.
16763 (combine_reloads): Likewise.
16764 (find_dummy_reload): Likewise.
16765 (reload_adjust_reg_for_mode): Likewise.
16766 * reload1.c (find_reload_regs): Likewise.
16767 (forget_old_reloads_1): Likewise.
16768 (reload_reg_free_for_value_p): Likewise.
16769 (reload_adjust_reg_for_temp): Likewise.
16770 (emit_reload_insns): Likewise.
16771 (delete_output_reload): Likewise.
16772 * sel-sched.c (choose_best_reg_1): Likewise.
16773 (choose_best_pseudo_reg): Likewise.
16775 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16776 Alan Hayward <alan.hayward@arm.com>
16777 David Sherwood <david.sherwood@arm.com>
16779 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
16780 * target.def (slow_unaligned_access): New hook.
16781 * targhooks.h (default_slow_unaligned_access): Declare.
16782 * targhooks.c (default_slow_unaligned_access): New function.
16783 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
16784 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
16785 * doc/tm.texi: Regenerate.
16786 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
16787 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
16788 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
16790 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
16791 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
16793 (rs6000_slow_unaligned_access): New function.
16794 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16795 (expand_block_compare): Likewise.
16796 (expand_strn_compare): Likewise.
16797 (rs6000_rtx_costs): Likewise.
16798 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
16799 (riscv_slow_unaligned_access): Likewise.
16800 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
16801 (riscv_slow_unaligned_access_p): ...this and make static.
16802 (riscv_option_override): Update accordingly.
16803 (riscv_slow_unaligned_access): New function.
16804 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16805 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
16806 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16807 (rs6000_slow_unaligned_access): New function.
16808 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16809 (rs6000_rtx_costs): Likewise.
16810 * config/rs6000/rs6000-string.c (expand_block_compare)
16811 (expand_strn_compare): Use targetm.slow_unaligned_access instead
16812 of SLOW_UNALIGNED_ACCESS.
16813 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
16814 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
16815 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
16816 of SLOW_UNALIGNED_ACCESS.
16817 * expmed.c (simple_mem_bitfield_p): Likewise.
16818 * expr.c (alignment_for_piecewise_move): Likewise.
16819 (emit_group_load_1): Likewise.
16820 (emit_group_store): Likewise.
16821 (copy_blkmode_from_reg): Likewise.
16822 (emit_push_insn): Likewise.
16823 (expand_assignment): Likewise.
16824 (store_field): Likewise.
16825 (expand_expr_real_1): Likewise.
16826 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16827 * lra-constraints.c (simplify_operand_subreg): Likewise.
16828 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16829 * gimple-ssa-store-merging.c: Likewise in block comment at start
16831 * tree-ssa-strlen.c: Include target.h.
16832 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
16833 of SLOW_UNALIGNED_ACCESS.
16834 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
16836 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16838 PR rtl-optimization/82185
16839 * expmed.c (emit_store_flag_int): Only test tem if it has been
16842 2017-09-12 Richard Biener <rguenther@suse.de>
16844 PR middle-end/82149
16845 * match.pd ((FTYPE) N CMP CST): Fix typo.
16847 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16849 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16851 (mips_near_type_p): Add 'short_call' attribute as a synonym
16853 * doc/extend.texi (short_call): Document new function attribute.
16855 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16858 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16859 assertion check that in the condition.
16860 (get_atomic_generic_size): Likewise. Before testing if parameter
16861 has pointer type, if it has array type, call for C++
16862 default_conversion to perform array-to-pointer conversion.
16864 2017-09-12 Richard Biener <rguenther@suse.de>
16866 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16867 for operations we cannot scalarize.
16869 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16871 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16872 vectors heap vectors. Clean up comments.
16873 Make visited_bbs a reference.
16874 (profitable_jump_thread_path): Make GC
16875 vectors heap vectors. Clean up comments.
16877 (convert_and_register_jump_thread_path): Make GC vectors heap
16879 (check_subpath_and_update_thread_path): Same. Clean up comments.
16880 Make visited_bbs a reference.
16881 (handle_phi): Abstract common code to to
16882 register_jump_thread_path_if_profitable.
16883 Rename VAR_BB to DEF_BB.
16885 Make GC vectors heap vectors.
16886 Make visited_bbs a reference.
16887 (handle_assignment): Same.
16888 (register_jump_thread_path_if_profitable): New.
16889 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16891 Make GC vectors heap vectors. Clean up comments.
16892 Make visited_bbs a reference.
16893 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16894 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16897 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16900 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16901 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16903 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16905 Revert r251800 and r251799.
16907 2017-09-11 Martin Jambor <mjambor@suse.cz>
16910 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16911 arguments in advance.
16912 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16913 use it to find predecessor edges.
16914 (naive_outof_ssa): Collect vector of predecessors.
16916 2017-09-08 Jason Merrill <jason@redhat.com>
16918 PR c++/70029 - ICE with ref-qualifier and -flto
16919 * langhooks.h (struct lang_hooks_for_types): Add
16920 copy_lang_qualifiers.
16921 * attribs.c (build_type_attribute_qual_variant): Use it.
16922 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16924 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16925 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16927 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16930 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16931 (*mulsi3_sp64): New instruction.
16932 (mulsi3): New expander.
16934 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16936 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16938 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16940 * sancov.c: Include memmodel.h.
16942 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16945 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16948 2017-09-07 Carl Love <cel@us.ibm.com>
16950 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16951 the sldi instruction.
16953 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16955 * sancov.c: Include tm_p.h.
16957 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16960 * output.h (switch_to_other_text_partition): New declaration.
16961 * varasm.c (switch_to_other_text_partition): New function.
16962 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16963 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16964 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16965 to the other text partition before emitting LCL label and switch back
16966 after emitting the word after it.
16968 2017-09-07 Richard Biener <rguenther@suse.de>
16970 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16971 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16972 critical edges here, after loop init.
16973 (pass_data_pre): Remove PROP_no_crit_edges flags.
16974 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16975 for valueization of call args to avoid leaking VN_TOP.
16976 (visit_use): Assert we do not visit default defs.
16977 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16978 Use error_mark_node to more easily detect leaking VN_TOP.
16979 All default-defs are varying, not VN_TOP. Mark them visited.
16980 (run_scc_vn): Make code match comment.
16982 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16984 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16985 OPTION_MASK_FLOAT128_KEYWORD.
16986 (POWERPC_MASKS): Likewise.
16987 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16988 support for the -mfloat128-type option, and make -mfloat128
16989 default on PowerPC Linux systems. Define or undefine
16990 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16991 Define __float128 to be __ieee128 if IEEE 128-bit support is
16992 enabled, or undefine it.
16993 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16994 Delete defining __FLOAT128_TYPE__.
16995 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16996 -mfloat128-type option and make -mfloat128 default on PowerPC
16998 (TARGET_FLOAT128_TYPE): Likewise.
16999 (-mfloat128-type): Likewise.
17000 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17001 Delete the -mfloat128-type option and make -mfloat128 default on
17002 PowerPC Linux systems. Always use __ieee128 to be the keyword for
17003 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
17004 128-bit floating point is enabled. Change tests from using
17005 -mfloat128-type to -mfloat128.
17006 (rs6000_mangle_type): Use the correct mangling for the __float128
17007 type even if normal long double is restricted to 64-bits.
17008 (floatn_mode): Enable the _Float128 type by default on VSX Linux
17010 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
17011 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
17012 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
17014 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
17015 documentation for -mfloat128.
17017 2017-09-06 Olivier Hainque <hainque@adacore.com>
17019 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
17021 2017-09-06 Wish Wu <wishwu007@gmail.com>
17022 Jakub Jelinek <jakub@redhat.com>
17024 * asan.c (initialize_sanitizer_builtins): Add
17025 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
17026 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
17027 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
17028 BT_FN_VOID_UINT64_PTR variables.
17029 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
17030 (BT_FN_VOID_UINT16_UINT16): Likewise.
17031 (BT_FN_VOID_UINT32_UINT32): Likewise.
17032 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
17033 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
17034 (BT_FN_VOID_UINT64_PTR): Likewise.
17035 * common.opt (flag_sanitize_coverage): New variable.
17036 (fsanitize-coverage=trace-pc): Remove.
17037 (fsanitize-coverage=): Add.
17038 * flag-types.h (enum sanitize_coverage_code): New enum.
17039 * fold-const.c (fold_range_test): Disable non-short-circuit
17040 optimization if flag_sanitize_coverage.
17041 (fold_truth_andor): Likewise.
17042 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
17043 * opts.c (COVERAGE_SANITIZER_OPT): Define.
17044 (coverage_sanitizer_opts): New array.
17045 (get_closest_sanitizer_option): Add OPTS argument, handle also
17046 OPT_fsanitize_coverage_.
17047 (parse_sanitizer_options): Adjusted to also handle
17048 OPT_fsanitize_coverage_.
17049 (common_handle_option): Add OPT_fsanitize_coverage_.
17050 * sancov.c (instrument_comparison, instrument_switch): New function.
17051 (sancov_pass): Add trace-cmp support.
17052 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
17053 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
17054 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
17055 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
17056 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
17057 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
17058 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
17059 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
17060 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
17061 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
17063 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
17065 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
17066 error. Only quit immediately if parsing is complete.
17067 (BEGIN): Initialize fatal_err and parse_done.
17068 (begin fpu, end fpu): Check number of arguments.
17069 (begin arch, end arch): Likewise.
17070 (begin cpu, end cpu): Likewise.
17071 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
17072 (optalias): Likewise.
17074 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
17076 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
17077 * config/arm/arm-isa.h: Delete. Move definitions to ...
17078 * arm-cpus.in: ... here. Use new feature and fgroup values.
17079 * config/arm/arm.c (arm_option_override): Use lower case for feature
17081 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
17082 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
17083 * config/arm/parsecpu.awk (END): Add new command 'isa'.
17085 (print_isa_bits_for): New function.
17086 (gen_isa): New function.
17087 (gen_comm_data): Use print_isa_bits_for.
17088 (define feature): New keyword.
17089 (define fgroup): New keyword.
17090 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
17091 (arm-isa.h): Add rule to generate file.
17092 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
17093 case for feature bit names.
17095 2017-09-06 Richard Biener <rguenther@suse.de>
17097 * tree-ssa-pre.c (NECESSARY): Remove.
17098 (create_expression_by_pieces): Do not touch pass-local flags.
17099 (insert_into_preds_of_block): Likewise.
17100 (do_pre_regular_insertion): Likewise.
17101 (eliminate_insert): Likewise.
17102 (eliminate_dom_walker::before_dom_children): Likewise.
17103 (fini_eliminate): Do not look at inserted_exprs.
17104 (mark_operand_necessary): Remove.
17105 (remove_dead_inserted_code): Replace with simple work-list
17106 algorithm based on inserted_exprs and SSA uses.
17107 (pass_pre::execute): Re-order fini_eliminate and
17108 remove_dead_inserted_code.
17110 2017-09-06 Olivier Hainque <hainque@adacore.com>
17112 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
17113 for VxWorks 7. Adjust surrounding comments.
17115 2017-09-06 Richard Biener <rguenther@suse.de>
17117 * gimple-ssa-strength-reduction.c
17118 (find_candidates_dom_walker::before_dom_children): Also allow
17121 2017-09-06 Richard Biener <rguenther@suse.de>
17123 PR tree-optimization/82108
17124 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
17125 for gap in the non-permutation SLP case.
17127 2017-09-06 Martin Jambor <mjambor@suse.cz>
17129 PR tree-optimization/82078
17130 * tree-sra.c (sort_and_splice_var_accesses): Move call to
17131 add_access_to_work_queue...
17132 (build_accesses_from_assign): ...here.
17133 (propagate_all_subaccesses): Make sure racc is the group
17134 representative, if there is one.
17136 2017-09-06 Jakub Jelinek <jakub@redhat.com>
17138 PR middle-end/82095
17139 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
17142 2017-09-06 Richard Biener <rguenther@suse.de>
17144 * gimple-ssa-strength-reduction.c
17145 (find_candidates_dom_walker::before_doom_children): Use a
17146 type and not a mode check.
17148 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
17151 * config/arm/predicates.md (arm_general_adddi_operand): Create new
17153 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
17155 2017-09-05 Jeff Law <law@redhat.com>
17157 PR tree-optimization/64910
17158 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
17159 cases where we have 3 or more operands.
17161 2017-09-05 Jakub Jelinek <jakub@redhat.com>
17163 PR middle-end/81768
17164 * omp-low.c (lower_omp_for): Recompute tree invariant if
17165 gimple_omp_for_initial/final is ADDR_EXPR.
17167 PR middle-end/81768
17168 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
17169 into gimple val before gimplification fo the COND_EXPR.
17171 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
17173 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
17174 REGION_COPY argument.
17175 (thread_through_all_blocks): Remove unused argument to
17176 duplicate_thread_path.
17178 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17179 Alan Hayward <alan.hayward@arm.com>
17180 David Sherwood <david.sherwood@arm.com>
17182 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
17183 Take a scalar_mode rather than a machine_mode.
17184 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17185 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
17186 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17187 (aarch64_gen_adjusted_ldpstp): Likewise.
17188 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
17190 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17191 Alan Hayward <alan.hayward@arm.com>
17192 David Sherwood <david.sherwood@arm.com>
17194 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
17195 Take a scalar_int_mode instead of a machine_mode.
17196 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
17197 (aarch64_output_scalar_simd_mov_immediate): Likewise.
17198 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
17199 (aarch64_simd_attr_length_rglist): Delete.
17200 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
17201 a scalar_int_mode instead of a machine_mode.
17202 (aarch64_add_offset): Likewise.
17203 (aarch64_internal_mov_immediate): Likewise
17204 (aarch64_add_constant_internal): Likewise.
17205 (aarch64_add_constant): Likewise.
17206 (aarch64_movw_imm): Likewise.
17207 (aarch64_rtx_arith_op_extract_p): Likewise.
17208 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
17209 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
17210 Remove assert that the mode isn't a vector.
17211 (aarch64_output_scalar_simd_mov_immediate): Likewise.
17212 (aarch64_expand_mov_immediate): Update calls after above changes.
17213 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
17214 (aarch64_and_bitmask_imm): Check for scalar integer modes.
17215 (aarch64_move_imm): Likewise.
17216 (aarch64_can_const_movi_rtx_p): Likewise.
17217 (aarch64_strip_extend): Likewise.
17218 (aarch64_extr_rtx_p): Likewise.
17219 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
17220 a CONST_INT when the mode parameter is VOIDmode.
17221 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
17223 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17225 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
17226 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17227 (bitwise_type_for_mode): Update accordingly.
17229 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17231 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
17232 * stor-layout.c (mode_for_size_tree): Likewise.
17233 (mode_for_array): Update accordingly.
17234 (layout_decl): Likewise.
17235 (compute_record_mode): Likewise. Only set the mode once.
17237 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17239 * target.def (get_mask_mode): Change return type to opt_mode.
17241 * doc/tm.texi: Regenerate.
17242 * targhooks.h (default_get_mask_mode): Return an opt_mode.
17243 * targhooks.c (default_get_mask_mode): Likewise.
17244 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17245 * optabs-query.c (can_vec_mask_load_store_p): Update use of
17246 targetm.get_mask_mode.
17247 * tree.c (build_truth_vector_type): Likewise.
17249 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17251 * machmode.h (mode_for_vector): Return an opt_mode.
17252 * stor-layout.c (mode_for_vector): Likewise.
17253 (mode_for_int_vector): Update accordingly.
17254 (layout_type): Likewise.
17255 * config/i386/i386.c (emit_memmov): Likewise.
17256 (ix86_expand_set_or_movmem): Likewise.
17257 (ix86_expand_vector_init): Likewise.
17258 (ix86_get_mask_mode): Likewise.
17259 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
17261 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
17262 * expmed.c (extract_bit_field_1): Likewise.
17263 * expr.c (expand_expr_real_2): Likewise.
17264 * optabs-query.c (can_vec_perm_p): Likewise.
17265 (can_vec_mask_load_store_p): Likewise.
17266 * optabs.c (expand_vec_perm): Likewise.
17267 * targhooks.c (default_get_mask_mode): Likewise.
17268 * tree-vect-stmts.c (vectorizable_store): Likewise.
17269 (vectorizable_load): Likewise.
17270 (get_vectype_for_scalar_type_and_size): Likewise.
17272 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17274 * machmode.h (mode_for_int_vector): New function.
17275 * stor-layout.c (mode_for_int_vector): Likewise.
17276 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
17277 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
17278 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17279 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17280 (s390_expand_vcond): Likewise.
17282 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17284 * machmode.h (opt_machine_mode): New type.
17285 (opt_mode<T>): Allow construction from anything that can be
17287 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
17288 (mode_for_size): Return an opt_machine_mode.
17289 * stor-layout.c (mode_for_size): Likewise.
17290 (mode_for_size_tree): Update call accordingly.
17291 (bitwise_mode_for_mode): Likewise.
17292 (make_fract_type): Likewise.
17293 (make_accum_type): Likewise.
17294 * caller-save.c (replace_reg_with_saved_mem): Update call
17296 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17297 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17298 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17299 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17300 * expmed.c (extract_bit_field_1): Likewise.
17301 * reload.c (get_secondary_mem): Likewise.
17302 * varasm.c (assemble_integer): Likewise.
17303 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
17306 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17308 * machmode.h (decimal_float_mode_for_size): New function.
17309 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
17310 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
17311 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
17312 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
17313 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
17314 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
17316 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17318 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
17319 (get_builtin_sync_mode): Likewise.
17320 (expand_ifn_atomic_compare_exchange): Likewise.
17321 (expand_builtin_atomic_clear): Likewise.
17322 (expand_builtin_atomic_test_and_set): Likewise.
17323 (fold_builtin_atomic_always_lock_free): Likewise.
17324 * calls.c (compute_argument_addresses): Likewise.
17325 (emit_library_call_value_1): Likewise.
17326 (store_one_arg): Likewise.
17327 * combine.c (combine_instructions): Likewise.
17328 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
17329 * config/arm/arm.c (arm_function_value): Likewise.
17330 (aapcs_allocate_return_reg): Likewise.
17331 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
17332 * config/i386/i386.c (construct_container): Likewise.
17333 (ix86_gimplify_va_arg): Likewise.
17334 (ix86_expand_sse_cmp): Likewise.
17335 (emit_memmov): Likewise.
17336 (emit_memset): Likewise.
17337 (expand_small_movmem_or_setmem): Likewise.
17338 (ix86_expand_pextr): Likewise.
17339 (ix86_expand_pinsr): Likewise.
17340 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
17341 * config/microblaze/microblaze.c (microblaze_block_move_straight):
17343 * config/mips/mips.c (mips_function_value_1) Likewise.
17344 (mips_block_move_straight): Likewise.
17345 (mips_expand_ins_as_unaligned_store): Likewise.
17346 * config/powerpcspe/powerpcspe.c
17347 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17348 (rs6000_darwin64_record_arg_flush): Likewise.
17349 * config/rs6000/rs6000.c
17350 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17351 (rs6000_darwin64_record_arg_flush): Likewise.
17352 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
17353 (sparc_function_value_1): Likewise.
17354 * config/spu/spu.c (adjust_operand): Likewise.
17355 (spu_emit_branch_or_set): Likewise.
17356 (arith_immediate_p): Likewise.
17357 * emit-rtl.c (gen_lowpart_common): Likewise.
17358 * expr.c (expand_expr_real_1): Likewise.
17359 * function.c (assign_parm_setup_block): Likewise.
17360 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
17361 * reload1.c (alter_reg): Likewise.
17362 * stor-layout.c (mode_for_vector): Likewise.
17363 (layout_type): Likewise.
17365 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17367 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
17368 (spu_convert_move): Likewise.
17369 * lower-subreg.c (resolve_simple_move): Likewise.
17371 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17374 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
17375 define_insn to a define_expand.
17376 (altivec_vsum2sws_direct): New define_insn.
17377 (altivec_vsumsws): Convert from a define_insn to a define_expand.
17379 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17381 * config/arm/arm.c (arm_option_params_internal): Improve setting of
17384 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
17389 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
17390 realign stack if stack alignment needed is less than incoming
17393 2017-09-05 Marek Polacek <polacek@redhat.com>
17396 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
17399 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17401 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
17403 2017-09-05 Richard Biener <rguenther@suse.de>
17405 PR tree-optimization/82084
17406 * fold-const.c (can_native_encode_string_p): Handle wide characters.
17408 2017-09-05 Richard Biener <rguenther@suse.de>
17410 PR tree-optimization/82102
17411 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
17413 2017-09-05 Martin Liska <mliska@suse.cz>
17415 PR tree-optimization/82032
17416 * tree-cfg.c (generate_range_test): New function.
17417 * tree-cfg.h (generate_range_test): Declared here.
17418 * tree-cfgcleanup.c (convert_single_case_switch): New function.
17419 (cleanup_control_expr_graph): Use it.
17420 * tree-switch-conversion.c (try_switch_expansion): Remove
17422 (emit_case_nodes): Use generate_range_test.
17424 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17427 * config/i386/i386.md (*<btsc><mode>_mask): Add
17428 TARGET_USE_BT to insn constraint.
17429 (*btr<mode>_mask): Ditto.
17431 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
17433 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
17434 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
17436 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
17439 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
17440 TARGET_NEON and TARGET_IWMMXT.
17441 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
17442 TARGET_NEON and TARGET_IWMMXT.
17443 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
17445 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17447 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
17448 (ix86_rewrite_tls_address): Ditto.
17449 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
17450 (ix86_rewrite_tls_address_1): Ditto.
17451 (ix86_rewrite_tls_address): Ditto.
17452 * config/i386/predicates.md (tls_address_pattern): New predicate.
17453 * config/i386/i386.md (TLS address splitter): New splitter.
17455 2017-09-04 Richard Biener <rguenther@suse.de>
17457 PR tree-optimization/82084
17458 * fold-const.h (can_native_encode_string_p): Declare.
17459 * fold-const.c (can_native_encode_string_p): Factor out from ...
17460 (native_encode_string): ... here.
17461 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
17462 vectorizing stores from constants we later cannot handle.
17464 2017-09-04 Marek Polacek <polacek@redhat.com>
17467 * doc/invoke.texi: Update -Wtautological-compare documentation.
17469 2017-09-04 Jeff Law <law@redhat.com>
17471 PR tree-optimization/64910
17472 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
17473 swap the first and last operand if the last is a constant.
17475 2017-09-04 Marek Polacek <polacek@redhat.com>
17478 * convert.c (do_narrow): When sanitizing signed integer overflows,
17479 bail out for signed types.
17480 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
17482 2017-09-04 Richard Biener <rguenther@suse.de>
17484 PR tree-optimization/82060
17485 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17486 Move devirtualization after stmt folding and before EH/AB/noreturn
17487 cleanup to get the stmt refs canonicalized. Use a bool instead
17488 of gimple_modified_p since that doesn't work for NOPs. Schedule
17489 NOPs generated by folding for removal.
17491 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17492 Alan Hayward <alan.hayward@arm.com>
17493 David Sherwood <david.sherwood@arm.com>
17495 * coretypes.h (pad_direction): New enum.
17496 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
17497 (FUNCTION_ARG_PADDING): Likewise.
17498 * target.def (function_arg_padding): New hook.
17499 * targhooks.h (default_function_arg_padding): Declare.
17500 * targhooks.c (default_function_arg_padding): New function.
17501 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
17502 (TARGET_FUNCTION_ARG_PADDING): ...this.
17503 * doc/tm.texi: Regenerate.
17504 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
17505 instead of direction.
17506 (compute_argument_addresses): Likewise.
17507 (load_register_parameters): Likewise.
17508 (emit_library_call_value_1): Likewise.
17509 (store_one_arg): Use targetm.calls.function_arg_padding instead
17510 of FUNCTION_ARG_PADDING.
17511 (must_pass_in_stack_var_size_or_pad): Likewise.
17512 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
17513 (emit_group_store): Likewise.
17514 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
17515 instead of FUNCTION_ARG_PADDING.
17516 (emit_push_insn): Likewise, and propagate enum change throughout
17518 * function.h (direction): Delete.
17519 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
17521 * function.c (assign_parm_find_stack_rtl): Likewise.
17522 (assign_parm_setup_block_p): Likewise.
17523 (assign_parm_setup_block): Likewise.
17524 (gimplify_parameters): Likewise.
17525 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
17526 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
17528 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
17529 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17530 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
17531 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
17532 (aarch64_function_arg_padding): ...this new function.
17533 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
17534 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17535 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
17536 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17537 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
17538 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17539 (arm_pad_arg_upward): Replace with...
17540 (arm_function_arg_padding): ...this new function.
17541 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
17543 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
17544 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
17545 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17546 (ia64_hpux_function_arg_padding): Replace with...
17547 (ia64_function_arg_padding): ...this new function. Use pad_direction
17548 instead of direction. Check for TARGET_HPUX.
17549 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
17550 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17551 (iq2000_function_arg_padding): New function.
17552 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
17553 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
17554 (mips_function_arg_padding): ...this new function.
17555 (mips_pad_reg_upward): Update accordingly.
17556 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17557 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
17558 targetm.calls.function_arg_padding.
17559 (FUNCTION_ARG_PADDING): Delete.
17560 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17561 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
17562 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17563 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
17564 (nios2_block_reg_padding): Return pad_direction instead of direction.
17565 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
17566 instead of direction.
17567 (nios2_function_arg_padding): Likewise. Make static.
17568 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17569 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
17570 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
17571 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
17572 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17573 (pa_function_arg_padding): Make static. Return pad_direction instead
17575 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
17576 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17577 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
17578 instead of direction. Use targetm.calls.function_arg_padding.
17579 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
17580 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17581 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
17582 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
17583 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
17585 (function_arg_padding): Rename to...
17586 (rs6000_function_arg_padding): ...this. Make static. Return
17587 pad_direction instead of direction.
17588 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17589 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
17590 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17591 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
17592 instead of direction. Use targetm.calls.function_arg_padding.
17593 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
17594 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17595 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
17596 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
17597 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17598 (function_arg_padding): Rename to...
17599 (rs6000_function_arg_padding): ...this. Make static. Return
17600 pad_direction instead of direction.
17601 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17602 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
17603 * config/s390/s390.c (s390_function_arg_padding): New function.
17604 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17605 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
17606 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
17607 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17608 (function_arg_padding): Rename to...
17609 (sparc_function_arg_padding): ...this. Make static. Return
17610 pad_direction instead of direction.
17611 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
17612 * config/spu/spu.c (spu_function_arg_padding): New function.
17613 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17614 * system.h (FUNCTION_ARG_PADDING): Poison.
17616 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17617 Alan Hayward <alan.hayward@arm.com>
17618 David Sherwood <david.sherwood@arm.com>
17620 * target.def (modes_tieable_p): New hook.
17621 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
17622 (TARGET_MODES_TIEABLE_P): ...this.
17623 * doc/tm.texi.in: Regenerate.
17624 * hooks.h (hook_bool_mode_mode_true): Declare.
17625 * hooks.c (hook_bool_mode_mode_true): New function.
17626 * combine.c (subst): Use targetm.modes_tieable_p instead of
17628 * dse.c (find_shift_sequence): Likewise.
17629 * expmed.c (extract_low_bits): Likewise.
17630 * lower-subreg.c: Include target.h.
17631 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
17633 * rtlanal.c (rtx_cost): Likewise.
17634 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
17635 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
17636 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
17637 (TARGET_MODES_TIEABLE_P): Redefine.
17638 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
17639 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
17640 (TARGET_MODES_TIEABLE_P): Redefine.
17641 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
17642 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
17643 (arc_modes_tieable_p): New function.
17644 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
17645 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
17646 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
17647 (arm_modes_tieable_p): Make static.
17648 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
17649 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
17650 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
17651 (TARGET_MODES_TIEABLE_P): Redefine.
17652 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
17653 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
17654 (TARGET_MODES_TIEABLE_P): Redefine.
17655 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
17656 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
17657 (cr16_modes_tieable_p): New function.
17658 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
17659 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
17660 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
17661 (TRULY_NOOP_TRUNCATION): Update comment.
17662 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
17663 (TRULY_NOOP_TRUNCATION): Update comment.
17664 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
17665 (frv_modes_tieable_p): New function.
17666 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
17667 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
17668 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
17669 (TARGET_MODES_TIEABLE_P): Redefine.
17670 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
17671 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
17672 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
17673 (TARGET_MODES_TIEABLE_P): Redefine.
17674 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
17675 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
17676 (ia64_modes_tieable_p): New function.
17677 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
17678 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
17679 (iq2000_modes_tieable_p): New function.
17680 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
17681 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
17682 (lm32_modes_tieable_p): New function.
17683 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
17684 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
17685 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
17686 (TARGET_MODES_TIEABLE_P): Redefine.
17687 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
17688 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
17689 (m32r_modes_tieable_p): New function.
17690 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
17691 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
17692 (m68k_modes_tieable_p): New function.
17693 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
17694 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
17695 (mcore_modes_tieable_p): New function.
17696 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
17697 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
17699 (TARGET_MODES_TIEABLE_P): Redefine.
17700 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
17701 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
17702 * config/mips/mips.c (mips_modes_tieable_p): Make static.
17703 (TARGET_MODES_TIEABLE_P): Redefine.
17704 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
17705 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
17706 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
17707 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
17708 (mn10300_modes_tieable_p): ...this and make static.
17709 (TARGET_MODES_TIEABLE_P): Redefine.
17710 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
17711 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
17712 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
17713 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
17714 (msp430_modes_tieable_p): Make static.
17715 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
17716 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
17717 (TARGET_MODES_TIEABLE_P): Redefine.
17718 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
17719 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
17720 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
17721 (TARGET_MODES_TIEABLE_P): Redefine.
17722 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
17723 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
17724 * config/pa/pa.c (pa_modes_tieable_p): Make static.
17725 (TARGET_MODES_TIEABLE_P): Redefine.
17726 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
17727 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
17728 (pdp11_modes_tieable_p): New function.
17729 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
17730 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
17731 (rs6000_modes_tieable_p): New function.
17732 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17733 * config/powerpcspe/powerpcspe.md: Update comment.
17734 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
17735 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
17736 (TARGET_MODES_TIEABLE_P): Redefine.
17737 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
17738 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
17739 (rl78_modes_tieable_p): New function.
17740 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
17741 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
17742 (rs6000_modes_tieable_p): New function.
17743 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17744 * config/rs6000/rs6000.md: Update comment.
17745 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
17746 * config/rx/rx.c (rx_modes_tieable_p): New function.
17747 (TARGET_MODES_TIEABLE_P): Redefine.
17748 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
17749 * config/s390/s390.c (s390_modes_tieable_p): New function.
17750 (TARGET_MODES_TIEABLE_P): Redefine.
17751 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
17752 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
17753 (sh_modes_tieable_p): New function.
17754 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
17755 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
17756 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
17757 (sparc_modes_tieable_p): Make static.
17758 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
17759 * config/spu/spu.c (spu_modes_tieable_p): New function.
17760 (TARGET_MODES_TIEABLE_P): Redefine.
17761 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
17762 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
17763 (TARGET_MODES_TIEABLE_P): Redefine.
17764 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
17765 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
17766 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
17767 * config/v850/v850.c (v850_modes_tieable_p): New function.
17768 (TARGET_MODES_TIEABLE_P): Redefine.
17769 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
17770 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
17771 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
17772 (visium_modes_tieable_p): New function.
17773 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
17774 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
17775 (xtensa_modes_tieable_p): New function.
17776 * system.h (MODES_TIEABLE_P): Poison.
17778 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17779 Alan Hayward <alan.hayward@arm.com>
17780 David Sherwood <david.sherwood@arm.com>
17782 * target.def (hard_regno_mode_ok): New hook.
17783 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
17784 (TARGET_HARD_REGNO_MODE_OK): ...this.
17785 * doc/tm.texi.in: Regenerate.
17786 * hooks.h (hook_bool_uint_mode_true): Declare.
17787 * hooks.c (hook_bool_uint_mode_true): New function.
17788 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
17789 HARD_REGNO_MODE_OK.
17790 * genpreds.c (write_insn_preds_c): Add an include of target.h.
17791 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
17792 instead of HARD_REGNO_MODE_OK.
17793 * caller-save.c: Include target.h.
17794 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
17795 HARD_REGNO_MODE_OK.
17796 * combine.c (can_combine_p): Likewise.
17797 (combinable_i3pat): Likewise.
17798 (can_change_dest_mode): Likewise.
17799 * expr.c (init_expr_target): Likewise.
17800 (convert_move): Likewise.
17801 (convert_modes): Likewise.
17802 * ira.c (setup_prohibited_class_mode_regs): Likewise.
17803 (setup_prohibited_mode_move_regs): Likewise.
17804 * ira.h (target_ira): Likewise.
17805 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17806 * lra-constraints.c (process_alt_operands): Likewise.
17807 (split_reg): Likewise.
17808 * recog.c (peep2_find_free_register): Likewise.
17809 * ree.c (combine_reaching_defs): Likewise.
17810 * regcprop.c (maybe_mode_change): Likewise.
17811 * reginfo.c (init_reg_sets_1): Likewise.
17812 (choose_hard_reg_mode): Likewise.
17813 (simplifiable_subregs): Likewise.
17814 * regrename.c (check_new_reg_p): Likewise.
17815 * reload.c (find_valid_class): Likewise.
17816 (find_valid_class_1): Likewise.
17817 (reload_inner_reg_of_subreg): Likewise.
17818 (push_reload): Likewise.
17819 (combine_reloads): Likewise.
17820 (find_dummy_reload): Likewise.
17821 (find_reloads): Likewise.
17822 * reload1.c (find_reg): Likewise.
17823 (set_reload_reg): Likewise.
17824 (allocate_reload_reg): Likewise.
17825 (choose_reload_regs): Likewise.
17826 (reload_adjust_reg_for_temp): Likewise.
17827 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
17828 (simplify_subreg_regno): Likewise.
17829 * sel-sched.c (init_regs_for_mode): Likewise.
17830 * varasm.c (make_decl_rtl): Likewise.
17831 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
17832 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
17833 HARD_REGNO_MODE_OK.
17834 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
17835 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
17836 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17837 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
17838 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
17839 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17840 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
17841 (arc_mode_class): Delete.
17842 (HARD_REGNO_MODE_OK): Delete.
17843 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17844 (arc_hard_regno_mode_ok): Rename old array to...
17845 (arc_hard_regno_mode_ok_modes): ...this.
17846 (arc_conditional_register_usage): Update accordingly.
17847 (arc_mode_class): Make static.
17848 (arc_hard_regno_mode_ok): New function.
17849 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17850 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17851 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17852 (arm_hard_regno_mode_ok): Make static.
17853 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17854 HARD_REGNO_MODE_OK.
17855 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17856 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17857 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17859 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17860 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17861 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17862 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17863 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17864 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17865 * config/bfin/predicates.md (valid_reg_operand): Use
17866 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17867 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17868 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17869 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17870 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17871 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17872 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17873 (cr16_hard_regno_mode_ok): Make static and return a bool.
17874 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17875 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17876 (cris_hard_regno_mode_ok): New function.
17877 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17878 (epiphany_mode_class): Delete.
17879 (HARD_REGNO_MODE_OK): Delete.
17880 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17881 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17882 (hard_regno_mode_ok): Rename to...
17883 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17884 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17885 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17886 HARD_REGNO_MODE_OK.
17887 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17888 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17889 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17890 (frv_hard_regno_mode_ok): Make static and return a bool.
17891 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17892 HARD_REGNO_MODE_OK.
17893 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17894 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17895 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17896 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17898 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17899 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17900 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17901 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17903 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17904 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17905 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17906 (ia64_hard_regno_mode_ok): New function.
17907 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17908 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17909 (iq2000_hard_regno_mode_ok): New function.
17910 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17911 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17912 (lm32_hard_regno_mode_ok): New function.
17913 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17914 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17915 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17916 instead of HARD_REGNO_MODE_OK.
17917 (m32c_hard_regno_ok): Rename to...
17918 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17919 (m32c_cannot_change_mode_class): Update accordingly.
17920 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17921 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17922 (m32r_mode_class): Delete.
17923 (HARD_REGNO_MODE_OK): Delete.
17924 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17925 (m32r_hard_regno_mode_ok): Rename to...
17926 (m32r_hard_regno_modes): ...this.
17927 (m32r_mode_class): Make static.
17928 (m32r_hard_regno_mode_ok): New function.
17929 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17930 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17931 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17932 (m68k_hard_regno_mode_ok): Make static.
17933 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17934 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17935 (mcore_hard_regno_mode_ok): New function.
17936 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17937 (HARD_REGNO_MODE_OK): Delete.
17938 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17940 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17941 (microblaze_hard_regno_mode_ok): New function.
17942 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17943 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17944 (mips_hard_regno_mode_ok): Delete.
17945 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17946 (mips_hard_regno_mode_ok_p): ...this and make static.
17947 (mips_hard_regno_mode_ok_p): Rename to...
17948 (mips_hard_regno_mode_ok_uncached): ...this.
17949 (mips_hard_regno_mode_ok): New function.
17950 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17951 of HARD_REGNO_MODE_OK.
17952 (mips_option_override): Update after above name changes.
17953 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17954 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17955 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17956 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17957 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17958 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17959 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17960 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17961 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17962 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17963 (msp430_hard_regno_mode_ok): Make static and return a bool.
17964 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17965 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17966 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17968 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17969 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17970 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17971 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17972 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17973 (PA_HARD_REGNO_MODE_OK): ...this
17974 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17975 (PA_HARD_REGNO_MODE_OK): ...this.
17976 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17977 (pa_hard_regno_mode_ok): New function.
17978 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17979 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17980 (pdp11_hard_regno_mode_ok): New function.
17981 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17982 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17984 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17986 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17987 (rs6000_hard_regno_mode_ok): Rename to...
17988 (rs6000_hard_regno_mode_ok_uncached): ...this.
17989 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17990 (rs6000_hard_regno_mode_ok): New function.
17991 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17992 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17993 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17994 (riscv_hard_regno_mode_ok): ...this and make static.
17995 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17996 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17997 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17998 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17999 (rl78_hard_regno_mode_ok): Make static and return bool.
18000 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
18001 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
18003 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
18004 (TARGET_HARD_REGNO_MODE_OK): Redefine.
18005 (rs6000_hard_regno_mode_ok): Rename to...
18006 (rs6000_hard_regno_mode_ok_uncached): ...this.
18007 (rs6000_init_hard_regno_mode_ok): Update accordingly.
18008 (rs6000_hard_regno_mode_ok): New function.
18009 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
18010 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
18011 (TARGET_HARD_REGNO_MODE_OK): Redefine.
18012 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
18013 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
18014 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
18015 (TARGET_HARD_REGNO_MODE_OK): Redefine.
18016 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
18017 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
18018 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
18019 (sh_hard_regno_mode_ok): Make static.
18020 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
18021 instead of HARD_REGNO_MODE_OK.
18022 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
18023 (sparc_mode_class): Delete.
18024 (HARD_REGNO_MODE_OK): Delete.
18025 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
18026 (hard_regno_mode_classes): Make static.
18027 (sparc_mode_class): Likewise.
18028 (sparc_hard_regno_mode_ok): New function.
18029 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
18030 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
18031 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
18033 (TARGET_HARD_REGNO_MODE_OK): Redefine.
18034 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
18035 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
18036 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
18037 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
18038 (TARGET_HARD_REGNO_MODE_OK): Redefine.
18039 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
18040 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
18041 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
18042 (visium_hard_regno_mode_ok): New function.
18043 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
18044 instead of HARD_REGNO_MODE_OK.
18045 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
18046 (HARD_REGNO_MODE_OK): Delete.
18047 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
18048 (xtensa_hard_regno_mode_ok_p): ...this and make static.
18049 (xtensa_option_override): Update accordingly.
18050 (TARGET_HARD_REGNO_MODE_OK): Redefine.
18051 (xtensa_hard_regno_mode_ok): New function.
18052 * system.h (HARD_REGNO_MODE_OK): Poison.
18054 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
18055 Alan Hayward <alan.hayward@arm.com>
18056 David Sherwood <david.sherwood@arm.com>
18058 * target.def (hard_regno_call_part_clobbered): New hook.
18059 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
18060 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
18061 * doc/tm.texi: Regenerate.
18062 * hooks.h (hook_bool_uint_mode_false): Declare.
18063 * hooks.c (hook_bool_uint_mode_false): New function.
18064 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18065 * cselib.c (cselib_process_insn): Use
18066 targetm.hard_regno_call_part_clobbered instead of
18067 HARD_REGNO_CALL_PART_CLOBBERED.
18068 * ira-conflicts.c (ira_build_conflicts): Likewise.
18069 * ira-costs.c (ira_tune_allocno_costs): Likewise.
18070 * lra-constraints.c (need_for_call_save_p): Likewise.
18071 * lra-lives.c: Include target.h.
18072 (check_pseudos_live_through_calls): Use
18073 targetm.hard_regno_call_part_clobbered instead of
18074 HARD_REGNO_CALL_PART_CLOBBERED.
18075 * regcprop.c: Include target.h.
18076 (copyprop_hardreg_forward_1): Use
18077 targetm.hard_regno_call_part_clobbered instead of
18078 HARD_REGNO_CALL_PART_CLOBBERED.
18079 * reginfo.c (choose_hard_reg_mode): Likewise.
18080 * regrename.c (check_new_reg_p): Likewise.
18081 * reload.c (find_equiv_reg): Likewise.
18082 * reload1.c (emit_reload_insns): Likewise.
18083 * sched-deps.c (deps_analyze_insn): Likewise.
18084 * sel-sched.c (init_regs_for_mode): Likewise.
18085 (mark_unavailable_hard_regs): Likewise.
18086 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
18087 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18088 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
18090 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18091 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18092 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
18094 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
18096 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18097 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18098 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
18100 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18101 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18102 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
18104 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18105 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
18107 * config/powerpcspe/powerpcspe.c
18108 (rs6000_hard_regno_call_part_clobbered): New function.
18109 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18110 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18111 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
18113 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18114 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18115 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
18117 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18118 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18119 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
18121 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
18122 Alan Hayward <alan.hayward@arm.com>
18123 David Sherwood <david.sherwood@arm.com>
18125 * rtl.h (subreg_memory_offset): Declare.
18126 * emit-rtl.c (subreg_memory_offset): New function.
18127 * expmed.c (store_bit_field_1): Use it.
18128 * expr.c (undefined_operand_subword_p): Likewise.
18129 * simplify-rtx.c (simplify_subreg): Likewise.
18131 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
18133 PR rtl-optimization/57448
18136 * optabs.c (expand_atomic_load): Place compiler memory barriers if
18137 using atomic_load pattern.
18138 (expand_atomic_store): Likewise.
18140 2017-09-04 Jakub Jelinek <jakub@redhat.com>
18143 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
18144 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
18145 handling. Use replace_call_with_value with NULL instead of
18146 gsi_replace, unlink_stmt_vdef and release_defs.
18148 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
18151 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
18153 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
18156 * rtl.h (emit_library_call_value_1): Declare.
18157 (emit_library_call): Replace declaration with a series of overloads.
18158 Remove the parameter count argument.
18159 (emit_library_call_value): Likewise.
18160 * calls.c (emit_library_call_value_1): Make global. Replace varargs
18161 with an "rtx_mode_t *".
18162 (emit_library_call_value): Delete.
18163 (emit_library_call): Likewise.
18164 * asan.c (asan_emit_stack_protection): Update calls accordingly.
18165 (asan_emit_allocas_unpoison): Likewise.
18166 * builtins.c (expand_builtin_powi): Likewise.
18167 (expand_asan_emit_allocas_unpoison): Likewise.
18168 * cfgexpand.c (expand_main_function): Likewise.
18169 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
18170 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
18171 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
18172 * config/arm/arm.c (arm_trampoline_init): Likewise.
18173 (arm_call_tls_get_addr): Likewise.
18174 (arm_expand_divmod_libfunc): Likewise.
18175 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
18176 (smulsi3_highpart): Likewise.
18177 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
18178 (c6x_expand_compare): Likewise.
18179 (c6x_expand_movmem): Likewise.
18180 * config/frv/frv.c (frv_trampoline_init): Likewise.
18181 * config/i386/i386.c (ix86_trampoline_init): Likewise.
18182 (ix86_expand_divmod_libfunc): Likewise.
18183 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
18184 (ia64_expand_compare): Likewise.
18185 (ia64_profile_hook): Likewise.
18186 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
18187 (nonlocal_goto): Likewise.
18188 (restore_stack_nonlocal): Likewise.
18189 * config/m32r/m32r.c (block_move_call): Likewise.
18190 (m32r_trampoline_init): Likewise.
18191 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
18192 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
18193 (m68k_call_m68k_read_tp): Likewise.
18194 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
18195 (microblaze_expand_divide): Likewise.
18196 * config/mips/mips.h (mips_args): Likewise.
18197 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
18198 (MIPS_ICACHE_SYNC): Likewise.
18199 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
18200 (nios2_trampoline_init): Likewise.
18201 * config/pa/pa.c (hppa_tls_call): Likewise.
18202 (pa_trampoline_init): Likewise.
18203 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
18204 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
18205 (expand_strn_compare): Likewise.
18206 (rs6000_generate_compare): Likewise.
18207 (rs6000_expand_float128_convert): Likewise.
18208 (output_profile_hook): Likewise.
18209 (rs6000_trampoline_init): Likewise.
18210 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
18211 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
18212 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
18213 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
18214 (rs6000_generate_compare): Likewise.
18215 (rs6000_expand_float128_convert): Likewise.
18216 (output_profile_hook): Likewise.
18217 (rs6000_trampoline_init): Likewise.
18218 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
18219 * config/sh/sh.c (sh_trampoline_init): Likewise.
18220 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
18221 (sparc_emit_float_lib_cmp): Likewise.
18222 (sparc32_initialize_trampoline): Likewise.
18223 (sparc64_initialize_trampoline): Likewise.
18224 (sparc_profile_hook): Likewise.
18225 * config/spu/spu.c (ea_load_store): Likewise.
18226 * config/spu/spu.md (floatunssidf2): Likewise.
18227 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
18228 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
18229 * config/visium/visium.c (expand_block_move_4): Likewise.
18230 (expand_block_move_2): Likewise.
18231 (expand_block_move_1): Likewise.
18232 (expand_block_set_4): Likewise.
18233 (expand_block_set_2): Likewise.
18234 (expand_block_set_1): Likewise.
18235 (visium_trampoline_init): Likewise.
18236 (visium_profile_hook): Likewise.
18237 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
18238 (xtensa_setup_frame_addresses): Likewise.
18239 (xtensa_trampoline_init): Likewise.
18240 * except.c (sjlj_emit_function_enter): Likewise.
18241 (sjlj_emit_function_exit): Likewise.
18242 * explow.c (allocate_dynamic_stack_space): Likewise.
18243 (probe_stack_range): Likewise.
18244 * expr.c (convert_mode_scalar): Likewise.
18245 * optabs.c (expand_binop): Likewise.
18246 (expand_twoval_binop_libfunc): Likewise.
18247 (expand_unop): Likewise.
18248 (prepare_cmp_insn): Likewise.
18249 (prepare_float_lib_cmp): Likewise.
18250 (expand_float): Likewise.
18251 (expand_fix): Likewise.
18252 (expand_fixed_convert): Likewise.
18253 (maybe_emit_sync_lock_test_and_set): Likewise.
18254 (expand_atomic_compare_and_swap): Likewise.
18255 (expand_mem_thread_fence): Likewise.
18256 (expand_atomic_fetch_op): Likewise.
18258 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
18260 * doc/generic.texi (OpenACC): Adjust URL.
18261 * doc/invoke.texi (C Dialect Options): Ditto.
18263 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
18265 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
18266 predicate for operand 1. Add (m,<S>) constraint.
18267 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
18268 Prevent memory operand 1 with register operand 2.
18270 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
18272 PR rtl-optimization/82024
18273 * combine.c (try_combine): If the combination result is a PARALLEL,
18274 and we only need to retain the SET in there that would be placed
18275 at I2, check that we can place that at I3 instead, before doing so.
18277 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18280 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
18282 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
18283 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
18286 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
18287 Jeff Law <law@redhat.com>
18289 * varasm.c (bss_initializer_p): Do not put constants into .bss
18290 (categorize_decl_for_section): Handle bss_initializer_p returning
18291 false when DECL_INITIAL is NULL.
18293 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18296 * config/s390/s390.c (s390_can_inline_p): New function.
18298 2017-09-01 Jeff Law <law@redhat.com>
18300 PR tree-optimization/82052
18301 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18302 Always initialize the returned slot after a hash table miss
18303 when INSERT is true.
18305 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
18307 * config/s390/s390.md (mem_signal_fence): Remove.
18308 * doc/md.texi (mem_signal_fence): Remove.
18309 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
18311 * target-insns.def (mem_signal_fence): Remove.
18313 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18316 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
18319 * asan.c (create_odr_indicator): Strip name encoding from assembler
18320 name before appending it after __odr_asan_.
18322 2017-09-01 Martin Liska <mliska@suse.cz>
18324 PR tree-optimization/82059
18325 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
18326 frequency only when an edge is redirected.
18328 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18330 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
18331 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
18332 (arc_conditional_register_usage): Remove ARC600 lp_count
18334 (arc_file_start): Emit Tag_ARC_CPU_variation.
18335 (arc_can_use_doloop_p): New conditions to use ZOLs.
18336 (hwloop_fail): New function.
18337 (hwloop_optimize): Likewise.
18338 (hwloop_pattern_reg): Likewise.
18339 (arc_doloop_hooks): New struct, to be used with reorg_loops.
18340 (arc_reorg_loops): New function, calls reorg_loops.
18341 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
18342 (arc600_corereg_hazard): Remove ZOL checking, case handled by
18344 (arc_loop_hazard): Remove function, functionality moved into
18346 (arc_hazard): Remove arc_loop_hazard call.
18347 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
18348 into hwloop_optimize.
18349 (arc_label_align): Remove ZOL handling.
18350 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
18351 * config/arc/arc.md (doloop_begin): Remove pattern.
18352 (doloop_begin_i): Likewise.
18353 (doloop_end_i): Likewise.
18354 (doloop_fallback): Likewise.
18355 (doloop_fallback_m): Likewise.
18356 (doloop_end): Reimplement expand.
18357 (arc_lp): New pattern for LP instruction.
18358 (loop_end): New pattern.
18359 (loop_fail): Likewise.
18360 (decrement_and_branch_until_zero): Likewise.
18361 * config/arc/arc.opt (mlpc-width): New option.
18362 * doc/invoke.texi (mlpc-width): Document option.
18364 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18366 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
18367 (arc_ccfsm_advance): Fix checking for delay slots.
18368 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
18370 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18372 * config/arc/arc.md (movqi_insn): Add stores to save constant long
18374 (movhi_insn): Update store instruction constraint which are saving
18375 6-bit short immediates.
18376 (movsi_insn): Consider also short scaled load operations.
18377 (zero_extendhisi2_i): Use Usd constraint instead of T.
18378 (extendhisi2_i): Add q constraint.
18379 (arc_clzsi2): Add type and length attributes.
18380 (arc_ctzsi2): Likewise.
18381 * config/arc/constraints.md (Usc): Update constraint, the
18382 assembler can parse two relocations for a single instruction.
18384 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18386 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
18387 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
18389 2017-08-31 Olivier Hainque <hainque@adacore.com>
18391 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
18392 match as powerpc-wrs-vxworks*.
18394 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
18396 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
18397 register constraint for by-element operand.
18398 (aarch64_mls_elt_merge<mode>): Likewise.
18400 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18402 * config/arc/arc.c (arc_can_follow_jump): Check for short
18405 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18407 * config.gcc: Use g.opt for arc.
18408 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
18409 functionality moved to ...
18410 (legitimate_scaled_address_p): New function, ...here.
18411 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
18412 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
18413 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
18415 (arc_override_options): Handle G option.
18416 (arc_output_pic_addr_const): Correct function definition.
18417 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
18418 (arc_decl_anon_ns_mem_p): Delete.
18419 (arc_in_small_data_p): Overhaul this function to take into
18420 consideration the value given via G option.
18421 (arc_rewrite_small_data_1): Renamed and corrected old
18422 arc_rewrite_small_data function.
18423 (arc_rewrite_small_data): New function.
18424 (small_data_pattern): Don't use pic_offset_table_rtx.
18425 * config/arc/arc.h (CC1_SPEC): Recognize G option.
18426 * config/arc/simdext.md (movmisalignv2hi): Use
18427 prepare_move_operands function.
18429 (movmisalign*): Likewise.
18430 * doc/invoke.texi (ARC options): Document -G option.
18432 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18434 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
18436 * config/arc/arc.c (arc_print_operand): Output scalled address for
18437 sdata whenever is possible.
18438 (arc_in_small_data_p): Allow sdata for 64bit datum when double
18439 load/stores are available.
18440 (compact_sda_memory_operand): Check for the alignment required by
18441 code density instructions.
18442 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
18444 * config/arc/constraints.md (Usd): Update constraint.
18445 (Us0): New constraint.
18446 (Usc): Update constraint.
18448 2017-08-31 Richard Biener <rguenther@suse.de>
18450 PR middle-end/82054
18451 * dwarf2out.c (dwarf2out_early_global_decl): Process each
18452 function only once.
18454 2017-08-31 Tamar Christina <tamar.christina@arm.com>
18456 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
18457 Resize type_signature.
18459 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18460 Alan Hayward <alan.hayward@arm.com>
18461 David Sherwood <david.sherwood@arm.com>
18463 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
18464 subregs whose inner modes can be stored in GPRs.
18465 (aarch64_classify_index): Likewise.
18467 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18468 Alan Hayward <alan.hayward@arm.com>
18469 David Sherwood <david.sherwood@arm.com>
18471 * config/aarch64/iterators.md (V_cmp_result): Rename to...
18472 (V_INT_EQUIV): ...this.
18473 (v_cmp_result): Rename to...
18474 (v_int_equiv): ...this.
18475 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
18476 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
18477 (copysign<mode>3): Likewise.
18478 (aarch64_simd_bsl<mode>_internal): Likewise.
18479 (aarch64_simd_bsl<mode>): Likewise.
18480 (vec_cmp<mode><mode>): Likewise.
18481 (vcond<mode><mode>): Likewise.
18482 (vcond<v_cmp_mixed><mode>): Likewise.
18483 (vcondu<mode><v_cmp_mixed>): Likewise.
18484 (aarch64_cm<optab><mode>): Likewise.
18485 (aarch64_cmtst<mode>): Likewise.
18486 (aarch64_fac<optab><mode>): Likewise.
18487 (vec_perm_const<mode>): Likewise.
18488 (vcond_mask_<mode><v_cmp_result>): Rename to...
18489 (vcond_mask_<mode><v_int_equiv>): ...this.
18490 (vec_cmp<mode><v_cmp_result>): Rename to...
18491 (vec_cmp<mode><v_int_equiv>): ...this.
18493 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18494 Alan Hayward <alan.hayward@arm.com>
18495 David Sherwood <david.sherwood@arm.com>
18497 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
18499 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
18500 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
18501 (UNSPEC_LD4_DREG): New unspecs.
18502 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
18503 (aarch64_ld2<mode>_dreg_be): Replace with...
18504 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
18506 (aarch64_ld3<mode>_dreg_le)
18507 (aarch64_ld3<mode>_dreg_be): Replace with...
18508 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
18510 (aarch64_ld4<mode>_dreg_le)
18511 (aarch64_ld4<mode>_dreg_be): Replace with...
18512 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
18515 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18517 PR tree-optimization/81987
18518 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
18519 insert an initializer in a location not dominated by the stride
18522 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18524 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
18525 on the entire header of the finally block in the fallthru case.
18527 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18529 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
18531 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
18533 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
18534 rs6000_emit_move_from_cr and call renamed function.
18535 (rs6000_emit_prologue): Call renamed functions.
18536 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
18537 movesi_from_cr, remove volatile CRs.
18539 2017-08-30 Jon Beniston <jon@beniston.com>
18540 Richard Biener <rguenther@suse.de>
18542 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
18543 of VECTOR_MODE_P check.
18544 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
18545 element vector types.
18547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18549 * df.h (df_read_modify_subreg_p): Remove in favor of...
18550 * rtl.h (read_modify_subreg_p): ...this new function. Take a
18551 const_rtx instead of an rtx.
18552 * cprop.c (local_cprop_find_used_regs): Update accordingly.
18553 * df-problems.c (df_word_lr_mark_ref): Likewise.
18554 * ira-lives.c (mark_pseudo_reg_live): Likewise.
18555 (mark_pseudo_reg_dead): Likewise.
18556 (mark_ref_dead): Likewise.
18557 * reginfo.c (init_subregs_of_mode): Likewise.
18558 * sched-deps.c (sched_analyze_1): Likewise.
18559 * df-scan.c (df_def_record_1): Likewise.
18560 (df_uses_record): Likewise.
18561 (df_read_modify_subreg_p): Remove in favor of...
18562 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
18563 const_rtx instead of an rtx.
18565 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18566 Alan Hayward <alan.hayward@arm.com>
18567 David Sherwood <david.sherwood@arm.com>
18569 * rtl.h (partial_subreg_p): New function.
18570 * caller-save.c (save_call_clobbered_regs): Use it.
18571 * calls.c (expand_call): Likewise.
18572 * combine.c (combinable_i3pat): Likewise.
18573 (simplify_set): Likewise.
18574 (make_extraction): Likewise.
18575 (make_compound_operation_int): Likewise.
18576 (gen_lowpart_or_truncate): Likewise.
18577 (force_to_mode): Likewise.
18578 (make_field_assignment): Likewise.
18579 (reg_truncated_to_mode): Likewise.
18580 (record_truncated_value): Likewise.
18581 (move_deaths): Likewise.
18582 * cse.c (record_jump_cond): Likewise.
18583 (cse_insn): Likewise.
18584 * cselib.c (cselib_lookup_1): Likewise.
18585 * expmed.c (extract_bit_field_using_extv): Likewise.
18586 * function.c (assign_parm_setup_reg): Likewise.
18587 * ifcvt.c (noce_convert_multiple_sets): Likewise.
18588 * ira-build.c (create_insn_allocnos): Likewise.
18589 * lra-coalesce.c (merge_pseudos): Likewise.
18590 * lra-constraints.c (match_reload): Likewise.
18591 (simplify_operand_subreg): Likewise.
18592 (curr_insn_transform): Likewise.
18593 * lra-lives.c (process_bb_lives): Likewise.
18594 * lra.c (new_insn_reg): Likewise.
18595 (lra_substitute_pseudo): Likewise.
18596 * regcprop.c (mode_change_ok): Likewise.
18597 (maybe_mode_change): Likewise.
18598 (copyprop_hardreg_forward_1): Likewise.
18599 * reload.c (push_reload): Likewise.
18600 (find_reloads): Likewise.
18601 (find_reloads_subreg_address): Likewise.
18602 * reload1.c (alter_reg): Likewise.
18603 (eliminate_regs_1): Likewise.
18604 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18606 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
18608 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
18609 back to if statements, including unpack.
18611 2017-08-30 Martin Liska <mliska@suse.cz>
18613 PR inline-asm/82001
18614 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
18616 (func_checker::compare_asm_inputs_outputs): ... this function.
18617 (func_checker::compare_gimple_asm): Use the function to compare
18618 also ASM constrains.
18619 * ipa-icf-gimple.h: Rename the function.
18621 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18622 Alan Hayward <alan.hayward@arm.com>
18623 David Sherwood <david.sherwood@arm.com>
18625 * coretypes.h (complex_mode): New type.
18626 * gdbhooks.py (build_pretty_printer): Handle it.
18627 * machmode.h (complex_mode): New class.
18628 (complex_mode::includes_p): New function.
18629 (is_complex_int_mode): Likewise.
18630 (is_complex_float_mode): Likewise.
18631 * genmodes.c (get_mode_class): Handle complex mode classes.
18632 * function.c (expand_function_end): Use is_complex_int_mode.
18634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18635 Alan Hayward <alan.hayward@arm.com>
18636 David Sherwood <david.sherwood@arm.com>
18638 * coretypes.h (scalar_mode_pod): New typedef.
18639 * gdbhooks.py (build_pretty_printer): Handle it.
18640 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
18641 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
18642 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
18643 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
18644 as_a <scalar_mode>.
18646 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18647 Alan Hayward <alan.hayward@arm.com>
18648 David Sherwood <david.sherwood@arm.com>
18650 * machmode.h (mode_for_vector): Take a scalar_mode instead
18652 * stor-layout.c (mode_for_vector): Likewise.
18653 * explow.c (promote_mode): Use as_a <scalar_mode>.
18654 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
18656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18657 Alan Hayward <alan.hayward@arm.com>
18658 David Sherwood <david.sherwood@arm.com>
18660 * target.def (preferred_simd_mode): Take a scalar_mode
18661 instead of a machine_mode.
18662 * targhooks.h (default_preferred_simd_mode): Likewise.
18663 * targhooks.c (default_preferred_simd_mode): Likewise.
18664 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
18665 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
18666 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
18667 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
18668 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
18669 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
18670 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
18671 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
18673 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
18674 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
18675 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
18676 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
18677 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
18678 * doc/tm.texi: Regenerate.
18679 * optabs-query.c (can_vec_mask_load_store_p): Return false for
18682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18683 Alan Hayward <alan.hayward@arm.com>
18684 David Sherwood <david.sherwood@arm.com>
18686 * target.def (scalar_mode_supported_p): Take a scalar_mode
18687 instead of a machine_mode.
18688 * targhooks.h (default_scalar_mode_supported_p): Likewise.
18689 * targhooks.c (default_scalar_mode_supported_p): Likewise.
18690 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
18691 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18692 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
18693 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
18694 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
18695 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
18696 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
18697 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
18698 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
18699 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
18700 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
18701 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
18703 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
18704 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
18705 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18706 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18707 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
18709 * doc/tm.texi: Regenerate.
18711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18712 Alan Hayward <alan.hayward@arm.com>
18713 David Sherwood <david.sherwood@arm.com>
18715 * coretypes.h (opt_scalar_mode): New typedef.
18716 * gdbhooks.py (build_pretty_printers): Handle it.
18717 * machmode.h (mode_iterator::get_2xwider): Add overload for
18719 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
18721 * expr.c (convert_mode_scalar): Likewise.
18722 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18723 * optabs.c (expand_float): Likewise.
18724 (expand_fix): Likewise.
18725 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18727 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18728 Alan Hayward <alan.hayward@arm.com>
18729 David Sherwood <david.sherwood@arm.com>
18731 * optabs.c (expand_float): Explicitly check for scalars before
18732 using a branching expansion.
18733 (expand_fix): Likewise.
18735 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18736 Alan Hayward <alan.hayward@arm.com>
18737 David Sherwood <david.sherwood@arm.com>
18739 * expr.c (convert_mode): Split scalar handling out into...
18740 (convert_mode_scalar): ...this new function. Treat the modes
18743 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18744 Alan Hayward <alan.hayward@arm.com>
18745 David Sherwood <david.sherwood@arm.com>
18747 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
18749 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
18751 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18752 Alan Hayward <alan.hayward@arm.com>
18753 David Sherwood <david.sherwood@arm.com>
18755 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
18756 rather than a machine_mode.
18757 (fixed_from_string): Likewise.
18758 (fixed_convert): Likewise.
18759 (fixed_convert_from_int): Likewise.
18760 (fixed_convert_from_real): Likewise.
18761 (real_convert_from_fixed): Likewise.
18762 * fixed-value.c (fixed_from_double_int): Likewise.
18763 (fixed_from_string): Likewise.
18764 (fixed_convert): Likewise.
18765 (fixed_convert_from_int): Likewise.
18766 (fixed_convert_from_real): Likewise.
18767 (real_convert_from_fixed): Likewise.
18768 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
18770 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18771 Alan Hayward <alan.hayward@arm.com>
18772 David Sherwood <david.sherwood@arm.com>
18774 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
18775 of separate mode class checks. Do not allow vector modes here.
18776 (immed_wide_int_const): Use as_a <scalar_mode>.
18777 * explow.c (trunc_int_for_mode): Likewise.
18778 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
18779 (wi::shwi): Likewise.
18780 (wi::min_value): Likewise.
18781 (wi::max_value): Likewise.
18782 * dwarf2out.c (loc_descriptor): Likewise.
18783 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
18784 for CONST_WIDE_INT.
18786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18787 Alan Hayward <alan.hayward@arm.com>
18788 David Sherwood <david.sherwood@arm.com>
18790 * tree.h (SCALAR_TYPE_MODE): New macro.
18791 * expr.c (expand_expr_addr_expr_1): Use it.
18792 (expand_expr_real_2): Likewise.
18793 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
18794 (fold_convert_const_fixed_from_int): Likewise.
18795 (fold_convert_const_fixed_from_real): Likewise.
18796 (native_encode_fixed): Likewise
18797 (native_encode_complex): Likewise
18798 (native_encode_vector): Likewise.
18799 (native_interpret_fixed): Likewise.
18800 (native_interpret_real): Likewise.
18801 (native_interpret_complex): Likewise.
18802 (native_interpret_vector): Likewise.
18803 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
18804 (simd_clone_adjust_argument_types): Likewise.
18805 (simd_clone_init_simd_arrays): Likewise.
18806 (simd_clone_adjust): Likewise.
18807 * stor-layout.c (layout_type): Likewise.
18808 * tree.c (build_minus_one_cst): Likewise.
18809 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18810 * tree-inline.c (estimate_move_cost): Likewise.
18811 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
18812 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
18813 (vectorizable_reduction): Likewise.
18814 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
18815 (vect_recog_mixed_size_cond_pattern): Likewise.
18816 (check_bool_pattern): Likewise.
18817 (adjust_bool_pattern): Likewise.
18818 (search_type_for_mask_1): Likewise.
18819 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
18820 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18821 (vectorizable_load): Likewise.
18822 (vectorizable_store): Likewise.
18823 * ubsan.c (ubsan_encode_value): Likewise.
18824 * varasm.c (output_constant): Likewise.
18826 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18827 Alan Hayward <alan.hayward@arm.com>
18828 David Sherwood <david.sherwood@arm.com>
18830 * coretypes.h (scalar_mode): New class.
18831 * machmode.h (scalar_mode): Likewise.
18832 (scalar_mode::includes_p): New function.
18833 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
18834 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
18835 * genmodes.c (get_mode_class): Handle remaining scalar modes.
18836 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
18837 * expmed.c (store_bit_field_1): Likewise.
18838 (extract_bit_field_1): Likewise.
18839 * expr.c (write_complex_part): Likewise.
18840 (read_complex_part): Likewise.
18841 (emit_move_complex_push): Likewise.
18842 (expand_expr_real_2): Likewise.
18843 * function.c (assign_parm_setup_reg): Likewise.
18844 (assign_parms_unsplit_complex): Likewise.
18845 * optabs.c (expand_binop): Likewise.
18846 * rtlanal.c (subreg_get_info): Likewise.
18847 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18848 * varasm.c (output_constant_pool_2): Likewise.
18850 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18851 Alan Hayward <alan.hayward@arm.com>
18852 David Sherwood <david.sherwood@arm.com>
18854 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18856 (expmed_mult_highpart_optab): Likewise.
18857 (expmed_mult_highpart): Likewise.
18859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18860 Alan Hayward <alan.hayward@arm.com>
18861 David Sherwood <david.sherwood@arm.com>
18863 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18864 instead of a machine_mode.
18865 (builtin_memset_read_str): Likewise.
18866 * builtins.c (c_readstr): Likewise.
18867 (builtin_memcpy_read_str): Likewise.
18868 (builtin_strncpy_read_str): Likewise.
18869 (builtin_memset_read_str): Likewise.
18870 (builtin_memset_gen_str): Likewise.
18871 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18872 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18873 instead of a machine_mode.
18874 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18876 (make_extraction): Likewise.
18877 (try_widen_shift_mode): Take and return scalar_int_modes instead
18879 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18880 a scalar_int_mode instead of a machine_mode.
18881 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18882 (avr_addr_space_pointer_mode): Likewise.
18883 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18884 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18885 (msp430_unwind_word_mode): Likewise.
18886 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18887 (spu_addr_space_pointer_mode): Likewise.
18888 (spu_addr_space_address_mode): Likewise.
18889 (spu_libgcc_cmp_return_mode): Likewise.
18890 (spu_libgcc_shift_count_mode): Likewise.
18891 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18892 (rl78_addr_space_pointer_mode): Likewise.
18893 (fl78_unwind_word_mode): Likewise.
18894 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18896 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18897 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18898 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18899 (mips_valid_pointer_mode): Likewise.
18900 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18901 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18902 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18904 (ft32_addr_space_address_mode): Likewise.
18905 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18906 scalar_int_mode instead of a machine_mode.
18907 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18909 (m32c_addr_space_address_mode): Likewise.
18910 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18911 (rs6000_eh_return_filter_mode): Likewise.
18912 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18913 (rs6000_eh_return_filter_mode): Likewise.
18914 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18915 (s390_libgcc_shift_count_mode): Likewise.
18916 (s390_unwind_word_mode): Likewise.
18917 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18919 * target.def (mode_rep_extended): Likewise.
18920 (valid_pointer_mode): Likewise.
18921 (addr_space.valid_pointer_mode): Likewise.
18922 (eh_return_filter_mode): Return a scalar_int_mode rather than
18924 (libgcc_cmp_return_mode): Likewise.
18925 (libgcc_shift_count_mode): Likewise.
18926 (unwind_word_mode): Likewise.
18927 (addr_space.pointer_mode): Likewise.
18928 (addr_space.address_mode): Likewise.
18929 * doc/tm.texi: Regenerate.
18930 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18932 (do_jump): Use scalar_int_mode for local variables.
18933 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18934 rather than a machine_mode.
18935 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18936 (scompare_loc_descriptor_wide): Likewise.
18937 (scompare_loc_descriptor_narrow): Likewise.
18938 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18940 * except.c (sjlj_emit_dispatch_table): Likewise.
18941 (expand_builtin_eh_copy_values): Likewise.
18942 * explow.c (convert_memory_address_addr_space_1): Likewise.
18943 Take a scalar_int_mode rather than a machine_mode.
18944 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18945 than a machine_mode.
18946 (memory_address_addr_space): Use scalar_int_mode for local variables.
18947 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18948 rather than a machine_mode.
18949 * expmed.c (mask_rtx): Likewise.
18950 (init_expmed_one_conv): Likewise.
18951 (expand_mult_highpart_adjust): Likewise.
18952 (extract_high_half): Likewise.
18953 (expmed_mult_highpart_optab): Likewise.
18954 (expmed_mult_highpart): Likewise.
18955 (expand_smod_pow2): Likewise.
18956 (expand_sdiv_pow2): Likewise.
18957 (emit_store_flag_int): Likewise.
18958 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18960 (extract_low_bits): Likewise.
18961 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18963 * expr.c (pieces_addr::adjust): Likewise.
18964 (can_store_by_pieces): Likewise.
18965 (store_by_pieces): Likewise.
18966 (clear_by_pieces_1): Likewise.
18967 (expand_expr_addr_expr_1): Likewise.
18968 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18969 (expand_expr_real_1): Likewise.
18970 (try_casesi): Likewise.
18971 * final.c (shorten_branches): Likewise.
18972 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18973 type of "mode" to machine_mode.
18974 * internal-fn.c (expand_arith_overflow_result_store): Take a
18975 scalar_int_mode rather than a machine_mode.
18976 (expand_mul_overflow): Use scalar_int_mode for local variables.
18977 * loop-doloop.c (doloop_modify): Likewise.
18978 (doloop_optimize): Likewise.
18979 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18980 than a machine_mode.
18981 (expand_doubleword_shift_condmove): Likewise.
18982 (expand_doubleword_shift): Likewise.
18983 (expand_doubleword_clz): Likewise.
18984 (expand_doubleword_popcount): Likewise.
18985 (expand_doubleword_parity): Likewise.
18986 (expand_absneg_bit): Use scalar_int_mode for local variables.
18987 (prepare_float_lib_cmp): Likewise.
18988 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18989 rather than a machine_mode.
18990 (convert_memory_address_addr_space): Likewise.
18991 (get_mode_bounds): Likewise.
18992 (get_address_mode): Return a scalar_int_mode rather than a
18994 * rtlanal.c (get_address_mode): Likewise.
18995 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18996 than a machine_mode.
18997 * targhooks.c (default_mode_rep_extended): Likewise.
18998 (default_valid_pointer_mode): Likewise.
18999 (default_addr_space_valid_pointer_mode): Likewise.
19000 (default_eh_return_filter_mode): Return a scalar_int_mode rather
19001 than a machine_mode.
19002 (default_libgcc_cmp_return_mode): Likewise.
19003 (default_libgcc_shift_count_mode): Likewise.
19004 (default_unwind_word_mode): Likewise.
19005 (default_addr_space_pointer_mode): Likewise.
19006 (default_addr_space_address_mode): Likewise.
19007 * targhooks.h (default_eh_return_filter_mode): Likewise.
19008 (default_libgcc_cmp_return_mode): Likewise.
19009 (default_libgcc_shift_count_mode): Likewise.
19010 (default_unwind_word_mode): Likewise.
19011 (default_addr_space_pointer_mode): Likewise.
19012 (default_addr_space_address_mode): Likewise.
19013 (default_mode_rep_extended): Take a scalar_int_mode rather than
19015 (default_valid_pointer_mode): Likewise.
19016 (default_addr_space_valid_pointer_mode): Likewise.
19017 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
19019 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
19020 rather than a machine_mode.
19021 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
19022 for local variables.
19023 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19024 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
19025 than a machine_mode.
19027 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19028 Alan Hayward <alan.hayward@arm.com>
19029 David Sherwood <david.sherwood@arm.com>
19031 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
19032 the mode argument to scalar_int_mode.
19033 (do_jump_by_parts_zero_rtx): Likewise.
19034 (do_jump_by_parts_equality_rtx): Likewise.
19035 (do_jump_by_parts_greater): Take a mode argument.
19036 (do_jump_by_parts_equality): Likewise.
19037 (do_jump_1): Update calls accordingly.
19039 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19040 Alan Hayward <alan.hayward@arm.com>
19041 David Sherwood <david.sherwood@arm.com>
19043 * is-a.h (safe_dyn_cast): New function.
19044 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
19045 (jump_table_for_label): Likewise.
19046 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
19047 instead of an rtx_insn *.
19048 (shorten_branches): Use dyn_cast instead of LABEL_P and
19049 JUMP_TABLE_DATA_P. Use jump_table_for_label and
19050 rtx_jump_table_data::get_data_mode.
19051 (final_scan_insn): Likewise.
19053 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19054 Alan Hayward <alan.hayward@arm.com>
19055 David Sherwood <david.sherwood@arm.com>
19057 * combine.c (try_combine): Use is_a <scalar_int_mode> when
19058 trying to combine a full-register integer set with a subreg
19061 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19062 Alan Hayward <alan.hayward@arm.com>
19063 David Sherwood <david.sherwood@arm.com>
19065 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
19066 that is always either address_mode or pointer_mode.
19068 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19069 Alan Hayward <alan.hayward@arm.com>
19070 David Sherwood <david.sherwood@arm.com>
19072 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
19073 when the two are known to be equal.
19075 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19076 Alan Hayward <alan.hayward@arm.com>
19077 David Sherwood <david.sherwood@arm.com>
19079 * simplify-rtx.c (simplify_const_unary_operation): Use
19080 is_a <scalar_int_mode> instead of checking for a nonzero
19081 precision. Forcibly convert op_mode to a scalar_int_mode
19082 in that case. More clearly differentiate the operand and
19083 result modes and use the former when deciding what the value
19084 of a count-bits operation should be. Use is_int_mode instead
19085 of checking for a MODE_INT. Remove redundant check for whether
19086 this mode has a zero precision.
19088 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19089 Alan Hayward <alan.hayward@arm.com>
19090 David Sherwood <david.sherwood@arm.com>
19092 * optabs.c (widen_leading): Change the type of the mode argument
19093 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
19094 (widen_bswap): Likewise.
19095 (expand_parity): Likewise.
19096 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
19097 (expand_ffs): Likewise.
19098 (epand_unop): Check for scalar integer modes before calling the
19101 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19102 Alan Hayward <alan.hayward@arm.com>
19103 David Sherwood <david.sherwood@arm.com>
19105 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
19107 (expand_expr_real_1): Update call accordingly.
19109 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19110 Alan Hayward <alan.hayward@arm.com>
19111 David Sherwood <david.sherwood@arm.com>
19113 * expmed.c (store_bit_field_using_insv): Add op0_mode and
19114 value_mode arguments. Use scalar_int_mode internally.
19115 (store_bit_field_1): Rename the new integer mode from imode
19116 to op0_mode and use it instead of GET_MODE (op0). Update calls
19117 to store_split_bit_field, store_bit_field_using_insv and
19118 store_fixed_bit_field.
19119 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
19120 Use scalar_int_mode internally. Use a bit count rather than a mode
19121 when calculating the largest bit size for get_best_mode.
19122 Update calls to store_split_bit_field and store_fixed_bit_field_1.
19123 (store_fixed_bit_field_1): Add mode and value_mode arguments.
19124 Remove assertion that OP0 has a scalar integer mode.
19125 (store_split_bit_field): Add op0_mode and value_mode arguments.
19126 Update calls to extract_fixed_bit_field.
19127 (extract_bit_field_using_extv): Add an op0_mode argument.
19128 Use scalar_int_mode internally.
19129 (extract_bit_field_1): Rename the new integer mode from imode to
19130 op0_mode and use it instead of GET_MODE (op0). Update calls to
19131 extract_split_bit_field, extract_bit_field_using_extv and
19132 extract_fixed_bit_field.
19133 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
19134 to extract_split_bit_field and extract_fixed_bit_field_1.
19135 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
19136 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
19137 on the target mode.
19138 (extract_split_bit_field): Add an op0_mode argument. Update call
19139 to extract_fixed_bit_field.
19141 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19142 Alan Hayward <alan.hayward@arm.com>
19143 David Sherwood <david.sherwood@arm.com>
19145 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
19146 * explow.c (hard_function_value): Likewise.
19147 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
19148 convert_to_mode call outside the loop.
19149 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
19150 for the mode iterator. Require the mode specified by max_pieces
19152 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
19154 (copy_blkmode_to_reg): Likewise.
19155 (set_storage_via_setmem): Likewise.
19156 * optabs.c (prepare_cmp_insn): Likewise.
19157 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19158 * stor-layout.c (finish_bitfield_representative): Likewise.
19160 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19161 Alan Hayward <alan.hayward@arm.com>
19162 David Sherwood <david.sherwood@arm.com>
19164 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
19165 * expr.c (convert_move): Use them.
19166 (convert_modes): Likewise.
19167 (store_expr_with_bounds): Likewise.
19169 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19170 Alan Hayward <alan.hayward@arm.com>
19171 David Sherwood <david.sherwood@arm.com>
19173 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
19174 parameter for the mode of "x". Remove the "known_x", "known_mode"
19175 and "known_ret" arguments. Change the type of the mode argument
19176 to scalar_int_mode.
19177 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
19178 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
19179 (reg_num_sign_bit_copies_for_combine): Likewise.
19180 * rtlanal.c (nonzero_bits1): Likewise.
19181 (num_sign_bit_copies1): Likewise.
19182 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
19183 (reg_num_sign_bit_copies_general): Likewise.
19184 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
19185 (reg_nonzero_bits_general): Likewise.
19187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19188 Alan Hayward <alan.hayward@arm.com>
19189 David Sherwood <david.sherwood@arm.com>
19191 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
19192 than in subroutines. Return 1 for non-integer modes.
19193 (cached_num_sign_bit_copies): Change the type of the mode parameter
19194 to scalar_int_mode.
19195 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
19196 classes. Handle CONST_INT_P first and then check whether X also
19197 has a scalar integer mode. Check the same thing for inner registers
19198 of a SUBREG and for values that are being extended or truncated.
19200 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19201 Alan Hayward <alan.hayward@arm.com>
19202 David Sherwood <david.sherwood@arm.com>
19204 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
19205 in subroutines. Return the mode mask for non-integer modes.
19206 (cached_nonzero_bits): Change the type of the mode parameter
19207 to scalar_int_mode.
19208 (nonzero_bits1): Likewise. Remove early exit for other mode
19209 classes. Handle CONST_INT_P first and then check whether X
19210 also has a scalar integer mode.
19212 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19213 Alan Hayward <alan.hayward@arm.com>
19214 David Sherwood <david.sherwood@arm.com>
19216 * expr.c (widest_int_mode_for_size): Make the comment match the code.
19217 Return a scalar_int_mode and assert that the size is greater than
19219 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
19220 (op_by_pieces_d::op_by_pieces_d): Likewise.
19221 (op_by_pieces_d::run): Likewise.
19222 (can_store_by_pieces): Likewise.
19224 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19225 Alan Hayward <alan.hayward@arm.com>
19226 David Sherwood <david.sherwood@arm.com>
19228 * combine.c (extract_left_shift): Add a mode argument and update
19230 (make_compound_operation_int): Change the type of the mode parameter
19231 to scalar_int_mode and update the call to extract_left_shift.
19233 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19234 Alan Hayward <alan.hayward@arm.com>
19235 David Sherwood <david.sherwood@arm.com>
19237 * combine.c (simplify_and_const_int): Change the type of the mode
19238 parameter to scalar_int_mode.
19239 (simplify_and_const_int_1): Likewise. Update recursive call.
19241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19242 Alan Hayward <alan.hayward@arm.com>
19243 David Sherwood <david.sherwood@arm.com>
19245 * combine.c (simplify_compare_const): Check that the mode is a
19246 scalar_int_mode (rather than VOIDmode) before testing its
19248 (simplify_comparison): Move COMPARISON_P handling out of the
19249 loop and restrict the latter part of the loop to scalar_int_modes.
19250 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
19251 and when considering SUBREG_REGs. Use is_int_mode instead of
19252 checking GET_MODE_CLASS against MODE_INT.
19254 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19255 Alan Hayward <alan.hayward@arm.com>
19256 David Sherwood <david.sherwood@arm.com>
19258 * combine.c (try_widen_shift_mode): Move check for equal modes to...
19259 (simplify_shift_const_1): ...here. Use scalar_int_mode for
19260 shift_unit_mode and for modes involved in scalar shifts.
19262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19263 Alan Hayward <alan.hayward@arm.com>
19264 David Sherwood <david.sherwood@arm.com>
19266 * combine.c (force_int_to_mode): New function, split out from...
19267 (force_to_mode): ...here. Keep xmode up-to-date and use it
19268 instead of GET_MODE (x).
19270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19271 Alan Hayward <alan.hayward@arm.com>
19272 David Sherwood <david.sherwood@arm.com>
19274 * optabs-query.h (extraction_insn::struct_mode): Change type to
19275 opt_scalar_int_mode and update comment.
19276 (extraction_insn::field_mode): Change type to scalar_int_mode.
19277 (extraction_insn::pos_mode): Likewise.
19278 * combine.c (make_extraction): Update accordingly.
19279 * optabs-query.c (get_traditional_extraction_insn): Likewise.
19280 (get_optab_extraction_insn): Likewise.
19281 * recog.c (simplify_while_replacing): Likewise.
19282 * expmed.c (narrow_bit_field_mem): Change the type of the mode
19283 parameter to opt_scalar_int_mode.
19285 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19286 Alan Hayward <alan.hayward@arm.com>
19287 David Sherwood <david.sherwood@arm.com>
19289 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
19290 to a scalar_int_mode instead of a machine_mode.
19291 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
19292 (get_best_mode): Return a boolean and use a pointer argument to store
19293 the selected mode. Replace the limit mode parameter with a bit limit.
19294 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
19295 for the values returned by bit_field_mode_iterator::next_mode.
19296 (store_bit_field): Update call to get_best_mode.
19297 (store_fixed_bit_field): Likewise.
19298 (extract_fixed_bit_field): Likewise.
19299 * expr.c (optimize_bitfield_assignment_op): Likewise.
19300 * fold-const.c (optimize_bit_field_compare): Likewise.
19301 (fold_truth_andor_1): Likewise.
19302 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
19303 Update for new type of m_mode.
19304 (get_best_mode): As above.
19306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19307 Alan Hayward <alan.hayward@arm.com>
19308 David Sherwood <david.sherwood@arm.com>
19310 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
19311 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
19312 (store_bit_field): Check is_a <scalar_int_mode> before calling
19313 strict_volatile_bitfield_p.
19314 (extract_bit_field): Likewise.
19316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19317 Alan Hayward <alan.hayward@arm.com>
19318 David Sherwood <david.sherwood@arm.com>
19320 * target.def (cstore_mode): Return a scalar_int_mode.
19321 * doc/tm.texi: Regenerate.
19322 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
19323 * targhooks.h (default_cstore_mode): Likewise.
19324 * targhooks.c (default_cstore_mode): Likewise, using a forced
19326 * expmed.c (emit_cstore): Expect the target of the cstore to be
19329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19330 Alan Hayward <alan.hayward@arm.com>
19331 David Sherwood <david.sherwood@arm.com>
19333 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
19335 (niter_desc): Likewise mode.
19336 (iv_analyze): Add a mode parameter.
19338 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
19339 and change its type to scalar_int_mode.
19340 * loop-iv.c: Update commentary at head of file.
19341 (iv_constant): Pass the mode paraeter before the rtx it describes
19342 and change its type to scalar_int_mode. Remove VOIDmode handling.
19343 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
19344 (iv_extend): Likewise.
19345 (shorten_into_mode): Likewise.
19346 (iv_add): Use scalar_int_mode.
19347 (iv_mult): Likewise.
19348 (iv_shift): Likewise.
19349 (canonicalize_iv_subregs): Likewise.
19350 (get_biv_step_1): Pass the outer_mode parameter before the rtx
19351 it describes and change its mode to scalar_int_mode. Also change
19352 the type of the returned inner_mode to scalar_int_mode.
19353 (get_biv_step): Likewise, turning outer_mode from a pointer
19354 into a direct parameter. Update call to get_biv_step_1.
19355 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
19356 iv_constant and get_biv_step.
19357 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
19358 and change its type to scalar_int_mode. Don't initialise iv->mode
19359 to VOIDmode and remove later checks for its still being VOIDmode.
19360 Update calls to iv_analyze_op and iv_analyze_expr. Check
19361 is_a <scalar_int_mode> when changing the mode under consideration.
19362 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
19363 Update call to iv_analyze_expr.
19364 (iv_analyze_op): Add a mode parameter. Reject subregs whose
19365 inner register is not also a scalar_int_mode. Update call to
19367 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
19368 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
19369 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
19370 separate mode class checks. Update calls to iv_analyze. Remove
19371 fix-up of VOIDmodes after iv_analyze_biv.
19372 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
19373 don't have a scalar_int_mode. Update call to biv_p.
19375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19376 Alan Hayward <alan.hayward@arm.com>
19377 David Sherwood <david.sherwood@arm.com>
19379 * cfgexpand.c (convert_debug_memory_address): Use
19380 as_a <scalar_int_mode>.
19381 * combine.c (expand_compound_operation): Likewise.
19382 (make_extraction): Likewise.
19383 (change_zero_ext): Likewise.
19384 (simplify_comparison): Likewise.
19385 * cse.c (cse_insn): Likewise.
19386 * dwarf2out.c (minmax_loc_descriptor): Likewise.
19387 (mem_loc_descriptor): Likewise.
19388 (loc_descriptor): Likewise.
19389 * expmed.c (init_expmed_one_mode): Likewise.
19390 (synth_mult): Likewise.
19391 (emit_store_flag_1): Likewise.
19392 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
19393 of a comparison with size.
19394 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
19395 (reduce_to_bit_field_precision): Likewise.
19396 * function.c (expand_function_end): Likewise.
19397 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
19398 * loop-doloop.c (doloop_modify): Likewise.
19399 * optabs.c (expand_binop): Likewise.
19400 (expand_unop): Likewise.
19401 (expand_copysign_absneg): Likewise.
19402 (prepare_cmp_insn): Likewise.
19403 (maybe_legitimize_operand): Likewise.
19404 * recog.c (const_scalar_int_operand): Likewise.
19405 * rtlanal.c (get_address_mode): Likewise.
19406 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19407 (simplify_cond_clz_ctz): Likewise.
19408 * tree-nested.c (get_nl_goto_field): Likewise.
19409 * tree.c (build_vector_type_for_mode): Likewise.
19410 * var-tracking.c (use_narrower_mode): Likewise.
19412 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19413 Alan Hayward <alan.hayward@arm.com>
19414 David Sherwood <david.sherwood@arm.com>
19416 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
19417 * builtins.c (expand_builtin_signbit): Use it.
19418 * cfgexpand.c (expand_debug_expr): Likewise.
19419 * dojump.c (do_jump): Likewise.
19420 (do_compare_and_jump): Likewise.
19421 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
19422 * expmed.c (make_tree): Likewise.
19423 * expr.c (expand_expr_real_2): Likewise.
19424 (expand_expr_real_1): Likewise.
19425 (try_casesi): Likewise.
19426 * fold-const-call.c (fold_const_call_ss): Likewise.
19427 * fold-const.c (unextend): Likewise.
19428 (extract_muldiv_1): Likewise.
19429 (fold_single_bit_test): Likewise.
19430 (native_encode_int): Likewise.
19431 (native_encode_string): Likewise.
19432 (native_interpret_int): Likewise.
19433 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
19434 * internal-fn.c (expand_addsub_overflow): Likewise.
19435 (expand_neg_overflow): Likewise.
19436 (expand_mul_overflow): Likewise.
19437 (expand_arith_overflow): Likewise.
19438 * match.pd: Likewise.
19439 * stor-layout.c (layout_type): Likewise.
19440 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
19441 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
19442 * tree-ssanames.c (get_range_info): Likewise.
19443 * tree-switch-conversion.c (array_value_type) Likewise.
19444 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
19445 (vect_recog_divmod_pattern): Likewise.
19446 (vect_recog_mixed_size_cond_pattern): Likewise.
19447 * tree-vrp.c (extract_range_basic): Likewise.
19448 (simplify_float_conversion_using_ranges): Likewise.
19449 * tree.c (int_fits_type_p): Likewise.
19450 * ubsan.c (instrument_bool_enum_load): Likewise.
19451 * varasm.c (mergeable_string_section): Likewise.
19452 (narrowing_initializer_constant_valid_p): Likewise.
19453 (output_constant): Likewise.
19455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19456 Alan Hayward <alan.hayward@arm.com>
19457 David Sherwood <david.sherwood@arm.com>
19459 * machmode.h (NARROWEST_INT_MODE): New macro.
19460 * expr.c (alignment_for_piecewise_move): Use it instead of
19461 GET_CLASS_NARROWEST_MODE (MODE_INT).
19462 (push_block): Likewise.
19463 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
19465 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19468 Alan Hayward <alan.hayward@arm.com>
19469 David Sherwood <david.sherwood@arm.com>
19471 * postreload.c (move2add_valid_value_p): Change the type of the
19472 mode parameter to scalar_int_mode.
19473 (move2add_use_add2_insn): Add a mode parameter and use it instead
19475 (move2add_use_add3_insn): Likewise.
19476 (reload_cse_move2add): Update accordingly.
19478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19479 Alan Hayward <alan.hayward@arm.com>
19480 David Sherwood <david.sherwood@arm.com>
19482 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
19484 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
19485 * optabs.c (expand_unop): Likewise.
19487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19488 Alan Hayward <alan.hayward@arm.com>
19489 David Sherwood <david.sherwood@arm.com>
19491 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
19492 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
19493 (popcount_loc_descriptor): Likewise.
19494 (bswap_loc_descriptor): Likewise.
19495 (rotate_loc_descriptor): Likewise.
19496 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
19497 calling the functions above.
19499 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19500 Alan Hayward <alan.hayward@arm.com>
19501 David Sherwood <david.sherwood@arm.com>
19503 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
19505 (try_combine): Likewise.
19506 (simplify_if_then_else): Likewise.
19507 * cse.c (cse_insn): Likewise.
19508 * dwarf2out.c (mem_loc_descriptor): Likewise.
19509 * emit-rtl.c (gen_lowpart_common): Likewise.
19510 * simplify-rtx.c (simplify_truncation): Likewise.
19511 (simplify_binary_operation_1): Likewise.
19512 (simplify_const_relational_operation): Likewise.
19513 (simplify_ternary_operation): Likewise.
19514 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
19516 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19517 Alan Hayward <alan.hayward@arm.com>
19518 David Sherwood <david.sherwood@arm.com>
19520 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
19521 * reload.c (push_reload): Likewise.
19522 (find_reloads): Likewise.
19524 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19525 Alan Hayward <alan.hayward@arm.com>
19526 David Sherwood <david.sherwood@arm.com>
19528 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
19529 (make_compound_operation_int): Likewise.
19530 (change_zero_ext): Likewise.
19531 * expr.c (convert_move): Likewise.
19532 (convert_modes): Likewise.
19533 * fwprop.c (forward_propagate_subreg): Likewise.
19534 * loop-iv.c (get_biv_step_1): Likewise.
19535 * optabs.c (widen_operand): Likewise.
19536 * postreload.c (move2add_valid_value_p): Likewise.
19537 * recog.c (simplify_while_replacing): Likewise.
19538 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19539 (simplify_binary_operation_1): Likewise. Remove redundant
19540 mode equality check.
19542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19543 Alan Hayward <alan.hayward@arm.com>
19544 David Sherwood <david.sherwood@arm.com>
19546 * combine.c (combine_simplify_rtx): Add checks for
19547 is_a <scalar_int_mode>.
19548 (simplify_if_then_else): Likewise.
19549 (make_field_assignment): Likewise.
19550 (simplify_comparison): Likewise.
19551 * ifcvt.c (noce_try_bitop): Likewise.
19552 * loop-invariant.c (canonicalize_address_mult): Likewise.
19553 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19555 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19556 Alan Hayward <alan.hayward@arm.com>
19557 David Sherwood <david.sherwood@arm.com>
19559 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
19560 is_a <scalar_int_mode> instead of != BLKmode.
19562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19563 Alan Hayward <alan.hayward@arm.com>
19564 David Sherwood <david.sherwood@arm.com>
19566 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
19567 instead of != VOIDmode.
19568 * combine.c (if_then_else_cond): Likewise.
19569 (change_zero_ext): Likewise.
19570 * dwarf2out.c (mem_loc_descriptor): Likewise.
19571 (loc_descriptor): Likewise.
19572 * rtlanal.c (canonicalize_condition): Likewise.
19573 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
19575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19576 Alan Hayward <alan.hayward@arm.com>
19577 David Sherwood <david.sherwood@arm.com>
19579 * simplify-rtx.c (simplify_binary_operation_1): Use
19580 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
19582 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19583 Alan Hayward <alan.hayward@arm.com>
19584 David Sherwood <david.sherwood@arm.com>
19586 * wide-int.h (int_traits<unsigned char>) New class.
19587 (int_traits<unsigned short>) Likewise.
19588 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
19589 Use GET_MODE_UNIT_PRECISION and remove redundant test for
19591 * combine.c (set_nonzero_bits_and_sign_copies): Use
19592 is_a <scalar_int_mode>.
19593 (find_split_point): Likewise.
19594 (combine_simplify_rtx): Likewise.
19595 (simplify_logical): Likewise.
19596 (expand_compound_operation): Likewise.
19597 (expand_field_assignment): Likewise.
19598 (make_compound_operation): Likewise.
19599 (extended_count): Likewise.
19600 (change_zero_ext): Likewise.
19601 (simplify_comparison): Likewise.
19602 * dwarf2out.c (scompare_loc_descriptor): Likewise.
19603 (ucompare_loc_descriptor): Likewise.
19604 (minmax_loc_descriptor): Likewise.
19605 (mem_loc_descriptor): Likewise.
19606 (loc_descriptor): Likewise.
19607 * expmed.c (init_expmed_one_mode): Likewise.
19608 * lra-constraints.c (lra_constraint_offset): Likewise.
19609 * optabs.c (prepare_libcall_arg): Likewise.
19610 * postreload.c (move2add_note_store): Likewise.
19611 * reload.c (operands_match_p): Likewise.
19612 * rtl.h (load_extend_op): Likewise.
19613 * rtlhooks.c (gen_lowpart_general): Likewise.
19614 * simplify-rtx.c (simplify_truncation): Likewise.
19615 (simplify_unary_operation_1): Likewise.
19616 (simplify_binary_operation_1): Likewise.
19617 (simplify_const_binary_operation): Likewise.
19618 (simplify_const_relational_operation): Likewise.
19619 (simplify_subreg): Likewise.
19620 * stor-layout.c (bitwise_mode_for_mode): Likewise.
19621 * var-tracking.c (adjust_mems): Likewise.
19622 (prepare_call_arguments): Likewise.
19624 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19625 Alan Hayward <alan.hayward@arm.com>
19626 David Sherwood <david.sherwood@arm.com>
19628 * machmode.h (is_int_mode): New fuction.
19629 * combine.c (find_split_point): Use it.
19630 (combine_simplify_rtx): Likewise.
19631 (simplify_if_then_else): Likewise.
19632 (simplify_set): Likewise.
19633 (simplify_shift_const_1): Likewise.
19634 (simplify_comparison): Likewise.
19635 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
19636 * cse.c (notreg_cost): Likewise.
19637 (cse_insn): Likewise.
19638 * cselib.c (cselib_lookup_1): Likewise.
19639 * dojump.c (do_jump_1): Likewise.
19640 (do_compare_rtx_and_jump): Likewise.
19641 * dse.c (get_call_args): Likewise.
19642 * dwarf2out.c (rtl_for_decl_init): Likewise.
19643 (native_encode_initializer): Likewise.
19644 * expmed.c (emit_store_flag_1): Likewise.
19645 (emit_store_flag): Likewise.
19646 * expr.c (convert_modes): Likewise.
19647 (store_field): Likewise.
19648 (expand_expr_real_1): Likewise.
19649 * fold-const.c (fold_read_from_constant_string): Likewise.
19650 * gimple-ssa-sprintf.c (get_format_string): Likewise.
19651 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
19652 * optabs.c (expand_binop): Likewise.
19653 (expand_unop): Likewise.
19654 (expand_abs_nojump): Likewise.
19655 (expand_one_cmpl_abs_nojump): Likewise.
19656 * simplify-rtx.c (mode_signbit_p): Likewise.
19657 (val_signbit_p): Likewise.
19658 (val_signbit_known_set_p): Likewise.
19659 (val_signbit_known_clear_p): Likewise.
19660 (simplify_relational_operation_1): Likewise.
19661 * tree.c (vector_type_mode): Likewise.
19663 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19664 Alan Hayward <alan.hayward@arm.com>
19665 David Sherwood <david.sherwood@arm.com>
19667 * machmode.h (smallest_mode_for_size): Fix formatting.
19668 (smallest_int_mode_for_size): New function.
19669 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
19670 instead of smallest_mode_for_size.
19671 * combine.c (make_extraction): Likewise.
19672 * config/arc/arc.c (arc_expand_movmem): Likewise.
19673 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
19674 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
19675 * config/s390/s390.c (s390_expand_insv): Likewise.
19676 * config/sparc/sparc.c (assign_int_registers): Likewise.
19677 * config/spu/spu.c (spu_function_value): Likewise.
19678 (spu_function_arg): Likewise.
19679 * coverage.c (get_gcov_type): Likewise.
19680 (get_gcov_unsigned_t): Likewise.
19681 * dse.c (find_shift_sequence): Likewise.
19682 * expmed.c (store_bit_field_1): Likewise.
19683 * expr.c (convert_move): Likewise.
19684 (store_field): Likewise.
19685 * internal-fn.c (expand_arith_overflow): Likewise.
19686 * optabs-query.c (get_best_extraction_insn): Likewise.
19687 * optabs.c (expand_twoval_binop_libfunc): Likewise.
19688 * stor-layout.c (layout_type): Likewise.
19689 (initialize_sizetypes): Likewise.
19690 * targhooks.c (default_get_mask_mode): Likewise.
19691 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
19693 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19694 Alan Hayward <alan.hayward@arm.com>
19695 David Sherwood <david.sherwood@arm.com>
19697 * machmode.h (opt_mode::else_blk): New function.
19698 (int_mode_for_mode): Declare.
19699 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
19700 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
19702 * cfgexpand.c (expand_debug_expr): Likewise.
19703 * combine.c (gen_lowpart_or_truncate): Likewise.
19704 (gen_lowpart_for_combine): Likewise.
19705 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
19706 * config/avr/avr.c (avr_to_int_mode): Likewise.
19707 (avr_out_plus_1): Likewise.
19708 (avr_out_plus): Likewise.
19709 (avr_out_round): Likewise.
19710 * config/i386/i386.c (ix86_split_to_parts): Likewise.
19711 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
19712 (s390_expand_vcond): Likewise.
19713 * config/spu/spu.c (spu_split_immediate): Likewise.
19714 (spu_expand_mov): Likewise.
19715 * dse.c (get_stored_val): Likewise.
19716 * expmed.c (store_bit_field_1): Likewise.
19717 (convert_extracted_bit_field): Use int_mode_for_mode instead of
19719 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
19720 (extract_low_bits): Likewise.
19721 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
19722 handling rather than repeating the check.
19723 (emit_group_store): Likewise.
19724 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
19725 * optabs.c (expand_absneg_bit): Likewise.
19726 (expand_copysign_absneg): Likewise.
19727 (expand_copysign_bit): Likewise.
19728 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
19729 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19730 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
19731 * var-tracking.c (prepare_call_arguments): Likewise.
19732 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
19733 int_mode_for_mode instead of mode_for_size.
19734 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
19736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19737 Alan Hayward <alan.hayward@arm.com>
19738 David Sherwood <david.sherwood@arm.com>
19740 * machmode.h (int_mode_for_size): New function.
19741 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
19742 instead of mode_for_size.
19743 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
19745 * combine.c (expand_field_assignment): Use int_mode_for_size
19746 instead of mode_for_size.
19747 (make_extraction): Likewise.
19748 (simplify_shift_const_1): Likewise.
19749 (simplify_comparison): Likewise.
19750 * dojump.c (do_jump): Likewise.
19751 * dwarf2out.c (mem_loc_descriptor): Likewise.
19752 * emit-rtl.c (init_derived_machine_modes): Likewise.
19753 * expmed.c (flip_storage_order): Likewise.
19754 (convert_extracted_bit_field): Likewise.
19755 * expr.c (copy_blkmode_from_reg): Likewise.
19756 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
19757 * internal-fn.c (expand_mul_overflow): Likewise.
19758 * lower-subreg.c (simple_move): Likewise.
19759 * optabs-libfuncs.c (init_optabs): Likewise.
19760 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19761 * tree.c (vector_type_mode): Likewise.
19762 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
19763 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
19764 * tree-vect-generic.c (expand_vector_parallel): Likewise.
19765 * tree-vect-stmts.c (vectorizable_load): Likewise.
19766 (vectorizable_store): Likewise.
19768 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19769 Alan Hayward <alan.hayward@arm.com>
19770 David Sherwood <david.sherwood@arm.com>
19772 * coretypes.h (pod_mode): New type.
19773 (scalar_int_mode_pod): New typedef.
19774 * machmode.h (pod_mode): New class.
19775 (int_n_data_t::m): Change type to scalar_int_mode_pod.
19776 * genmodes.c (emit_mode_int_n): Update accordingly.
19777 * lower-subreg.h (target_lower_subreg): Change type to
19778 scalar_int_mode_pod.
19779 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
19780 scalar_int_mode_pod.
19782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19783 Alan Hayward <alan.hayward@arm.com>
19784 David Sherwood <david.sherwood@arm.com>
19786 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
19787 machine_mode to scalar_int_mode.
19788 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
19789 (rs6000_option_override_internal): Remove cast to int.
19790 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
19791 machine_mode to scalar_int_mode.
19792 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
19793 (rs6000_option_override_internal): Remove cast to int.
19794 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
19795 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
19797 * config/s390/s390.c (s390_expand_builtin): Likewise.
19798 * coretypes.h (scalar_int_mode): New type.
19799 (opt_scalar_int_mode): New typedef.
19800 * machmode.h (scalar_int_mode): New class.
19801 (scalar_int_mode::includes_p): New function.
19802 (byte_mode): Change type to scalar_int_mode.
19803 (word_mode): Likewise.
19804 (ptr_mode): Likewise.
19805 * emit-rtl.c (byte_mode): Likewise.
19806 (word_mode): Likewise.
19807 (ptr_mode): Likewise.
19808 (init_derived_machine_modes): Update accordingly.
19809 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
19810 and MODE_PARTIAL_INT.
19811 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
19812 opt_scalar_int_mode.
19814 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19815 Alan Hayward <alan.hayward@arm.com>
19816 David Sherwood <david.sherwood@arm.com>
19818 * target.def (libgcc_floating_mode_supported_p): Take a
19820 * doc/tm.texi: Regenerate.
19821 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
19823 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19824 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
19827 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19828 Alan Hayward <alan.hayward@arm.com>
19829 David Sherwood <david.sherwood@arm.com>
19831 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
19832 * doc/tm.texi: Regenerate.
19833 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
19834 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
19835 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
19836 * targhooks.h (default_floatn_mode): Likewise.
19837 * targhooks.c (default_floatn_mode): Likewise.
19838 * tree.c (build_common_tree_nodes): Update accordingly.
19840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19841 Alan Hayward <alan.hayward@arm.com>
19842 David Sherwood <david.sherwood@arm.com>
19844 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
19845 (mode_iterator::iterate_p): Likewise.
19846 (mode_iterator::get_wider): Likewise.
19847 * expr.c (init_expr_target): Use opt_scalar_float_mode.
19849 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19850 Alan Hayward <alan.hayward@arm.com>
19851 David Sherwood <david.sherwood@arm.com>
19853 * coretypes.h (opt_scalar_float_mode): New typedef.
19854 * machmode.h (float_mode_for_size): New function.
19855 * emit-rtl.c (double_mode): Delete.
19856 (init_emit_once): Use float_mode_for_size.
19857 * stor-layout.c (layout_type): Likewise.
19858 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19860 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19861 Alan Hayward <alan.hayward@arm.com>
19862 David Sherwood <david.sherwood@arm.com>
19864 * output.h (assemble_real): Take a scalar_float_mode.
19865 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19866 * config/arm/arm.md (consttable_4): Likewise.
19867 (consttable_8): Likewise.
19868 (consttable_16): Likewise.
19869 * config/mips/mips.md (consttable_float): Likewise.
19870 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19871 * varasm.c (assemble_real): Take a scalar_float_mode.
19872 (output_constant_pool_2): Update accordingly.
19873 (output_constant): Likewise.
19875 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19876 Alan Hayward <alan.hayward@arm.com>
19877 David Sherwood <david.sherwood@arm.com>
19879 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19880 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19881 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19882 (native_encode_real): Likewise.
19883 (native_interpret_real): Likewise.
19884 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19885 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19888 Alan Hayward <alan.hayward@arm.com>
19889 David Sherwood <david.sherwood@arm.com>
19891 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19892 <scalar_float_mode>. Simplify.
19893 (gen_extend_conv_libfunc): Likewise.
19895 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19896 Alan Hayward <alan.hayward@arm.com>
19897 David Sherwood <david.sherwood@arm.com>
19899 * coretypes.h (scalar_float_mode): New type.
19900 * machmode.h (mode_traits::from_int): Use machine_mode if
19901 USE_ENUM_MODES is defined.
19902 (is_a): New function.
19904 (dyn_cast): Likewise.
19905 (scalar_float_mode): New class.
19906 (scalar_float_mode::includes_p): New function.
19907 (is_float_mode): Likewise.
19908 * gdbhooks.py (MachineModePrinter): New class.
19909 (build_pretty_printer): Use it for scalar_float_mode.
19910 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19911 (format_helper::format_helper): Turn into a template.
19912 * genmodes.c (get_mode_class): New function.
19913 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19914 or machine_mode if none.
19915 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19916 as_a <scalar_float_mode>.
19917 * dwarf2out.c (mem_loc_descriptor): Likewise.
19918 (insert_float): Likewise.
19919 (add_const_value_attribute): Likewise.
19920 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19921 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19922 (expand_unop): Update accordingly.
19923 (expand_abs_nojump): Likewise.
19924 (expand_copysign_absneg): Take a scalar_float_mode.
19925 (expand_copysign_bit): Likewise.
19926 (expand_copysign): Update accordingly.
19928 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19929 Alan Hayward <alan.hayward@arm.com>
19930 David Sherwood <david.sherwood@arm.com>
19932 * coretypes.h (opt_mode): New class.
19933 * machmode.h (opt_mode): Likewise.
19934 (opt_mode::else_void): New function.
19935 (opt_mode::require): Likewise.
19936 (opt_mode::exists): Likewise.
19937 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19938 (GET_MODE_2XWIDER_MODE): Likewise.
19939 (mode_iterator::get_wider): Update accordingly.
19940 (mode_iterator::get_2xwider): Likewise.
19941 (mode_iterator::get_known_wider): Likewise, turning into a template.
19942 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19943 forcing a wider mode to exist.
19944 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19945 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19946 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19947 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19948 * lower-subreg.c (init_lower_subreg): Likewise.
19949 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19950 on the final iteration.
19951 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19952 a wider mode exists before asking for a move pattern.
19953 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19954 forcing a wider mode to exist.
19955 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19956 returning false if no such mode exists.
19957 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19958 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19959 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19960 Avoid checking for a MODE_INT if we already know the mode is not a
19962 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19963 forcing a wider mode to exist.
19964 (expmed_mult_highpart_optab): Likewise.
19965 (expmed_mult_highpart): Likewise.
19966 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19968 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19969 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19970 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19971 * internal-fn.c (expand_mul_overflow): Update use of
19972 GET_MODE_2XWIDER_MODE.
19973 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19974 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19975 GET_MODE_WIDER_MODE.
19976 (convert_plusminus_to_widen): Likewise.
19977 * tree-switch-conversion.c (array_value_type): Likewise.
19978 * var-tracking.c (emit_note_insn_var_location): Likewise.
19979 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19980 Return false inside rather than outside the loop if no wider mode
19982 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19983 and GET_MODE_2XWIDER_MODE
19984 (can_compare_p): Use else_void.
19985 * gdbhooks.py (OptMachineModePrinter): New class.
19986 (build_pretty_printer): Use it for opt_mode.
19988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19989 Alan Hayward <alan.hayward@arm.com>
19990 David Sherwood <david.sherwood@arm.com>
19992 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19993 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19995 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19996 Alan Hayward <alan.hayward@arm.com>
19997 David Sherwood <david.sherwood@arm.com>
19999 * machmode.h (mode_traits): New structure.
20000 (get_narrowest_mode): New function.
20001 (mode_iterator::start): Likewise.
20002 (mode_iterator::iterate_p): Likewise.
20003 (mode_iterator::get_wider): Likewise.
20004 (mode_iterator::get_known_wider): Likewise.
20005 (mode_iterator::get_2xwider): Likewise.
20006 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
20007 (FOR_EACH_MODE): Likewise.
20008 (FOR_EACH_MODE_FROM): Likewise.
20009 (FOR_EACH_MODE_UNTIL): Likewise.
20010 (FOR_EACH_WIDER_MODE): Likewise.
20011 (FOR_EACH_2XWIDER_MODE): Likewise.
20012 * builtins.c (expand_builtin_strlen): Use new mode iterators.
20013 * combine.c (simplify_comparison): Likewise
20014 * config/i386/i386.c (type_natural_mode): Likewise.
20015 * cse.c (cse_insn): Likewise.
20016 * dse.c (find_shift_sequence): Likewise.
20017 * emit-rtl.c (init_derived_machine_modes): Likewise.
20018 (init_emit_once): Likewise.
20019 * explow.c (hard_function_value): Likewise.
20020 * expmed.c (extract_fixed_bit_field_1): Likewise.
20021 (extract_bit_field_1): Likewise.
20022 (expand_divmod): Likewise.
20023 (emit_store_flag_1): Likewise.
20024 * expr.c (init_expr_target): Likewise.
20025 (convert_move): Likewise.
20026 (alignment_for_piecewise_move): Likewise.
20027 (widest_int_mode_for_size): Likewise.
20028 (emit_block_move_via_movmem): Likewise.
20029 (copy_blkmode_to_reg): Likewise.
20030 (set_storage_via_setmem): Likewise.
20031 (compress_float_constant): Likewise.
20032 * omp-low.c (omp_clause_aligned_alignment): Likewise.
20033 * optabs-query.c (get_best_extraction_insn): Likewise.
20034 * optabs.c (expand_binop): Likewise.
20035 (expand_twoval_unop): Likewise.
20036 (expand_twoval_binop): Likewise.
20037 (widen_leading): Likewise.
20038 (widen_bswap): Likewise.
20039 (expand_parity): Likewise.
20040 (expand_unop): Likewise.
20041 (prepare_cmp_insn): Likewise.
20042 (prepare_float_lib_cmp): Likewise.
20043 (expand_float): Likewise.
20044 (expand_fix): Likewise.
20045 (expand_sfix_optab): Likewise.
20046 * postreload.c (move2add_use_add2_insn): Likewise.
20047 * reg-stack.c (reg_to_stack): Likewise.
20048 * reginfo.c (choose_hard_reg_mode): Likewise.
20049 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
20050 * stor-layout.c (mode_for_size): Likewise.
20051 (smallest_mode_for_size): Likewise.
20052 (mode_for_vector): Likewise.
20053 (finish_bitfield_representative): Likewise.
20054 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
20055 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
20056 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20057 * var-tracking.c (prepare_call_arguments): Likewise.
20059 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20060 Alan Hayward <alan.hayward@arm.com>
20061 David Sherwood <david.sherwood@arm.com>
20063 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
20064 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
20065 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
20066 * machmode.h (mode_size): Move earlier in file.
20067 (mode_precision): Likewise.
20068 (mode_inner): Likewise.
20069 (mode_nunits): Likewise.
20070 (mode_unit_size): Likewise.
20071 (unit_unit_precision): Likewise.
20072 (mode_wider): Likewise.
20073 (mode_2xwider): Likewise.
20074 (machine_mode): New class.
20075 (mode_to_bytes): New function.
20076 (mode_to_bits): Likewise.
20077 (mode_to_precision): Likewise.
20078 (mode_to_inner): Likewise.
20079 (mode_to_unit_size): Likewise.
20080 (mode_to_unit_precision): Likewise.
20081 (mode_to_nunits): Likewise.
20082 (GET_MODE_SIZE): Use mode_to_bytes.
20083 (GET_MODE_BITSIZE): Use mode_to_bits.
20084 (GET_MODE_PRECISION): Use mode_to_precision.
20085 (GET_MODE_INNER): Use mode_to_inner.
20086 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
20087 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
20088 (GET_MODE_NUNITS): Use mode_to_nunits.
20089 * system.h (ALWAYS_INLINE): New macro.
20090 * config/powerpcspe/powerpcspe-c.c
20091 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
20092 int for arg1_mode and arg2_mode.
20094 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20095 Alan Hayward <alan.hayward@arm.com>
20096 David Sherwood <david.sherwood@arm.com>
20098 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
20099 Prefix mode names with E_ in case statements.
20100 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20101 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
20102 (aarch64_split_simd_move): Likewise.
20103 (aarch64_gen_storewb_pair): Likewise.
20104 (aarch64_gen_loadwb_pair): Likewise.
20105 (aarch64_gen_store_pair): Likewise.
20106 (aarch64_gen_load_pair): Likewise.
20107 (aarch64_get_condition_code_1): Likewise.
20108 (aarch64_constant_pool_reload_icode): Likewise.
20109 (get_rsqrte_type): Likewise.
20110 (get_rsqrts_type): Likewise.
20111 (get_recpe_type): Likewise.
20112 (get_recps_type): Likewise.
20113 (aarch64_gimplify_va_arg_expr): Likewise.
20114 (aarch64_simd_container_mode): Likewise.
20115 (aarch64_emit_load_exclusive): Likewise.
20116 (aarch64_emit_store_exclusive): Likewise.
20117 (aarch64_expand_compare_and_swap): Likewise.
20118 (aarch64_gen_atomic_cas): Likewise.
20119 (aarch64_emit_bic): Likewise.
20120 (aarch64_emit_atomic_swap): Likewise.
20121 (aarch64_emit_atomic_load_op): Likewise.
20122 (aarch64_evpc_trn): Likewise.
20123 (aarch64_evpc_uzp): Likewise.
20124 (aarch64_evpc_zip): Likewise.
20125 (aarch64_evpc_ext): Likewise.
20126 (aarch64_evpc_rev): Likewise.
20127 (aarch64_evpc_dup): Likewise.
20128 (aarch64_gen_ccmp_first): Likewise.
20129 (aarch64_gen_ccmp_next): Likewise.
20130 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
20131 (alpha_emit_xfloating_libcall): Likewise.
20132 (emit_insxl): Likewise.
20133 (alpha_arg_type): Likewise.
20134 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
20135 (arc_preferred_simd_mode): Likewise.
20136 (arc_secondary_reload): Likewise.
20137 (get_arc_condition_code): Likewise.
20138 (arc_print_operand): Likewise.
20139 (arc_legitimate_constant_p): Likewise.
20140 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20141 * config/arc/arc.md (casesi_load): Likewise.
20142 (casesi_compact_jump): Likewise.
20143 * config/arc/predicates.md (proper_comparison_operator): Likewise.
20144 (cc_use_register): Likewise.
20145 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20146 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
20147 (arm_init_iwmmxt_builtins): Likewise.
20148 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
20149 (neon_expand_vector_init): Likewise.
20150 (arm_attr_length_move_neon): Likewise.
20151 (maybe_get_arm_condition_code): Likewise.
20152 (arm_emit_vector_const): Likewise.
20153 (arm_preferred_simd_mode): Likewise.
20154 (arm_output_iwmmxt_tinsr): Likewise.
20155 (thumb1_output_casesi): Likewise.
20156 (thumb2_output_casesi): Likewise.
20157 (arm_emit_load_exclusive): Likewise.
20158 (arm_emit_store_exclusive): Likewise.
20159 (arm_expand_compare_and_swap): Likewise.
20160 (arm_evpc_neon_vuzp): Likewise.
20161 (arm_evpc_neon_vzip): Likewise.
20162 (arm_evpc_neon_vrev): Likewise.
20163 (arm_evpc_neon_vtrn): Likewise.
20164 (arm_evpc_neon_vext): Likewise.
20165 (arm_validize_comparison): Likewise.
20166 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
20167 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
20168 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
20169 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
20170 (c6x_preferred_simd_mode): Likewise.
20171 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
20172 (epiphany_rtx_costs): Likewise.
20173 * config/epiphany/predicates.md (proper_comparison_operator):
20175 * config/frv/frv.c (condexec_memory_operand): Likewise.
20176 (frv_emit_move): Likewise.
20177 (output_move_single): Likewise.
20178 (output_condmove_single): Likewise.
20179 (frv_hard_regno_mode_ok): Likewise.
20180 (frv_matching_accg_mode): Likewise.
20181 * config/h8300/h8300.c (split_adds_subs): Likewise.
20182 (h8300_rtx_costs): Likewise.
20183 (h8300_print_operand): Likewise.
20184 (compute_mov_length): Likewise.
20185 (output_logical_op): Likewise.
20186 (compute_logical_op_length): Likewise.
20187 (compute_logical_op_cc): Likewise.
20188 (h8300_shift_needs_scratch_p): Likewise.
20189 (output_a_shift): Likewise.
20190 (compute_a_shift_length): Likewise.
20191 (compute_a_shift_cc): Likewise.
20192 (expand_a_rotate): Likewise.
20193 (output_a_rotate): Likewise.
20194 * config/i386/i386.c (classify_argument): Likewise.
20195 (function_arg_advance_32): Likewise.
20196 (function_arg_32): Likewise.
20197 (function_arg_64): Likewise.
20198 (function_value_64): Likewise.
20199 (ix86_gimplify_va_arg): Likewise.
20200 (ix86_legitimate_constant_p): Likewise.
20201 (put_condition_code): Likewise.
20202 (split_double_mode): Likewise.
20203 (ix86_avx256_split_vector_move_misalign): Likewise.
20204 (ix86_expand_vector_logical_operator): Likewise.
20205 (ix86_split_idivmod): Likewise.
20206 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
20207 (ix86_build_const_vector): Likewise.
20208 (ix86_build_signbit_mask): Likewise.
20209 (ix86_match_ccmode): Likewise.
20210 (ix86_cc_modes_compatible): Likewise.
20211 (ix86_expand_branch): Likewise.
20212 (ix86_expand_sse_cmp): Likewise.
20213 (ix86_expand_sse_movcc): Likewise.
20214 (ix86_expand_int_sse_cmp): Likewise.
20215 (ix86_expand_vec_perm_vpermi2): Likewise.
20216 (ix86_expand_vec_perm): Likewise.
20217 (ix86_expand_sse_unpack): Likewise.
20218 (ix86_expand_int_addcc): Likewise.
20219 (ix86_split_to_parts): Likewise.
20220 (ix86_vectorize_builtin_gather): Likewise.
20221 (ix86_vectorize_builtin_scatter): Likewise.
20222 (avx_vpermilp_parallel): Likewise.
20223 (inline_memory_move_cost): Likewise.
20224 (ix86_tieable_integer_mode_p): Likewise.
20225 (x86_maybe_negate_const_int): Likewise.
20226 (ix86_expand_vector_init_duplicate): Likewise.
20227 (ix86_expand_vector_init_one_nonzero): Likewise.
20228 (ix86_expand_vector_init_one_var): Likewise.
20229 (ix86_expand_vector_init_concat): Likewise.
20230 (ix86_expand_vector_init_interleave): Likewise.
20231 (ix86_expand_vector_init_general): Likewise.
20232 (ix86_expand_vector_set): Likewise.
20233 (ix86_expand_vector_extract): Likewise.
20234 (emit_reduc_half): Likewise.
20235 (ix86_emit_i387_round): Likewise.
20236 (ix86_mangle_type): Likewise.
20237 (ix86_expand_round_sse4): Likewise.
20238 (expand_vec_perm_blend): Likewise.
20239 (canonicalize_vector_int_perm): Likewise.
20240 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
20241 (expand_vec_perm_1): Likewise.
20242 (expand_vec_perm_interleave3): Likewise.
20243 (expand_vec_perm_even_odd_pack): Likewise.
20244 (expand_vec_perm_even_odd_1): Likewise.
20245 (expand_vec_perm_broadcast_1): Likewise.
20246 (ix86_vectorize_vec_perm_const_ok): Likewise.
20247 (ix86_expand_vecop_qihi): Likewise.
20248 (ix86_expand_mul_widen_hilo): Likewise.
20249 (ix86_expand_sse2_abs): Likewise.
20250 (ix86_expand_pextr): Likewise.
20251 (ix86_expand_pinsr): Likewise.
20252 (ix86_preferred_simd_mode): Likewise.
20253 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
20254 * config/i386/sse.md (*andnot<mode>3): Likewise.
20255 (<mask_codefor><code><mode>3<mask_name>): Likewise.
20256 (*<code><mode>3): Likewise.
20257 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
20258 (ia64_expand_atomic_op): Likewise.
20259 (ia64_arg_type): Likewise.
20260 (ia64_mode_to_int): Likewise.
20261 (ia64_scalar_mode_supported_p): Likewise.
20262 (ia64_vector_mode_supported_p): Likewise.
20263 (expand_vec_perm_broadcast): Likewise.
20264 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
20265 (iq2000_function_arg_advance): Likewise.
20266 (iq2000_function_arg): Likewise.
20267 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
20268 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
20269 (m68k_libcall_value): Likewise.
20270 (m68k_function_value): Likewise.
20271 (sched_attr_op_type): Likewise.
20272 * config/mcore/mcore.c (mcore_output_move): Likewise.
20273 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
20275 (microblaze_function_arg): Likewise.
20276 * config/mips/mips.c (mips16_build_call_stub): Likewise.
20277 (mips_print_operand): Likewise.
20278 (mips_mode_ok_for_mov_fmt_p): Likewise.
20279 (mips_vector_mode_supported_p): Likewise.
20280 (mips_preferred_simd_mode): Likewise.
20281 (mips_expand_vpc_loongson_even_odd): Likewise.
20282 (mips_expand_vec_unpack): Likewise.
20283 (mips_expand_vi_broadcast): Likewise.
20284 (mips_expand_vector_init): Likewise.
20285 (mips_expand_vec_reduc): Likewise.
20286 (mips_expand_msa_cmp): Likewise.
20287 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
20288 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
20289 (cc_flags_for_mode): Likewise.
20290 * config/msp430/msp430.c (msp430_print_operand): Likewise.
20291 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
20292 (nds32_output_casesi_pc_relative): Likewise.
20293 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20294 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
20295 (nvptx_gen_unpack): Likewise.
20296 (nvptx_gen_pack): Likewise.
20297 (nvptx_gen_shuffle): Likewise.
20298 (nvptx_gen_wcast): Likewise.
20299 (nvptx_preferred_simd_mode): Likewise.
20300 * config/pa/pa.c (pa_secondary_reload): Likewise.
20301 * config/pa/predicates.md (base14_operand): Likewise.
20302 * config/powerpcspe/powerpcspe-c.c
20303 (altivec_resolve_overloaded_builtin): Likewise.
20304 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
20306 (rs6000_preferred_simd_mode): Likewise.
20307 (output_vec_const_move): Likewise.
20308 (rs6000_expand_vector_extract): Likewise.
20309 (rs6000_split_vec_extract_var): Likewise.
20310 (reg_offset_addressing_ok_p): Likewise.
20311 (rs6000_legitimate_offset_address_p): Likewise.
20312 (rs6000_legitimize_address): Likewise.
20313 (rs6000_emit_set_const): Likewise.
20314 (rs6000_const_vec): Likewise.
20315 (rs6000_emit_move): Likewise.
20316 (spe_build_register_parallel): Likewise.
20317 (rs6000_darwin64_record_arg_recurse): Likewise.
20318 (swap_selector_for_mode): Likewise.
20319 (spe_init_builtins): Likewise.
20320 (paired_init_builtins): Likewise.
20321 (altivec_init_builtins): Likewise.
20322 (do_load_for_compare): Likewise.
20323 (rs6000_generate_compare): Likewise.
20324 (rs6000_expand_float128_convert): Likewise.
20325 (emit_load_locked): Likewise.
20326 (emit_store_conditional): Likewise.
20327 (rs6000_output_function_epilogue): Likewise.
20328 (rs6000_handle_altivec_attribute): Likewise.
20329 (rs6000_function_value): Likewise.
20330 (emit_fusion_gpr_load): Likewise.
20331 (emit_fusion_p9_load): Likewise.
20332 (emit_fusion_p9_store): Likewise.
20333 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
20334 (fusion_gpr_mem_load): Likewise.
20335 (fusion_addis_mem_combo_load): Likewise.
20336 (fusion_addis_mem_combo_store): Likewise.
20337 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
20338 (fusion_gpr_mem_load): Likewise.
20339 (fusion_addis_mem_combo_load): Likewise.
20340 (fusion_addis_mem_combo_store): Likewise.
20341 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20343 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
20344 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20345 (rs6000_preferred_simd_mode): Likewise.
20346 (output_vec_const_move): Likewise.
20347 (rs6000_expand_vector_extract): Likewise.
20348 (rs6000_split_vec_extract_var): Likewise.
20349 (reg_offset_addressing_ok_p): Likewise.
20350 (rs6000_legitimate_offset_address_p): Likewise.
20351 (rs6000_legitimize_address): Likewise.
20352 (rs6000_emit_set_const): Likewise.
20353 (rs6000_const_vec): Likewise.
20354 (rs6000_emit_move): Likewise.
20355 (rs6000_darwin64_record_arg_recurse): Likewise.
20356 (swap_selector_for_mode): Likewise.
20357 (paired_init_builtins): Likewise.
20358 (altivec_init_builtins): Likewise.
20359 (rs6000_expand_float128_convert): Likewise.
20360 (emit_load_locked): Likewise.
20361 (emit_store_conditional): Likewise.
20362 (rs6000_output_function_epilogue): Likewise.
20363 (rs6000_handle_altivec_attribute): Likewise.
20364 (rs6000_function_value): Likewise.
20365 (emit_fusion_gpr_load): Likewise.
20366 (emit_fusion_p9_load): Likewise.
20367 (emit_fusion_p9_store): Likewise.
20368 * config/rx/rx.c (rx_gen_move_template): Likewise.
20369 (flags_from_mode): Likewise.
20370 * config/s390/predicates.md (s390_alc_comparison): Likewise.
20371 (s390_slb_comparison): Likewise.
20372 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
20373 (s390_vector_mode_supported_p): Likewise.
20374 (s390_cc_modes_compatible): Likewise.
20375 (s390_match_ccmode_set): Likewise.
20376 (s390_canonicalize_comparison): Likewise.
20377 (s390_emit_compare_and_swap): Likewise.
20378 (s390_branch_condition_mask): Likewise.
20379 (s390_rtx_costs): Likewise.
20380 (s390_secondary_reload): Likewise.
20381 (__SECONDARY_RELOAD_CASE): Likewise.
20382 (s390_expand_cs): Likewise.
20383 (s390_preferred_simd_mode): Likewise.
20384 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
20385 * config/sh/sh.c (sh_print_operand): Likewise.
20386 (dump_table): Likewise.
20387 (sh_secondary_reload): Likewise.
20388 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20389 * config/sh/sh.md (casesi_worker_1): Likewise.
20390 (casesi_worker_2): Likewise.
20391 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
20392 (fcc_comparison_operator): Likewise.
20393 * config/sparc/sparc.c (sparc_expand_move): Likewise.
20394 (emit_soft_tfmode_cvt): Likewise.
20395 (sparc_preferred_simd_mode): Likewise.
20396 (output_cbranch): Likewise.
20397 (sparc_print_operand): Likewise.
20398 (sparc_expand_vec_perm_bmask): Likewise.
20399 (vector_init_bshuffle): Likewise.
20400 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
20401 (spu_vector_mode_supported_p): Likewise.
20402 (spu_expand_insv): Likewise.
20403 (spu_emit_branch_or_set): Likewise.
20404 (spu_handle_vector_attribute): Likewise.
20405 (spu_builtin_splats): Likewise.
20406 (spu_builtin_extract): Likewise.
20407 (spu_builtin_promote): Likewise.
20408 (spu_expand_sign_extend): Likewise.
20409 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
20410 (tilegx_simd_int): Likewise.
20411 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
20412 (tilepro_simd_int): Likewise.
20413 * config/v850/v850.c (const_double_split): Likewise.
20414 (v850_print_operand): Likewise.
20415 (ep_memory_offset): Likewise.
20416 * config/vax/vax.c (vax_rtx_costs): Likewise.
20417 (vax_output_int_move): Likewise.
20418 (vax_output_int_add): Likewise.
20419 (vax_output_int_subtract): Likewise.
20420 * config/visium/predicates.md (visium_branch_operator): Likewise.
20421 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
20422 (visium_print_operand_address): Likewise.
20423 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20424 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
20425 (xtensa_expand_conditional_branch): Likewise.
20426 (xtensa_copy_incoming_a7): Likewise.
20427 (xtensa_output_literal): Likewise.
20428 * dfp.c (decimal_real_maxval): Likewise.
20429 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
20431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20432 Alan Hayward <alan.hayward@arm.com>
20433 David Sherwood <david.sherwood@arm.com>
20435 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
20436 (mode_nunits_inline): Likewise.
20437 (mode_inner_inline): Likewise.
20438 (mode_unit_size_inline): Likewise.
20439 (mode_unit_precision_inline): Likewise.
20440 (emit_insn_modes_h): Likewise. Also emit a #define of the
20442 (emit_mode_wider): Add an E_ prefix to mode names.
20443 (emit_mode_complex): Likewise.
20444 (emit_mode_inner): Likewise.
20445 (emit_mode_adjustments): Likewise.
20446 (emit_mode_int_n): Likewise.
20447 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
20448 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
20449 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
20450 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
20451 (CRC32_BUILTIN, ENTRY): Likewise.
20452 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
20453 (aarch64_pop_regs): Likewise.
20454 (aarch64_process_components): Likewise.
20455 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
20456 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
20457 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
20458 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
20459 * config/arm/arm.c (arm_init_libfuncs): Likewise.
20460 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
20462 * config/i386/i386-builtin.def (pcmpestr): Likewise.
20463 (pcmpistr): Likewise.
20464 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
20465 * config/mmix/mmix.c (mmix_output_condition): Likewise.
20466 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20468 * config/rl78/rl78.c (mduc_regs): Likewise.
20469 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20470 (htm_expand_builtin): Likewise.
20471 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
20472 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
20473 * config/xtensa/xtensa.c (print_operand): Likewise.
20474 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
20475 (NUM_MODE_VECTOR_INT): Likewise.
20476 * genoutput.c (null_operand): Likewise.
20477 (output_operand_data): Likewise.
20478 * genrecog.c (print_parameter_value): Likewise.
20479 * lra.c (debug_operand_data): Likewise.
20481 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20483 * dwarf2out.c (scompare_loc_descriptor_wide)
20484 (scompare_loc_descriptor_narrow): New functions, split out from...
20485 (scompare_loc_descriptor): ...here.
20486 * expmed.c (emit_store_flag_int): New function, split out from...
20487 (emit_store_flag): ...here.
20489 2017-08-30 Richard Biener <rguenther@suse.de>
20491 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
20492 (dwarf2out_early_finish): Move setting of AT_pubnames from
20493 early debug output to early finish.
20495 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20497 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
20498 and -mdata-region to the assembler.
20500 2017-08-30 Richard Biener <rguenther@suse.de>
20502 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
20504 (gen_subprogram_die): Add DW_AT_object_pointer only early.
20505 (dwarf2out_early_global_decl): Only generate a DIE for the
20506 abstract origin if it doesn't already exist or is a declaration DIE.
20507 (resolve_addr): Do not add the linkage name twice when
20508 generating a stub DIE for the DW_TAG_GNU_call_site target.
20510 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20512 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20513 Use machine_mode rather than int for arg1_mode.
20515 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20518 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
20519 that the second argument of the built-in functions to unpack
20520 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
20521 switch statement instead a lot of if statements.
20522 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
20523 Allow 64-bit values to be in Altivec registers as well as
20524 traditional floating point registers.
20525 (pack<mode>, FMOVE128_VSX iterator): Likewise.
20527 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
20529 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
20530 MAX_REGS_PER_ADDRESS == 1.
20532 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
20534 * config/i386/i386.opt (flag_fentry): Do not init to -1.
20535 * config/i386/i386.c (ix86_option_override_internal): Simplify
20536 setting of opts->x_flag_entry.
20538 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20539 Jakub Jelinek <jakub@redhat.com>
20540 Richard Biener <rguenther@suse.de>
20542 PR tree-optimization/81503
20543 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
20544 folded constant fits in the target type; reorder tests for clarity.
20546 2017-08-29 Martin Liska <mliska@suse.cz>
20548 * passes.def: Include pass_lower_switch.
20549 * stmt.c (dump_case_nodes): Remove and move to
20550 tree-switch-conversion.
20551 (case_values_threshold): Likewise.
20552 (expand_switch_as_decision_tree_p): Likewise.
20553 (emit_case_decision_tree): Likewise.
20554 (expand_case): Likewise.
20555 (balance_case_nodes): Likewise.
20556 (node_has_low_bound): Likewise.
20557 (node_has_high_bound): Likewise.
20558 (node_is_bounded): Likewise.
20559 (emit_case_nodes): Likewise.
20560 (struct simple_case_node): New struct.
20561 (add_case_node): Remove.
20562 (emit_case_dispatch_table): Use vector instead of case_list.
20563 (reset_out_edges_aux): Remove.
20564 (compute_cases_per_edge): Likewise.
20565 (expand_case): Build list of simple_case_node.
20566 (expand_sjlj_dispatch_table): Use it.
20567 * tree-switch-conversion.c (struct case_node): Moved from
20568 stmt.c and adjusted.
20569 (emit_case_nodes): Likewise.
20570 (node_has_low_bound): Likewise.
20571 (node_has_high_bound): Likewise.
20572 (node_is_bounded): Likewise.
20573 (case_values_threshold): Likewise.
20574 (reset_out_edges_aux): Likewise.
20575 (compute_cases_per_edge): Likewise.
20576 (add_case_node): Likewise.
20577 (dump_case_nodes): Likewise.
20578 (balance_case_nodes): Likewise.
20579 (expand_switch_as_decision_tree_p): Likewise.
20580 (emit_jump): Likewise.
20581 (emit_case_decision_tree): Likewise.
20582 (try_switch_expansion): Likewise.
20583 (do_jump_if_equal): Likewise.
20584 (emit_cmp_and_jump_insns): Likewise.
20585 (fix_phi_operands_for_edge): New function.
20586 (record_phi_operand_mapping): Likewise.
20587 (class pass_lower_switch): New pass.
20588 (pass_lower_switch::execute): New function.
20589 (make_pass_lower_switch): Likewise.
20590 (conditional_probability):
20591 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
20592 * tree-pass.h: Add make_pass_lower_switch.
20594 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20597 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
20600 2017-08-29 Richard Biener <rguenther@suse.de>
20602 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
20603 we do not add a DW_AT_inline attribute twice.
20604 (gen_subprogram_die): Remove code setting DW_AT_inline on
20605 DECL_ABSTRACT_P nodes.
20607 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
20609 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
20610 calls to internal functions.
20611 (gimplify_modify_expr): Likewise.
20612 * tree-call-cdce.c (use_internal_fn): Likewise.
20613 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
20614 (convert_to_divmod): Set the nothrow flag.
20615 * tree-if-conv.c (predicate_mem_writes): Likewise.
20616 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20617 (vectorizable_call): Likewise.
20618 (vectorizable_store): Likewise.
20619 (vectorizable_load): Likewise.
20620 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
20621 (vect_recog_mask_conversion_pattern): Likewise.
20623 2017-08-29 Martin Liska <mliska@suse.cz>
20626 * gcc.c (driver_handle_option): Add new argument.
20627 * opts-common.c (handle_option): Pass
20628 target_option_override_hook.
20629 * opts-global.c (lang_handle_option): Add new option.
20630 (set_default_handlers): Add new argument.
20631 (decode_options): Likewise.
20632 * opts.c (target_handle_option): Likewise.
20633 (common_handle_option): Call target_option_override_hook.
20634 * opts.h (struct cl_option_handler_func): Add hook for
20635 target option override.
20636 (struct cl_option_handlers): Likewise.
20637 (set_default_handlers): Add new argument.
20638 (decode_options): Likewise.
20639 (common_handle_option): Likewise.
20640 (target_handle_option): Likewise.
20641 * toplev.c (toplev::main): Pass targetm.target_option.override
20644 2017-08-29 Richard Biener <rguenther@suse.de>
20645 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
20647 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
20648 life to the active subtree.
20650 2017-08-28 Jeff Law <law@redhat.com>
20652 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
20653 derive_equivalences.
20654 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
20655 Code moved into....
20656 (edge_info::derive_equivalences): New private member function
20658 * tree-ssa-dom.c (class edge_info): Changed from a struct
20659 to a class. Add ctor/dtor, methods and data members.
20660 (edge_info::edge_info): Renamed from allocate_edge_info.
20661 Initialize additional members.
20662 (edge_info::~edge_info): New.
20663 (free_dom_edge_info): Delete the edge info.
20664 (record_edge_info): Use new class & associated member functions.
20665 Tighten forms for testing for edge equivalences.
20666 (record_temporary_equivalences): Iterate over the simple
20667 equivalences rather than assuming there's only one per edge.
20668 (cprop_into_successor_phis): Iterate over the simple
20669 equivalences rather than assuming there's only one per edge.
20670 (optimize_stmt): Use operand_equal_p rather than pointer
20671 equality for mini-DSE code.
20673 2017-08-28 Nathan Sidwell <nathan@acm.org>
20675 * gcc.c (execute): Fold SIGPIPE handling into switch
20676 statement. Adjust internal error message.
20678 2017-08-28 Richard Biener <rguenther@suse.de>
20681 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
20682 Do nothing for removed DIEs.
20684 2017-08-28 Richard Biener <rguenther@suse.de>
20686 PR tree-optimization/81977
20687 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
20690 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
20693 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
20694 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
20695 using targetm.gen_mem_thread_fence.
20697 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
20700 * config/i386/i386.md (*<btsc><mode>): Change operand 2
20701 predicate to register_operand. Reorder operands.
20702 (*btr<mode>): Ditto.
20703 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
20704 (*btr<mode>_mask): Ditto.
20706 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
20708 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
20709 * config/rs6000/xmmintrin.h: New file.
20710 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
20712 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20715 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
20716 parameter and_insn and return it.
20717 (recombine_lvx_pattern): Insert a copy to ensure availability of
20718 the base register of the copied masking operation at the point of
20719 the instruction replacement.
20720 (recombine_stvx_pattern): Likewise.
20722 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20724 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
20725 undocumented switches.
20726 (-mpower9-dform-vector): Likewise.
20727 (-mpower9-dform): Likewise.
20728 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
20729 comments to delete references to -mpower9-dform* switches.
20730 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
20731 Delete reference to -mpower9-dform* switches, test for
20732 -mpower9-vector instead.
20733 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
20734 (OTHER_P9_VECTOR_MASKS): Likewise.
20735 (POWERPC_MASKS): Likewise.
20736 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
20737 tests against -mpower9-dform* to -mpower9-vector. Delete code
20738 that checked for -mpower9-dform* consistancy with other options.
20739 Add test for -mpower9-misc to enable other power9 switches.
20740 (rs6000_init_hard_regno_mode_ok): Likewise.
20741 (rs6000_option_override_internal): Likewise.
20742 (rs6000_emit_prologue): Likewise.
20743 (rs6000_emit_epilogue): Likewise.
20744 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
20745 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
20746 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
20748 (emit_fusion_p9_store): Likewise.
20749 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
20750 resetting these macros if the assembler does not support ISA 3.0
20752 (TARGET_P9_DFORM_VECTOR): Likewise.
20753 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
20754 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
20756 2017-08-25 Alan Modra <amodra@gmail.com>
20758 PR rtl-optimization/81747
20759 * cse.c (cse_extended_basic_block): Don't attempt to record
20760 equivalences for degenerate conditional branches that branch
20761 to their fall-through.
20763 2017-08-24 Martin Sebor <msebor@redhat.com>
20765 PR middle-end/81908
20766 * gimple-fold.c (size_must_be_zero_p): New function.
20767 (gimple_fold_builtin_memory_op): Call it.
20769 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
20771 * config/rs6000/mm_malloc.h: New file.
20773 2017-08-24 Bin Cheng <bin.cheng@arm.com>
20775 PR tree-optimization/81913
20776 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
20777 analysis when either IVs in condition can wrap.
20779 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
20781 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
20782 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
20784 2017-08-24 Richard Biener <rguenther@suse.de>
20787 * targhooks.c (default_target_can_inline_p): Properly
20788 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
20789 is present and always compare.
20790 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
20791 infer -mfpmath=sse from TARGET_SSE_P.
20792 (ix86_can_inline_p): Properly use target_option_default_node when
20793 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
20795 2017-08-24 Richard Biener <rguenther@suse.de>
20798 * dwarf2out.c (output_die): Handle flag_generate_offload like
20800 (output_comp_unit): Likewise.
20801 (gen_array_type_die): Likewise.
20802 (dwarf2out_early_finish): Likewise.
20803 (note_variable_value_in_expr): Likewise.
20804 (dwarf2out_finish): Likewise. Adjust assert.
20805 * cgraphunit.c (symbol_table::compile): Move setting of
20806 flag_generate_offload earlier ...
20807 (symbol_table::finalize_compilation_unit): ... here, before
20808 early debug finalization.
20810 2017-08-24 Richard Biener <rguenther@suse.de>
20812 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
20813 and ipa-fnsummary.h.
20814 (ix86_can_inline_p): When ix86_fpmath flags do not match
20815 check whether the callee uses FP math at all.
20817 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
20819 PR middle-end/81931
20820 * tree-ssanames.c (get_nonzero_bits): Use element_precision
20821 instead of TYPE_PRECISION.
20823 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
20824 Alan Hayward <alan.hayward@arm.com>
20825 David Sherwood <david.sherwood@arm.com>
20827 * combine.c (make_extraction): Use subreg_offset_from_lsb.
20829 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
20831 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
20833 (ix86_frame::stack_realign_allocate): New field.
20834 (struct machine_frame_state): Modify comments.
20835 (machine_frame_state::sp_realigned_fp_end): New field.
20836 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
20837 layout calculation.
20838 (sp_valid_at): Add assertion to assure no attempt to access invalid
20839 offset of a realigned stack.
20840 (fp_valid_at): Likewise.
20841 (choose_baseaddr): Modify comments.
20842 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
20843 ix86_expand_prologue.
20844 (ix86_expand_prologue): Modify stack realignment and allocation.
20845 (ix86_expand_epilogue): Modify comments.
20846 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
20847 avx2_runtime, avx512f, and avx512f_runtime.
20849 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20851 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20852 (mstackrealign): Do not init to -1.
20853 * config/i386/i386.c (ix86_option_override_internal):
20854 Check opts_set, not opts when setting default value of
20855 opts->x_ix86_force_align_arg_pointer.
20857 2017-08-23 Richard Biener <rguenther@suse.de>
20859 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20860 lang_hooks.decl_printable_name.
20861 * print-rtl-function.c (print_rtx_function): Likewise.
20862 * tree-pretty-print.c (dump_function_header): Likewise.
20864 2017-08-23 Richard Biener <rguenther@suse.de>
20867 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20868 -g0 at compile-time.
20870 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20872 PR middle-end/19706
20873 * doc/sourcebuild.texi (Other hardware attributes):
20876 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20878 PR middle-end/19706
20879 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20880 Fix single-use check.
20882 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20884 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20886 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20888 * doc/install.texi: Modify to add more details on running selected
20891 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20893 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20894 is combined with -mabi=ms.
20895 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20898 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20900 PR tree-optimization/81488
20901 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20902 and cached_basis fields.
20903 (MAX_SPREAD): New constant.
20904 (alloc_cand_and_find_basis): Initialize new fields.
20905 (clear_visited): New function.
20906 (create_phi_basis_1): Rename from create_phi_basis, set visited
20907 and cached_basis fields.
20908 (create_phi_basis): New wrapper function.
20909 (phi_add_costs_1): Rename from phi_add_costs, add spread
20910 parameter, set visited field, short-circuit when limits reached.
20911 (phi_add_costs): New wrapper function.
20912 (record_phi_increments_1): Rename from record_phi_increments, set
20914 (record_phi_increments): New wrapper function.
20915 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20916 (phi_incr_cost): New wrapper function.
20917 (all_phi_incrs_profitable_1): Rename from
20918 all_phi_incrs_profitable, set visited field.
20919 (all_phi_incrs_profitable): New wrapper function.
20921 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20922 Alan Hayward <alan.hayward@arm.com>
20923 David Sherwood <david.sherwood@arm.com>
20925 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20926 that takes the outer and inner modes.
20927 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20928 comparison as the canonical test for a paradoxical subreg.
20929 * combine.c (simplify_set): Use paradoxical_subreg_p.
20930 (make_extraction): Likewise.
20931 (force_to_mode): Likewise.
20932 (rtx_equal_for_field_assignment_p): Likewise.
20933 (gen_lowpart_for_combine): Likewise.
20934 (simplify_comparison): Likewise.
20935 * cse.c (equiv_constant): Likewise.
20936 * expmed.c (store_bit_field_1): Likewise.
20937 * final.c (alter_subreg): Likewise.
20938 * fwprop.c (propagate_rtx): Likewise.
20939 (forward_propagate_subreg): Likewise.
20940 * ira-conflicts.c (ira_build_conflicts): Likewise.
20941 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20942 * lra-constraints.c (curr_insn_transform): Likewise.
20943 (split_reg): Likewise.
20944 * lra-eliminations.c (move_plus_up): Likewise.
20945 (lra_eliminate_regs_1): Likewise.
20946 * recog.c (general_operand): Likewise.
20947 * ree.c (combine_reaching_defs): Likewise.
20948 * reload.c (push_reload): Likewise.
20949 (find_reloads): Likewise.
20950 * reload1.c (elimination_effects): Likewise.
20951 (compute_reload_subreg_offset): Likewise.
20952 (choose_reload_regs): Likewise.
20953 * rtlanal.c (subreg_lsb_1): Likewise.
20954 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20955 (simplify_subreg): Likewise.
20956 * var-tracking.c (track_loc_p): Likewise.
20957 * emit-rtl.c (byte_lowpart_offset): Likewise.
20958 (paradoxical_subreg_p): Delete out-of-line definition.
20960 2017-08-22 Jeff Law <law@redhat.com>
20962 PR tree-optimization/81741
20963 PR tree-optimization/71947
20964 * tree-ssa-dom.c: Include tree-inline.h.
20965 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20966 equivalences if one is more expensive to compute than the other.
20967 * tree-ssa-scopedtables.h (class const_or_copies): Make
20968 record_const_or_copy_raw method private.
20969 (class avail_exprs_stack): New method simplify_binary_operation.
20970 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20971 avail_exprs_stack::simplify_binary_operation as needed.
20972 (avail_exprs_stack::simplify_binary_operation): New function.
20974 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20976 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20977 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20978 (DOT_SYMBOLS): Likewise.
20979 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20980 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20981 (RS6000_ABI_NAME): Likewise.
20982 (TARGET_CMODEL): Likewise.
20983 (TOC_SECTION_ASM_OP): Likewise.
20984 (SET_CMODEL): New macro.
20985 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20987 2017-08-22 Richard Biener <rguenther@suse.de>
20989 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20990 to allow for free-lang-data replacements similar to verify_type_variant.
20992 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20993 Alan Hayward <alan.hayward@arm.com>
20994 David Sherwood <david.sherwood@arm.com>
20996 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20997 VOIDmode for the LABEL_REF.
20999 2017-08-22 Richard Biener <rguenther@suse.de>
21001 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
21003 2017-08-22 Richard Biener <rguenther@suse.de>
21005 * common.opt (feliminate-dwarf2-dups): Ignore.
21006 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
21007 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
21008 same_die_p_wrap, compute_section_prefix,
21009 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
21010 (comdat_symbol_id, comdat_symbol_number): Likewise.
21011 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
21013 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
21014 (output_die): Mark unreachable path unreachable.
21015 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
21016 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
21017 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
21018 (dwarf2out_early_finish): Likewise.
21020 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
21022 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
21024 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
21027 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
21028 not VAR_P. Filter attribute warnings with OPT_Wattributes.
21029 (avr_attribute_table) <io, io_low, address>: Initialize
21030 .decl_required with true.
21032 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
21034 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
21035 undocumented debugging options.
21036 (-mvsx-scalar-double): Likewise.
21037 (-mallow-df-permute): Likewise.
21038 (-mvectorize-builtins): Likewise.
21039 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
21040 (rs6000_builtin_vectorized_function): Likewise.
21041 (rs6000_builtin_md_vectorized_function): Likewise.
21042 (rs6000_opt_vars): Likewise.
21044 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
21047 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
21048 (*btrq_imm): Rename from *btrq.
21049 (*btcq_imm): Rename from *btcq.
21050 (btsc): New code attribute.
21051 (*<btsc><mode>): New insn pattern.
21052 (*btr<mode>): Ditto.
21053 (*<btsc><mode>_mask): New insn_and_split pattern.
21054 (*btr<mode>_mask): Ditto.
21056 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21057 Alan Hayward <alan.hayward@arm.com>
21058 David Sherwood <david.sherwood@arm.com>
21060 * function.c (pad_below): Simplify padding calculation.
21062 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21063 Alan Hayward <alan.hayward@arm.com>
21064 David Sherwood <david.sherwood@arm.com>
21066 * target.def (function_prologue): Remove frame size argument.
21067 (function_epilogue): Likewise.
21068 * doc/tm.texi: Regenerate.
21069 * final.c (final_start_function): Update call to function_prologue.
21070 (final_end_function): Update call to function_epilogue.
21071 (default_function_pro_epilogue): Remove frame size argument.
21072 * output.h (default_function_pro_epilogue): Likewise.
21073 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
21074 (arm_output_function_prologue): Likewise.
21075 * config/frv/frv.c (frv_function_prologue): Likewise.
21076 (frv_function_epilogue): Likewise.
21077 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
21078 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
21079 (ia64_output_function_epilogue): Likewise.
21080 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
21081 (m32r_output_function_epilogue): Likewise.
21082 * config/microblaze/microblaze.c (microblaze_function_prologue)
21083 (microblaze_function_epilogue): Likewise.
21084 * config/mips/mips.c (mips_output_function_prologue): Likewise.
21085 (mips_output_function_epilogue): Likewise.
21086 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
21087 (mmix_target_asm_function_epilogue): Likewise.
21088 * config/msp430/msp430.c (msp430_start_function): Likewise.
21089 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
21090 (nds32_asm_function_epilogue): Likewise.
21091 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
21092 * config/pa/pa.c (pa_output_function_prologue): Likewise.
21093 (pa_output_function_epilogue): Likewise.
21094 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
21095 (rs6000_output_function_epilogue): Likewise.
21096 * config/rl78/rl78.c (rl78_start_function): Likewise.
21097 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
21098 (rs6000_output_function_epilogue): Likewise.
21099 * config/rx/rx.c (rx_output_function_prologue): Likewise.
21100 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
21101 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
21102 (sparc_asm_function_epilogue): Likewise.
21104 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21106 * tree.h (type_has_mode_precision_p): New function.
21107 * convert.c (convert_to_integer_1): Use it.
21108 * expr.c (expand_expr_real_2): Likewise.
21109 (expand_expr_real_1): Likewise.
21110 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
21111 * match.pd: Likewise.
21112 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
21113 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
21114 * tree-tailcall.c (process_assignment): Likewise.
21115 * tree-vect-loop.c (vectorizable_reduction): Likewise.
21116 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
21117 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
21118 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
21119 (vectorizable_assignment): Likewise.
21120 (vectorizable_shift): Likewise.
21121 (vectorizable_operation): Likewise.
21122 * tree-vrp.c (register_edge_assert_for_2): Likewise.
21124 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
21126 * match.pd: Add pow (C, x) simplification.
21128 2017-08-21 Richard Biener <rguenther@suse.de>
21130 PR tree-optimization/81900
21131 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
21132 for blocks with abnormal predecessors.
21133 (compute_antic): Do not set visited flag prematurely.
21135 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
21138 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
21140 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21142 * stor-layout.h (vector_type_mode): Move to...
21143 * tree.h (vector_type_mode): ...here.
21144 * stor-layout.c (vector_type_mode): Move to...
21145 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
21147 2017-08-21 Richard Biener <rguenther@suse.de>
21149 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
21150 register_external_die hooks.
21151 (debug_false_tree_charstarstar_uhwistar): Declare.
21152 (debug_nothing_tree_charstar_uhwi): Likewise.
21153 * debug.c (do_nothing_debug_hooks): Adjust.
21154 (debug_false_tree_charstarstar_uhwistar): New do nothing.
21155 (debug_nothing_tree_charstar_uhwi): Likewise.
21156 * dbxout.c (dbx_debug_hooks): Adjust.
21157 (xcoff_debug_hooks): Likewise.
21158 * sdbout.c (sdb_debug_hooks): Likewise.
21159 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21160 * dwarf2out.c (macinfo_label_base): New global.
21161 (dwarf2out_register_external_die): New function for the
21162 register_external_die hook.
21163 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
21164 (dwarf2_debug_hooks): Use them.
21165 (dwarf2_lineno_debug_hooks): Adjust.
21166 (struct die_struct): Add with_offset flag.
21167 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
21168 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
21169 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
21170 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
21171 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
21172 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
21173 defining section names for the early LTO debug variants.
21174 (reset_indirect_string): New helper.
21175 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
21176 (print_dw_val): Add support for offsetted symbol references.
21177 (get_ultimate_context): Split out from is_cxx.
21178 (is_cxx): Use get_ultimate_context.
21179 (is_fortran): Add decl overload.
21180 (compute_comp_unit_symbol): Split out worker from
21181 compute_section_prefix.
21182 (compute_section_prefix): Call compute_comp_unit_symbol and
21183 set comdat_type_p here.
21184 (output_die): Skip DIE symbol output for the LTO added one.
21185 Handle DIE symbol references with offset.
21186 (output_comp_unit): Guard section name mangling properly.
21187 For LTO debug sections emit a symbol at the section beginning
21188 which we use to refer to its DIEs.
21189 (add_abstract_origin_attribute): For DIEs registered via
21190 dwarf2out_register_external_die directly refer to the early
21191 DIE rather than indirectly through the shadow one we created.
21192 Remove obsolete call to dwarf2out_abstract_function for
21193 non-function/block origins.
21194 (gen_array_type_die): When generating early LTO debug do
21195 not emit DW_AT_string_length.
21196 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
21197 late when in LTO. As suggested place a gcc_unreachable for
21198 the DECL_ABSTRACT_P case.
21199 (gen_subprogram_die): Avoid another specification DIE
21200 for early built declarations/definitions for the late LTO case.
21201 (gen_variable_die): Add type references for late duplicated VLA dies
21203 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
21204 we have the abstract instance already.
21205 (process_scope_var): Adjust decl DIE contexts in LTO which
21206 first puts them in limbo.
21207 (gen_decl_die): Do not generate type DIEs late apart from
21208 types for VLAs or for decls we do not yet have a DIE. Do not
21209 call dwarf2out_abstract_function late.
21210 (dwarf2out_early_global_decl): Make sure to create DIEs
21211 for abstract instances of a decl first.
21212 (dwarf2out_late_global_decl): Adjust comment.
21213 (output_macinfo_op): With multiple macro sections use
21214 macinfo_label_base to distinguish labels.
21215 (output_macinfo): Likewise. Update macinfo_label_base.
21216 Pass in the line info label.
21217 (note_variable_value_in_expr): When generating LTO resolve
21218 all variable values here by generating DIEs as needed.
21219 (init_sections_and_labels): Add early LTO debug flag parameter
21220 and generate different sections and names if set. Add generation
21221 counter for the labels so we can have multiple of them.
21222 (reset_dies): Helper to allow DIEs to be output multiple times.
21223 (dwarf2out_finish): When outputting DIEs to the fat part of an
21224 LTO object first reset DIEs.
21225 (dwarf2out_early_finish): Output early DIEs when generating LTO.
21226 (modified_type_die): Check for decl_ultimate_origin being self
21228 (gen_type_die_with_usage): Likewise.
21229 (gen_typedef_die): Allow decl_ultimate_origin being self.
21230 (set_decl_abstract_flags): Remove.
21231 (set_block_abstract_flags): Likewise.
21232 (dwarf2out_abstract_function): Treat the early generated DIEs
21233 as the abstract copy and only add DW_AT_inline and
21234 DW_AT_artificial here and call set_decl_origin_self.
21235 If the DIE has an abstract origin don't do anything.
21236 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
21237 if we have none yet (Go fails to build one, PR78628).
21238 (variably_modified_type_p): Prevent endless recursion for Ada
21239 cyclic pointer types.
21240 * lto-streamer-in.c: Include debug.h.
21241 (dref_queue): New global.
21242 (lto_read_tree_1): Stream in DIE references.
21243 (lto_input_tree): Register DIE references.
21244 (input_function): Stream DECL_DEBUG_ARGS.
21245 * lto-streamer-out.c: Include debug.h.
21246 (lto_write_tree_1): Output DIE references.
21247 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
21248 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
21249 (output_function): Stream DECL_DEBUG_ARGS.
21250 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
21251 Stream DECL_ABSTRACT_ORIGIN.
21252 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
21253 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
21254 DECL_CONTEXT for file-scope decls.
21255 * lto-streamer.h (struct dref_entry): Declare.
21256 (dref_queue): Likewise.
21257 * cfgexpand.c (pass_expand::execute): Do not call the
21258 outlining_inline_function hook here.
21259 * lto-wrapper.c (debug_obj): New global.
21260 (tool_cleanup): Unlink it if required.
21261 (debug_objcopy): New function.
21262 (run_gcc): Handle early debug sections in the IL files by
21263 extracting them to separate files, partially linkin them and
21264 feeding the result back as result to the linker.
21265 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
21266 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
21267 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
21268 sections into a separate segment.
21269 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
21271 (darwin_asm_dwarf_section): Likewise.
21272 (darwin_asm_output_dwarf_offset): Likewise.
21273 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
21275 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21276 Alan Hayward <alan.hayward@arm.com>
21277 David Sherwood <david.sherwood@arm.com>
21279 * read-md.h (md_reader::record_potential_iterator_use): Replace
21280 pointer argument with an rtx and an index.
21281 * read-rtl.c (iterator_group::apply_iterator): Likewise.
21282 (apply_mode_iterator): Likewise.
21283 (apply_code_iterator): Likewise.
21284 (apply_int_iterator): Likewise.
21285 (apply_subst_iterator): Likewise.
21286 (record_iterator_use): Likewise.
21287 (record_attribute_use): Likewise.
21288 (md_reader::record_potential_iterator_use): Likewise. Update calls
21289 to record_iterator_use and apply_iterator.
21290 (iterator_use): Replace ptr with x and index.
21291 (attribute_use): Likewise.
21292 (apply_attribute_uses): Update calls to apply_iterator.
21293 (apply_iterators): Likewise. Update initialization of iterator_use.
21294 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
21295 and record_potential_iterator_use.
21296 (rtx_reader::read_rtx_operand): Likewise.
21298 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21299 Alan Hayward <alan.hayward@arm.com>
21300 David Sherwood <david.sherwood@arm.com>
21302 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
21305 2017-08-21 Richard Biener <rguenther@suse.de>
21307 PR middle-end/81884
21308 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
21309 at struct end conservatively when comparing common bases.
21311 2017-08-21 Richard Biener <rguenther@suse.de>
21313 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
21314 (mem_ref_in_stmt): Remove.
21315 (determine_max_movement): Use ref index to get at the reference.
21316 (invariantness_dom_walker::before_dom_children): Deal with
21317 lim data already initialized.
21318 (gather_mem_refs_stmt): Initialize lim data and record ref index.
21320 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21322 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
21323 (TARGET_ISA_ROUND): Ditto.
21324 (TARGET_ROUND): Ditto.
21325 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
21326 * config/i386/i386.md: Ditto.
21327 * config/i386/sse.md: Ditto.
21328 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
21329 with OPTION_MASK_ISA_SSE4_1.
21331 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21334 * doc/extend.texi (x86 Built-in Functions): Correct the name of
21335 __builtin_ia32_lzcnt_u16.
21337 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
21340 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
21341 (rs6000_set_current_function): Rewrite function to use it.
21343 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
21346 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
21347 and TYPE_WARN_IF_NOT_ALIGN.
21348 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
21349 (handle_warn_if_not_align): New.
21350 (place_union_field): Call handle_warn_if_not_align.
21351 (place_field): Call handle_warn_if_not_align.
21352 Copy TYPE_WARN_IF_NOT_ALIGN.
21353 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
21354 (layout_type): Likewise.
21355 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
21357 (tree_decl_common): Add warn_if_not_align.
21358 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
21359 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
21360 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
21361 (DECL_WARN_IF_NOT_ALIGN): Likewise.
21362 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
21363 * doc/extend.texi: Document warn_if_not_aligned attribute.
21364 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
21366 2017-08-17 Martin Liska <mliska@suse.cz>
21369 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
21370 (get_data_dependence): Use it as pointer type.
21371 (distribute_loop): Likewise.
21373 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21375 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
21376 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
21377 (p8_vmrgow_<mode>_direct): New define_insn.
21378 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
21379 handle endianness for vmrgew and vmrgow permute patterns.
21381 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21383 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
21384 * config/rs6000/rs6000-cpus.def: Remove comment.
21385 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
21386 (POWERPC_MASKS): Likewise.
21387 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
21388 use of TARGET_VSX_TIMODE.
21389 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
21390 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
21391 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
21392 (rs6000_option_override_internal): Remove dead code.
21393 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
21394 (rs6000_legitimize_reload_address): Likewise.
21395 (rs6000_legitimate_address_p): Likewise.
21396 (rs6000_opt_masks): Delete "vsx-timode".
21397 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
21398 from function comment.
21399 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
21400 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
21401 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
21403 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
21404 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
21405 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
21406 (VSX_TI): Likewise.
21408 (define_peephole2): Likewise.
21410 2017-08-17 Martin Sebor <msebor@redhat.com>
21413 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
21414 past the end of an array.
21415 (test_pp_format): Add test cases.
21417 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
21419 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
21420 missing ECF_NOTHROW flags.
21422 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21425 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
21426 operands residing in integer registers.
21427 (*vsx_le_perm_load_<mode>): Likewise.
21428 (*vsx_le_perm_store_<mode>): Likewise.
21429 (define_peephole2): Add peepholes to optimize the above.
21431 2017-08-17 Marek Polacek <polacek@redhat.com>
21433 PR middle-end/81814
21434 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
21435 to mimic what shorten_compare did. Change the return type to bool.
21436 (fold_cond_expr_with_comparison): Update call to
21437 operand_equal_for_comparison_p.
21438 (fold_ternary_loc): Likewise.
21440 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
21442 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
21444 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
21446 2017-08-17 Richard Biener <rguenther@suse.de>
21448 * tree-ssa-structalias.c (solve_graph): When propagating
21449 to successors update the graphs succ edges and avoid duplicate work.
21451 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
21454 * config/i386/i386.c (ix86_option_override_internal): Save target
21455 specific options after ix86_stack_protector_guard_reg was changed.
21457 2017-08-17 Richard Biener <rguenther@suse.de>
21459 PR tree-optimization/81827
21460 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
21461 (new_var_info): Initialize it conservatively.
21462 (get_call_vi): Mark register vars.
21463 (new_scalar_tmp_constraint_exp): Likewise.
21464 (handle_rhs_call): Likewise.
21465 (handle_const_call): Likewise.
21466 (create_function_info_for): Likewise.
21467 (solve_constraints): Sort varinfos to separate register from
21468 non-register vars to pack points-to solution bitmaps during
21471 2017-08-17 Marek Polacek <polacek@redhat.com>
21473 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
21475 2017-08-17 Richard Biener <rguenther@suse.de>
21477 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
21478 to true when overflow is undefined and we saturated the result.
21480 2017-08-17 Alan Modra <amodra@gmail.com>
21483 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
21484 Don't use store multiple if only one reg needs saving.
21485 (interesting_frame_related_regno): New function.
21486 (rs6000_frame_related): Don't emit frame info for regs that
21488 (rs6000_emit_epilogue): Likewise.
21490 2017-08-16 Nathan Sidwell <nathan@acm.org>
21492 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
21493 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
21494 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
21495 * tree.c (free_lang_data_in_type): Use else-if chain. Always
21496 clear TYPE_LANG_1. Remove obsolete member-function stripping.
21497 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
21498 (verify_type): Adjust for TYPE_BINFO move.
21499 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
21500 process TYPE_BINFO directly.
21501 (hash_tree): Likewise.
21502 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21504 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21507 2017-08-16 David Malcolm <dmalcolm@redhat.com>
21509 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
21511 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
21514 * config/i386/i386.md (*anddi_1_btr): Change predicates of
21515 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
21516 Add ix86_binary_operator_ok to insn constraint.
21517 (*iordi_1_bts): Ditto.
21518 (*xordi_1_btc): Ditto.
21519 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
21520 Update corresponding peephole2 pattern.
21524 2017-08-16 Bin Cheng <bin.cheng@arm.com>
21526 PR tree-optimization/81832
21527 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
21528 copy loop header which has IFN_LOOP_DIST_ALIAS call.
21530 2017-08-16 Marek Polacek <polacek@redhat.com>
21533 * fold-const.c (fold_indirect_ref_1): Restore original behavior
21534 regarding size_zero_node.
21536 2017-08-16 Martin Liska <mliska@suse.cz>
21539 * config.gcc: Respect previously set extra_objs in case
21542 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
21544 PR tree-optimization/81835
21545 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
21546 the phi SSA_NAME. Check that the condition in a COND_EXPR does
21547 not depend on the phi.
21549 2017-08-16 Alan Modra <amodra@gmail.com>
21551 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
21554 2017-08-16 Alan Modra <amodra@gmail.com>
21556 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
21557 (save_reg_p): ..into this. Update all callers.
21558 (first_reg_to_save): Simplify.
21560 2017-08-16 Alan Modra <amodra@gmail.com>
21562 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
21565 2017-08-15 Joseph Myers <joseph@codesourcery.com>
21569 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
21570 constant count if that count is less than 32.
21572 2017-08-15 Nathan Sidwell <nathan@acm.org>
21574 * gcc.c (execute): Emit friendlier message if inferior is killed
21575 by an external cause.
21577 2017-08-15 Richard Biener <rguenther@suse.de>
21579 PR tree-optimization/81790
21580 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
21581 CONSTRUCTORs from simplifying and VN.
21583 2017-08-14 Martin Sebor <msebor@redhat.com>
21585 * builtin-attrs.def: Add comments.
21587 2017-08-14 Martin Sebor <msebor@redhat.com>
21590 * doc/extend.texi (attribute nonstring): Document new attribute.
21592 2017-08-14 Martin Sebor <msebor@redhat.com>
21595 * tree-diagnostic.c (default_tree_printer): Handle %G.
21596 * gimple-pretty-print.h (percent_G_format): Declare new function.
21597 * gimple-pretty-print.c (percent_G_format): Define.
21598 * tree-pretty-print.c (percent_K_format): Add argument.
21600 2017-08-14 Martin Sebor <msebor@redhat.com>
21602 PR translation/79998
21603 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
21604 Remove a stray space.
21606 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
21609 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
21610 (*iordi_1_bts): Ditto.
21611 (*xordi_1_btc): Ditto.
21613 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21616 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
21617 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21619 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
21620 quoted strings, and make more translator-friendly.
21621 (darwin_rs6000_override_options): Likewise.
21622 (rs6000_option_override_internal): Likewise.
21623 (rs6000_return_in_memory): Fix overlong line.
21624 (init_cmulative_args): Use quoted strings, and make more
21625 translator-friendly.
21626 (rs6000_pass_by_reference): Fix overlong line.
21627 (def_builtin): Use quoted strings.
21628 (altivec_expand_predicate_builtin): Use quoted strings, and make
21629 more translator-friendly.
21630 (htm_expand_builtin): Use quoted strings.
21631 (cpu_expand_builtin): Use quoted strings, and make more
21632 translator-friendly.
21633 (altivec_expand_builtin): Likewise.
21634 (paired_expand_predicate_builtin): Likewise.
21635 (rs6000_invalid_builtin): Likewise.
21636 (builtin_function_type): Use quoted strings.
21637 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
21638 more translator-friendly.
21639 (rs6000_trampoline_init): Likewise.
21640 (rs6000_handle_altivec_attribute): Likewise.
21641 (rs6000_inner_target_options): Use quoted strings.
21642 (rs6000_disable_incompatible_switches): Likewise.
21643 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
21644 strings, and make more translator-friendly.
21645 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
21647 2017-08-14 Bin Cheng <bin.cheng@arm.com>
21649 PR tree-optimization/81799
21650 * tree-loop-distribution.c (version_loop_by_alias_check): Force
21651 cond_expr to simple gimple operand.
21653 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
21655 PR middle-end/46932
21656 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
21658 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
21662 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
21663 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
21664 TARGET_GASISR_PROLOGUES.
21665 * config/avr/avr.c (avr_option_override): Same.
21666 (avr_pass_pre_proep::execute): Same.
21668 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
21671 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
21672 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
21674 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
21676 * config/i386/i386.md (*load_tp_<mode>): Redefine as
21677 define_insn_and_split. Split to a memory load from 0 in
21678 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
21679 using PTR mode iterator.
21680 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
21681 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
21682 (*add_tp_<mode>): Redefine as define_insn_and_split.
21683 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
21684 address space. Merge with *add_tp_x32 using PTR mode iterator.
21685 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
21686 Split to an add with a memory load from 0 in
21687 DEFAULT_TLS_SEG_REG address space.
21689 2017-08-12 Andrew Pinski <apinski@cavium.com>
21691 * config/aarch64/aarch64-option-extensions.def (rdma):
21692 Fix feature string to what Linux prints out in /proc/cpuinfo.
21694 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
21697 * dwarf2out.c (modified_type_die): For C typedef types that have
21698 an ultimate origin, process the ultimate origin instead of the
21700 (gen_typedef_die): Assert that input DECLs have no ultimate
21702 (gen_type_die_with_usage): For typedef variants that have an
21703 ultimate origin, just call gen_decl_die on the original DECL.
21704 (process_scope_var): Avoid creating DIEs for local typedefs and
21705 concrete static variables.
21707 2017-08-12 Alan Modra <amodra@gmail.com>
21711 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
21712 match gnu-user.h startfile.
21713 (ENDFILE_LINUX_SPEC): Similarly.
21715 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
21718 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
21720 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
21722 2017-08-11 Tamar Christina <tamar.christina@arm.com>
21723 * config/aarch64/aarch64.md (mov<mode>): Change.
21724 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21725 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
21726 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
21728 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
21730 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
21731 for storage order barriers.
21733 2017-08-11 Martin Liska <mliska@suse.cz>
21736 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
21737 variables of void type.
21739 2017-08-11 Martin Liska <mliska@suse.cz>
21741 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
21742 TARGET_SUPPORTS_ALIASES.
21743 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
21744 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
21745 (optimize_weakref): Likewise.
21746 * symtab.c (symtab_node::noninterposable_alias): Likewise.
21747 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
21748 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
21750 2017-08-11 Martin Liska <mliska@suse.cz>
21753 * config/i386/i386.c (make_resolver_func): Do complete
21754 refactoring of the function.
21756 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21759 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
21760 * config/i386/i386.c (ix86_stack_protect_guard): Use
21761 ix86_stack_protect_guard_symbol_str to generate varible declaration.
21762 * doc/invoke.texi (x86 Options): Document
21763 -mstack-protector-guard-symbol= option.
21765 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21767 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
21768 * config/i386/i386.c (ix86_split_stack_guard): New function.
21769 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
21770 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
21771 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
21772 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
21773 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
21774 (split_stack_space_check): Call ix86_split_stack_guard.
21776 2017-08-10 Martin Sebor <msebor@redhat.com>
21778 * print-tree.c (print_node): Print location using the established
21780 Replace spaces with colons.
21781 (debug_raw, debug): Ditto.
21783 2017-08-10 Martin Sebor <msebor@redhat.com>
21786 * pretty-print.c (pp_format): Correct the handling of %s precision.
21788 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
21791 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21793 (ix86_finalize_stack_frame_flags): This. Also clear
21794 frame_pointer_needed if -fno-omit-frame-pointer is used without
21796 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21797 with ix86_finalize_stack_frame_flags.
21798 (ix86_expand_epilogue): Likewise.
21799 (ix86_expand_split_stack_prologue): Likewise.
21800 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
21802 2017-08-10 Martin Liska <mliska@suse.cz>
21805 * c-attribs.c (handle_target_attribute):
21806 Report warning for an empty string argument of target attribute.
21808 2017-08-09 Jakub Jelinek <jakub@redhat.com>
21811 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
21813 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
21814 or DECL_NONLOCAL labels.
21815 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
21816 or DECL_NONLOCAL labels here.
21818 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
21820 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
21821 to indicate when early gimple folding has been disabled.
21822 (rs6000_gimple_fold_builtin): Add debug content.
21823 (rs6000_invalid_builtin): Fix whitespace.
21824 (rs6000_expand_builtin): Fix whitespace.
21825 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
21827 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
21830 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
21831 SAVE_MULTIPLE if not all the registers that saves, should be saved.
21833 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
21835 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
21836 (qdf24xx): Likewise.
21837 * config/aarch64/aarch64.md: Include falkor.md.
21838 * config/aarch64/falkor.md: New.
21840 2017-08-09 Marek Polacek <polacek@redhat.com>
21843 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
21844 * diagnostic.c (emit_diagnostic): Add a comment.
21845 (emit_diagnostic_valist): New function.
21847 2017-08-09 Marek Polacek <polacek@redhat.com>
21850 * input.c (make_location): New overload.
21851 * input.h (make_location): Declare.
21853 2017-08-08 Alan Modra <amodra@gmail.com>
21854 H.J. Lu <hongjiu.lu@intel.com>
21857 * gcc.c (NO_PIE_SPEC): Delete.
21858 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21860 (LINK_PIE_SPEC): ..to here.
21861 (LINK_COMMAND_SPEC): Support -no-pie.
21862 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21863 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21865 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21866 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21867 (ENDFILE_CRTEND_SPEC): Similarly.
21869 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21872 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21873 (mstack-protector-guard-offset=): Ditto.
21874 * config/i386/i386.c (ix86_option_override): Handle
21875 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21877 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21878 ix86_stack_protect_guard_offset variables.
21879 (TARGET_STACK_PROTECT_GUARD): Always define.
21880 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21881 and -mstack-protector-guard-offset= options.
21883 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21885 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21886 boundary case for the last candidate.
21888 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21890 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21891 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21893 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21895 PR middle-end/19706
21896 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21897 * config/aarch64/aarch64-builtins.c
21898 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21899 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21900 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21902 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21903 Andrew Pinski <pinskia@gmail.com>
21905 PR middle-end/19706
21906 * internal-fn.def (XORSIGN): New.
21907 * optabs.def (xorsign_optab): New.
21908 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21909 (convert_expand_mult_copysign): New.
21910 (pass_optimize_widening_mul::execute): Call
21911 convert_expand_mult_copysign.
21913 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21915 PR tree-optimization/81354
21916 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21917 Insert on edges rather than explicitly creating landing pads.
21918 (analyze_candidates_and_replace): Commit edge inserts.
21920 2017-08-08 Richard Biener <rguenther@suse.de>
21922 PR middle-end/81719
21923 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21924 (expand_simple_operations): Also look through ADDR_EXPRs with
21925 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21927 2017-08-08 Richard Biener <rguenther@suse.de>
21929 PR tree-optimization/81723
21930 * tree-vect-slp.c (struct bst_traits): New hash traits.
21931 (bst_fail): New global.
21932 (vect_build_slp_tree_2): New worker, split out from ...
21933 (vect_build_slp_tree): ... this now wrapping it with using
21934 bst_fail set to cache SLP tree build fails. Properly handle
21936 (vect_analyze_slp_instance): Allocate and free bst_fail.
21938 2017-08-08 Martin Liska <mliska@suse.cz>
21941 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21942 LABEL_DECLs that can be from a different function.
21944 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21946 PR tree-optimization/81744
21947 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21948 loop's number of iterations.
21950 2017-08-08 Martin Liska <mliska@suse.cz>
21952 * asan.c: Include header files.
21953 * attribs.c (build_decl_attribute_variant): New function moved
21955 (build_type_attribute_qual_variant): Likewise.
21956 (cmp_attrib_identifiers): Likewise.
21957 (simple_cst_list_equal): Likewise.
21958 (omp_declare_simd_clauses_equal): Likewise.
21959 (attribute_value_equal): Likewise.
21960 (comp_type_attributes): Likewise.
21961 (build_type_attribute_variant): Likewise.
21962 (lookup_ident_attribute): Likewise.
21963 (remove_attribute): Likewise.
21964 (merge_attributes): Likewise.
21965 (merge_type_attributes): Likewise.
21966 (merge_decl_attributes): Likewise.
21967 (merge_dllimport_decl_attributes): Likewise.
21968 (handle_dll_attribute): Likewise.
21969 (attribute_list_equal): Likewise.
21970 (attribute_list_contained): Likewise.
21971 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21972 (lookup_attribute_by_prefix): Likewise.
21973 * bb-reorder.c: Include header files.
21974 * builtins.c: Likewise.
21975 * calls.c: Likewise.
21976 * cfgexpand.c: Likewise.
21977 * cgraph.c: Likewise.
21978 * cgraphunit.c: Likewise.
21979 * convert.c: Likewise.
21980 * dwarf2out.c: Likewise.
21981 * final.c: Likewise.
21982 * fold-const.c: Likewise.
21983 * function.c: Likewise.
21984 * gimple-expr.c: Likewise.
21985 * gimple-fold.c: Likewise.
21986 * gimple-pretty-print.c: Likewise.
21987 * gimple.c: Likewise.
21988 * gimplify.c: Likewise.
21989 * hsa-common.c: Likewise.
21990 * hsa-gen.c: Likewise.
21991 * internal-fn.c: Likewise.
21992 * ipa-chkp.c: Likewise.
21993 * ipa-cp.c: Likewise.
21994 * ipa-devirt.c: Likewise.
21995 * ipa-fnsummary.c: Likewise.
21996 * ipa-inline.c: Likewise.
21997 * ipa-visibility.c: Likewise.
21999 * lto-cgraph.c: Likewise.
22000 * omp-expand.c: Likewise.
22001 * omp-general.c: Likewise.
22002 * omp-low.c: Likewise.
22003 * omp-offload.c: Likewise.
22004 * omp-simd-clone.c: Likewise.
22005 * opts-global.c: Likewise.
22006 * passes.c: Likewise.
22007 * predict.c: Likewise.
22008 * sancov.c: Likewise.
22009 * sanopt.c: Likewise.
22010 * symtab.c: Likewise.
22011 * toplev.c: Likewise.
22012 * trans-mem.c: Likewise.
22013 * tree-chkp.c: Likewise.
22014 * tree-eh.c: Likewise.
22015 * tree-into-ssa.c: Likewise.
22016 * tree-object-size.c: Likewise.
22017 * tree-parloops.c: Likewise.
22018 * tree-profile.c: Likewise.
22019 * tree-ssa-ccp.c: Likewise.
22020 * tree-ssa-live.c: Likewise.
22021 * tree-ssa-loop.c: Likewise.
22022 * tree-ssa-sccvn.c: Likewise.
22023 * tree-ssa-structalias.c: Likewise.
22024 * tree-ssa.c: Likewise.
22025 * tree-streamer-in.c: Likewise.
22026 * tree-vectorizer.c: Likewise.
22027 * tree-vrp.c: Likewise.
22028 * tsan.c: Likewise.
22029 * ubsan.c: Likewise.
22030 * varasm.c: Likewise.
22031 * varpool.c: Likewise.
22032 * tree.c: Remove functions moved to attribs.[ch].
22033 * tree.h: Likewise.
22034 * config/aarch64/aarch64.c: Add attrs.h header file.
22035 * config/alpha/alpha.c: Likewise.
22036 * config/arc/arc.c: Likewise.
22037 * config/arm/arm.c: Likewise.
22038 * config/avr/avr.c: Likewise.
22039 * config/bfin/bfin.c: Likewise.
22040 * config/c6x/c6x.c: Likewise.
22041 * config/cr16/cr16.c: Likewise.
22042 * config/cris/cris.c: Likewise.
22043 * config/darwin.c: Likewise.
22044 * config/epiphany/epiphany.c: Likewise.
22045 * config/fr30/fr30.c: Likewise.
22046 * config/frv/frv.c: Likewise.
22047 * config/ft32/ft32.c: Likewise.
22048 * config/h8300/h8300.c: Likewise.
22049 * config/i386/winnt.c: Likewise.
22050 * config/ia64/ia64.c: Likewise.
22051 * config/iq2000/iq2000.c: Likewise.
22052 * config/lm32/lm32.c: Likewise.
22053 * config/m32c/m32c.c: Likewise.
22054 * config/m32r/m32r.c: Likewise.
22055 * config/m68k/m68k.c: Likewise.
22056 * config/mcore/mcore.c: Likewise.
22057 * config/microblaze/microblaze.c: Likewise.
22058 * config/mips/mips.c: Likewise.
22059 * config/mmix/mmix.c: Likewise.
22060 * config/mn10300/mn10300.c: Likewise.
22061 * config/moxie/moxie.c: Likewise.
22062 * config/msp430/msp430.c: Likewise.
22063 * config/nds32/nds32-isr.c: Likewise.
22064 * config/nds32/nds32.c: Likewise.
22065 * config/nios2/nios2.c: Likewise.
22066 * config/nvptx/nvptx.c: Likewise.
22067 * config/pa/pa.c: Likewise.
22068 * config/pdp11/pdp11.c: Likewise.
22069 * config/powerpcspe/powerpcspe.c: Likewise.
22070 * config/riscv/riscv.c: Likewise.
22071 * config/rl78/rl78.c: Likewise.
22072 * config/rx/rx.c: Likewise.
22073 * config/s390/s390.c: Likewise.
22074 * config/sh/sh.c: Likewise.
22075 * config/sol2.c: Likewise.
22076 * config/sparc/sparc.c: Likewise.
22077 * config/spu/spu.c: Likewise.
22078 * config/stormy16/stormy16.c: Likewise.
22079 * config/tilegx/tilegx.c: Likewise.
22080 * config/tilepro/tilepro.c: Likewise.
22081 * config/v850/v850.c: Likewise.
22082 * config/vax/vax.c: Likewise.
22083 * config/visium/visium.c: Likewise.
22084 * config/xtensa/xtensa.c: Likewise.
22086 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
22089 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
22090 constraints since the -mupper-regs-* switches have been
22092 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
22093 into a vector from a double word element that was extracted from
22094 another vector, and eliminate extra XXPERMDI instructions.
22095 (vsx_concat_<mode>_2): Likewise.
22096 (vsx_concat_<mode>_3): Likewise.
22097 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
22098 concat to allow optimizing inserts from previous extracts.
22100 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
22102 * config/i386/i386.c (ix86_stack_protect_guard): Generate
22103 memory reference to a SSP offset in TLS address space.
22104 (ix86_print_operand) <case '@'>: Remove.
22105 (ix86_print_operand_punct_valid_p): Remove '@' code.
22106 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
22107 UNSPEC_SP_TLS_TEST.
22108 (stack_tls_protect_set_<mode>): Remove.
22109 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
22110 (stack_tls_protect_test_<mode>): Remove.
22111 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
22113 2017-08-07 Olivier Hainque <hainque@adacore.com>
22116 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
22118 2017-08-07 Douglas Rupp <rupp@adacore.com>
22120 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
22121 variable was referenced as multidir in command.
22123 2017-08-07 Jakub Jelinek <jakub@redhat.com>
22126 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
22129 2017-08-07 Martin Liska <mliska@suse.cz>
22131 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
22132 * config/rl78/rl78.c: Add include of attribs.h.
22133 * config/sh/sh.c: Likewise.
22134 * config/v850/v850.c: Likewise.
22136 2017-08-07 Tom de Vries <tom@codesourcery.com>
22138 PR middle-end/78266
22139 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
22141 2017-08-07 Martin Liska <mliska@suse.cz>
22143 * config/mips/mips.c: Include attribs.h.
22145 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
22148 * doc/invoke.texi: Document change in behvaior for -Ofast for
22151 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
22153 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
22155 (aarch64_pop_regs): Likewise.
22156 (aarch64_gen_load_pair): Likewise.
22157 (aarch64_save_callee_saves): Likewise.
22158 (aarch64_restore_callee_saves): Likewise.
22160 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22162 * config/i386/i386.c: Revert the last change.
22164 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22167 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
22169 (ix86_finalize_stack_frame_flags): This. Also clear
22170 frame_pointer_needed if -fno-omit-frame-pointer is used without
22172 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
22173 with ix86_finalize_stack_frame_flags.
22174 (ix86_expand_epilogue): Likewise.
22175 (ix86_expand_split_stack_prologue): Likewise.
22177 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22180 * config/i386/i386.c (get_builtin_code_for_version): Set priority
22181 to P_AES for Westmere.
22183 2017-08-07 Jonathan Yong <10walls@gmail.com>
22185 * config/i386/mingw.opt (fset-stack-executable): Removed.
22186 * config/i386/cygming.opt (fset-stack-executable): Moved
22188 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
22190 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
22192 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
22194 2017-08-07 Marek Polacek <polacek@redhat.com>
22196 PR middle-end/81737
22197 * fold-const.c (fold_indirect_ref_1): Check type_domain.
22199 2017-08-07 Martin Liska <mliska@suse.cz>
22201 * attribs.h (canonicalize_attr_name): New function.
22202 (cmp_attribs): Move from c-format.c and adjusted.
22203 (is_attribute_p): Moved from tree.h.
22204 * tree-inline.c: Add new includes.
22205 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
22206 (private_is_attribute_p): Remove.
22207 (private_lookup_attribute): Likewise.
22208 (private_lookup_attribute_by_prefix): Simplify.
22209 (remove_attribute): Use is_attribute_p.
22210 * tree.h: Remove removed declarations.
22212 2017-08-07 Jakub Jelinek <jakub@redhat.com>
22214 PR middle-end/81698
22215 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
22216 instead of computing it in the function. Formatting fix.
22217 (expand_case): Don't rely on default_edge being the first edge,
22218 clear it if removing it, pass default_edge to
22219 emit_case_dispatch_table.
22220 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
22223 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
22225 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
22226 insn in the function, emit NOP after the insn.
22228 2017-08-06 Tom de Vries <tom@codesourcery.com>
22230 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
22233 2017-08-06 Tom de Vries <tom@codesourcery.com>
22235 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
22238 2017-08-04 Yury Gribov <tetra2005@gmail.com>
22240 PR tree-optimization/57371
22241 * match.pd: New pattern.
22243 2017-08-04 Marek Polacek <polacek@redhat.com>
22245 PR middle-end/81695
22246 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
22247 perform the computation in offset_int.
22249 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22251 PR tree-optimization/81136
22252 * tree-vectorizer.h: Include tree-hash-traits.h.
22253 (vec_base_alignments): New typedef.
22254 (vec_info): Add a base_alignments field.
22255 (vect_record_base_alignments): Declare.
22256 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
22258 (DR_IS_CONDITIONAL_IN_STMT): New macro.
22259 (create_data_ref): Add an is_conditional_in_stmt argument.
22260 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
22261 the is_conditional_in_stmt field.
22262 (data_ref_loc): Add an is_conditional_in_stmt field.
22263 (get_references_in_stmt): Set the is_conditional_in_stmt field.
22264 (find_data_references_in_stmt): Update call to create_data_ref.
22265 (graphite_find_data_references_in_stmt): Likewise.
22266 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
22267 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
22268 (vect_record_base_alignment): New function.
22269 (vect_record_base_alignments): Likewise.
22270 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
22271 for nested statements even if we fail to compute a misalignment.
22272 Use pooled base alignments for unconditional references.
22273 (vect_find_same_alignment_drs): Compare base addresses instead
22275 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
22276 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
22278 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22280 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
22281 Add an explicit name for the enum. Use auto_vec for slp_instances
22282 and grouped_stores.
22283 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
22285 (_bb_vec_info): Add a constructor and destructor.
22286 (vinfo_for_stmt): Return NULL for uids of -1 as well.
22287 (destroy_loop_vec_info): Delete.
22288 (vect_destroy_datarefs): Likewise.
22289 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
22290 (vec_info::vec_info): New function.
22291 (vec_info::~vec_info): Likewise.
22292 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
22293 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
22294 destroy_loop_vec_info.
22295 * tree-vect-loop.c (new_loop_vec_info): Replace with...
22296 (_loop_vec_info::_loop_vec_info): ...this.
22297 (destroy_loop_vec_info): Replace with...
22298 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
22299 the stmt_vec_infos. Leave handling of vec_info information to its
22300 destructor. Remove explicit vector releases.
22301 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
22302 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
22303 * tree-vect-slp.c (new_bb_vec_info): Replace with...
22304 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
22305 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
22306 (destroy_bb_vec_info): Replace with...
22307 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
22308 information to its destructor.
22309 (vect_slp_analyze_bb_1): Use new and delete instead of
22310 new_bb_vec_info and destroy_bb_vec_info.
22311 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
22314 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22316 * tree-data-ref.h (subscript): Add access_fn field.
22317 (data_dependence_relation): Add could_be_independent_p.
22318 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
22319 (same_access_functions): Move to tree-data-ref.c.
22320 * tree-data-ref.c (ref_contains_union_access_p): New function.
22321 (access_fn_component_p): Likewise.
22322 (access_fn_components_comparable_p): Likewise.
22323 (dr_analyze_indices): Add a reference to access_fn_component_p.
22324 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
22326 (constant_access_functions): Likewise.
22327 (add_other_self_distances): Likewise.
22328 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
22329 (initialize_data_dependence_relation): Use XCNEW and remove
22330 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
22331 of access functions that have the same type. Allow the
22332 subsequence to end with different bases in some circumstances.
22333 Record the chosen access functions in SUB_ACCESS_FN.
22334 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
22335 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
22336 (subscript_dependence_tester_1): Likewise dra and drb.
22337 (build_classic_dist_vector): Update calls accordingly.
22338 (subscript_dependence_tester): Likewise.
22339 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
22340 DDR_COULD_BE_INDEPENDENT_P.
22341 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
22342 comp_alias_ddrs instead of may_alias_ddrs.
22343 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
22345 (vect_analyze_data_ref_dependence): Use it if
22346 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
22347 distance vectors if that fails.
22348 (dependence_distance_ge_vf): New function.
22349 (vect_prune_runtime_alias_test_list): Use it. Don't clear
22350 LOOP_VINFO_MAY_ALIAS_DDRS.
22352 2017-08-04 Richard Biener <rguenther@suse.de>
22354 PR middle-end/81705
22355 * fold-const.c (fold_binary_loc): Properly restrict
22356 minus_var0 && minus_var1 case when associating undefined overflow
22359 2017-08-04 Tom de Vries <tom@codesourcery.com>
22361 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
22363 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22365 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22366 Don't start diagnostic messages with a capital letter.
22367 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22369 (rs6000_invalid_builtin): Likewise.
22370 (rs6000_trampoline_init): Likewise.
22372 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22375 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
22376 after setting changeable df flags.
22378 2017-08-03 Richard Biener <rguenther@suse.de>
22380 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
22381 up if the use is in USE - X.
22383 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
22385 * toplev.c (dumpfile.h): New include.
22386 (internal_error_reentered): New static function. Use it...
22387 (internal_error_function): ...here to handle reentered internal_error.
22389 2017-08-03 Richard Biener <rguenther@suse.de>
22391 PR middle-end/81148
22392 * fold-const.c (split_tree): Add minus_var and minus_con
22393 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
22394 here but always use minus_*.
22395 (associate_trees): Assert we never associate with MINUS_EXPR
22396 and NULL first operand. Do not recurse for PLUS_EXPR operands
22397 when associating as MINUS_EXPR either.
22398 (fold_binary_loc): Track minus_var and minus_con.
22400 2017-08-03 Tom de Vries <tom@codesourcery.com>
22403 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
22404 ACCEL_COMPILER, apply finish_options on
22405 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
22407 2017-08-03 Tom de Vries <tom@codesourcery.com>
22410 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
22411 function_entry_patch_area_size > 0.
22413 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22416 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
22417 instead of 10??LU, perform unit multiplication in wide_int,
22418 don't change alloc_object_size_limit if the limit is larger
22421 PR tree-optimization/81655
22422 PR tree-optimization/81588
22423 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
22424 the case when ranges[i].low and high are 1 for unsigned type with
22427 PR middle-end/81052
22428 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
22429 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
22431 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22433 * tree-vrp.h: Add include guard.
22435 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
22438 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
22439 (ud2): New insn pattern.
22440 * config/i386/i386.c (ix86_expand_epilogue):
22441 For naked functions, generate ud2 instead of trap insn.
22443 2017-08-02 Marek Polacek <polacek@redhat.com>
22446 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
22448 2017-08-02 Tom de Vries <tom@codesourcery.com>
22449 Cesar Philippidis <cesar@codesourcery.com>
22451 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
22452 Add missing edge probabilities.
22454 2017-08-02 Tamar Christina <tamar.christina@arm.com>
22456 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
22457 Correct endianness.
22459 2017-08-02 Jakub Jelinek <jakub@redhat.com>
22461 PR middle-end/79499
22462 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
22463 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
22464 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
22465 prologue_seq sequences - if any.
22467 2017-08-02 Richard Biener <rguenther@suse.de>
22469 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
22470 via vectors if supported, integer extracts via punning if supported
22471 or otherwise vector extracts.
22473 2017-08-02 Richard Biener <rguenther@suse.de>
22475 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
22477 (bitmap_insert_into_set): ... this.
22479 2017-08-02 Richard Biener <rguenther@suse.de>
22481 PR tree-optimization/81633
22483 2015-08-17 Alan Hayward <alan.hayward@arm.com>
22485 PR tree-optimization/71752
22486 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
22488 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
22490 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
22491 (machine_function::call_ms2sysv_pad_out): Remove field.
22492 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
22493 (ix86_compute_frame_layout): Likewise.
22495 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
22498 * config/i386/i386.c (ix86_set_func_type): Disallow naked
22499 attribute with interrupt attribute.
22501 2017-08-01 Andrew Pinski <apinski@cavium.com>
22503 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
22504 BIT_INSERT_EXPR's operand 1
22505 to see if the types precision matches.
22507 2017-08-01 Martin Liska <mliska@suse.cz>
22509 PR middle-end/70140
22510 * builtins.c (expand_builtin_memcpy_args): Remove.
22511 (expand_builtin_memcpy): Call newly added function
22512 expand_builtin_memory_copy_args.
22513 (expand_builtin_memcpy_with_bounds): Likewise.
22514 (expand_builtin_mempcpy): Remove last argument.
22515 (expand_builtin_mempcpy_with_bounds): Likewise.
22516 (expand_builtin_memory_copy_args): New function created from
22517 expand_builtin_mempcpy_args with small modifications.
22518 (expand_builtin_mempcpy_args): Remove.
22519 (expand_builtin_stpcpy): Remove unused argument.
22520 (expand_builtin): Likewise.
22521 (expand_builtin_with_bounds): Likewise.
22523 2017-08-01 Martin Liska <mliska@suse.cz>
22526 Make mempcpy more optimal (PR middle-end/70140).
22528 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22531 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
22532 __builtin_vec_cmpne verify both arguments are compatible vectors
22533 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
22534 verify arg1_type is a pointer or array type. For __builtin_vec_st,
22535 move computation of aligned to after checking the argument types.
22539 * config/rs6000/vsx.md (vextract_fp_from_shorth,
22540 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
22543 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
22544 Doug Rupp <rupp@adacore.com>
22545 Olivier Hainque <hainque@adacore.com>
22547 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
22548 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
22550 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
22551 for TARGET_OS_CPP_BUILTIN.
22552 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
22553 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
22555 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
22556 passing required abi options to the assembler for EABI configurations.
22557 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
22558 of .text.hot and .text.unlikely sections for kernel modules when
22559 using ARM style exceptions.
22560 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
22561 options. Add EXTRA_CC1_SPEC.
22562 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
22564 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
22566 (ARM_TARGET2_DWARF_FORMAT): Define.
22567 * config/arm/t-vxworks: Adjust multilib control to removal of the
22568 Diab command line options.
22570 2017-08-01 Martin Liska <mliska@suse.cz>
22572 PR gcov-profile/81561
22573 * gcov.c (unblock): Make unblocking safe as we need to preserve
22574 index correspondence of blocks and block_lists.
22576 2017-08-01 Richard Biener <rguenther@suse.de>
22578 PR tree-optimization/81181
22579 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
22580 (compute_antic): ... end of iteration here.
22582 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
22584 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
22585 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
22586 (ftree-slp-vectorize): Likewise.
22587 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
22588 can no longer be set independent of flag_tree_loop_vectorize.
22589 * omp-general.c (emp_max_vf): Likewise.
22590 * opts.c (enable_fdo_optimizations): Remove references to
22591 flag_tree_vectorize, these are now implicit.
22592 (common_handle_option): Remove handling for OPT_ftree_vectorize,
22593 and leave it for the options machinery.
22595 2017-08-01 Martin Liska <mliska@suse.cz>
22597 PR middle-end/70140
22598 * builtins.c (expand_builtin_memcpy_args): Remove.
22599 (expand_builtin_memcpy): Call newly added function
22600 expand_builtin_memory_copy_args.
22601 (expand_builtin_memcpy_with_bounds): Likewise.
22602 (expand_builtin_mempcpy): Remove last argument.
22603 (expand_builtin_mempcpy_with_bounds): Likewise.
22604 (expand_builtin_memory_copy_args): New function created from
22605 expand_builtin_mempcpy_args with small modifications.
22606 (expand_builtin_mempcpy_args): Remove.
22607 (expand_builtin_stpcpy): Remove unused argument.
22608 (expand_builtin): Likewise.
22609 (expand_builtin_with_bounds): Likewise.
22611 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22614 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
22615 print "ds:" only for immediates in generic address space.
22617 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22620 * config/i386/i386.c (ix86_funciton_naked): New prototype.
22621 (ix86_function_ok_for_sibcall): Return false for naked functions.
22623 2017-08-01 Richard Biener <rguenther@suse.de>
22625 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
22626 (compute_antic): Seed worklist with exit block predecessors.
22627 * cfganal.c (dfs_find_deadend): For a cycle return the source
22628 of the edge closing it.
22630 2017-08-01 Tamar Christina <tamar.christina@arm.com>
22632 * config/aarch64/aarch64.c
22633 (aarch64_can_const_movi_rtx_p): Move 0 check.
22635 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22637 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
22638 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
22640 * match.pd: Ditto in address comparison pattern.
22642 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22644 PR tree-optimization/81627
22645 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
22646 closed ssa form for store-store chain.
22648 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22650 PR tree-optimization/81620
22651 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
22652 for store-store chain.
22654 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22656 PR tree-optimization/81588
22657 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
22658 ranges[i].in_p, invert comparison code ccode. For >/>=,
22659 swap rhs1 and rhs2 and comparison code unconditionally,
22660 for </<= don't do that. Don't swap rhs1/rhs2 again if
22661 ranges[i].in_p, instead invert comparison code ccode if
22662 opcode or oe->rank is BIT_IOR_EXPR.
22665 * optabs.def (vec_extract_optab, vec_init_optab): Change from
22666 a direct optab to conversion optab.
22667 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
22668 with GET_MODE_INNER as last argument instead of optab_handler.
22669 * expmed.c (extract_bit_field_1): Likewise. Use vector from
22670 vector extraction if possible and optab is available.
22671 * expr.c (store_constructor): Use convert_optab_handler instead
22672 of optab_handler. Use vector initialization from smaller
22673 vectors if possible and optab is available.
22674 * tree-vect-stmts.c (vectorizable_load): Likewise.
22675 * doc/md.texi (vec_extract, vec_init): Document that the optabs
22676 now have two modes.
22677 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
22678 of vec_init from half-sized vectors with the same element mode.
22679 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
22680 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
22681 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
22682 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
22683 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
22684 after mode in gen_vec_extract* calls.
22685 (vec_extract<mode>): Renamed to ...
22686 (vec_extract<mode><ssescalarmodelower>): ... this.
22687 (vec_extract<mode><ssehalfvecmodelower>): New expander.
22688 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
22689 element mode after mode in gen_vec_init* calls.
22690 (VEC_INIT_HALF_MODE): New mode iterator.
22691 (vec_init<mode>): Renamed to ...
22692 (vec_init<mode><ssescalarmodelower>): ... this.
22693 (vec_init<mode><ssehalfvecmodelower>): New expander.
22694 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
22695 (vec_extractv2sfsf): ... this.
22696 (vec_initv2sf): Renamed to ...
22697 (vec_initv2sfsf): ... this.
22698 (vec_extractv2si): Renamed to ...
22699 (vec_extractv2sisi): ... this.
22700 (vec_initv2si): Renamed to ...
22701 (vec_initv2sisi): ... this.
22702 (vec_extractv4hi): Renamed to ...
22703 (vec_extractv4hihi): ... this.
22704 (vec_initv4hi): Renamed to ...
22705 (vec_initv4hihi): ... this.
22706 (vec_extractv8qi): Renamed to ...
22707 (vec_extractv8qiqi): ... this.
22708 (vec_initv8qi): Renamed to ...
22709 (vec_initv8qiqi): ... this.
22710 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
22711 (vec_init<mode>): Renamed to ...
22712 (vec_init<mode><VEC_base_l>): ... this.
22713 (vec_extract<mode>): Renamed to ...
22714 (vec_extract<mode><VEC_base_l>): ... this.
22715 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
22716 (vec_initv2sfsf): ... this.
22717 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
22718 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22719 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
22720 element mode after mode in gen_vec_init* calls.
22721 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
22722 (vec_init<mode><Vel>): ... this.
22723 (vec_extract<mode>): Renamed to ...
22724 (vec_extract<mode><Vel>): ... this.
22725 * config/aarch64/iterators.md (Vel): New mode attribute.
22726 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
22727 Add element mode after mode in gen_vec_extract* calls.
22728 * config/s390/vector.md (non_vec_l): New mode attribute.
22729 (vec_extract<mode>): Renamed to ...
22730 (vec_extract<mode><non_vec_l>): ... this.
22731 (vec_init<mode>): Renamed to ...
22732 (vec_init<mode><non_vec_l>): ... this.
22733 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
22734 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
22736 * config/arm/iterators.md (V_elem_l): New mode attribute.
22737 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
22738 (vec_extract<mode><V_elem_l>): ... this.
22739 (vec_extractv2di): Renamed to ...
22740 (vec_extractv2didi): ... this.
22741 (vec_init<mode>): Renamed to ...
22742 (vec_init<mode><V_elem_l>): ... this.
22743 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
22744 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
22745 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
22746 Add element mode after gen_vec_extract* calls.
22747 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
22748 (vec_init<mode><unitmode>): ... this.
22749 (vec_extract<mode>): Renamed to ...
22750 (vec_extract<mode><unitmode>): ... this.
22751 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
22752 (vec_init<mode><unitmode>): ... this.
22753 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
22754 (vec_initv2sfsf): ... this.
22755 (vec_extractv2sf): Renamed to ...
22756 (vec_extractv2sfsf): ... this.
22757 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
22758 Add element mode after gen_vec_extract* calls.
22759 * config/mips/mips.md (unitmode): New mode iterator.
22760 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
22761 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
22762 * config/spu/spu.md (inner_l): New mode attribute.
22763 (vec_init<mode>): Renamed to ...
22764 (vec_init<mode><inner_l>): ... this.
22765 (vec_extract<mode>): Renamed to ...
22766 (vec_extract<mode><inner_l>): ... this.
22767 * config/sparc/sparc.md (veltmode): New mode iterator.
22768 (vec_init<VMALL:mode>): Renamed to ...
22769 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
22770 * config/ia64/vect.md (vec_initv2si): Renamed to ...
22771 (vec_initv2sisi): ... this.
22772 (vec_initv2sf): Renamed to ...
22773 (vec_initv2sfsf): ... this.
22774 (vec_extractv2sf): Renamed to ...
22775 (vec_extractv2sfsf): ... this.
22776 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
22777 (vec_init<mode>): Renamed to ...
22778 (vec_init<mode><VEC_base_l>): ... this.
22779 (vec_extract<mode>): Renamed to ...
22780 (vec_extract<mode><VEC_base_l>): ... this.
22781 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
22782 (vec_initv2sfsf): ... this.
22783 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
22784 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22785 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
22786 gen_vec_init* calls.
22788 2017-08-01 Richard Biener <rguenther@suse.de>
22790 PR tree-optimization/81297
22791 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
22792 TREE_OVERFLOW from INTEGER_CSTs.
22794 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
22796 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
22798 2017-07-31 Carl Love <cel@us.ibm.com>
22800 * config/rs6000/rs6000-c: Add support for built-in functions
22801 vector signed char vec_xl_be (signed long long, signed char *);
22802 vector unsigned char vec_xl_be (signed long long, unsigned char *);
22803 vector signed int vec_xl_be (signed long long, signed int *);
22804 vector unsigned int vec_xl_be (signed long long, unsigned int *);
22805 vector signed long long vec_xl_be (signed long long, signed long long *);
22806 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
22807 vector signed short vec_xl_be (signed long long, signed short *);
22808 vector unsigned short vec_xl_be (signed long long, unsigned short *);
22809 vector double vec_xl_be (signed long long, double *);
22810 vector float vec_xl_be (signed long long, float *);
22811 * config/rs6000/altivec.h (vec_xl_be): Add #define.
22812 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
22813 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
22815 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
22816 (altivec_expand_builtin): Add switch statement to call
22817 altivec_expand_xl_be for each builtin.
22818 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
22819 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
22820 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
22821 __builtin_vsx_le_be_v16qi.
22822 * doc/extend.texi: Update the built-in documentation file for the
22823 new built-in functions.
22825 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22828 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
22830 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22832 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22834 * config.gcc: Add z14.
22835 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
22836 CPU model numbers for z13s and z14.
22837 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
22839 * config/s390/s390-opts.h (enum processor_type): Rename
22840 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22841 * config/s390/s390.c (processor_table): Add field for CPU name to
22842 be passed to Binutils.
22843 (s390_asm_output_machine_for_arch): Use the new field in
22844 processor_table for Binutils.
22845 (s390_expand_builtin): Replace arch12 with z14.
22846 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22847 (s390_get_sched_attrmask): Likewise.
22848 (s390_get_unit_mask): Likewise.
22849 * config/s390/s390.opt: Add z14 to processor_type enum.
22851 2017-07-31 Martin Jambor <mjambor@suse.cz>
22854 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22855 regardless of optimization level.
22857 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22858 Martin Liska <mliska@suse.cz>
22860 * predict.def: Remove old comment and adjust probability.
22861 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22862 PREDICT statements.
22864 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22867 * config/i386/i386.c (ix86_function_naked): New function.
22868 (ix86_can_use_return_insn_p): Return false for naked functions.
22869 (ix86_expand_prologue): Skip prologue for naked functions.
22870 (ix86_expand_epilogue): Skip epilogue for naked functions
22871 and emit trap instruction.
22872 (ix86_warn_func_return): New function.
22873 (ix86_attribute_table): Add "naked" attribute specification.
22874 (TARGET_WARN_FUNC_RETURN): Define.
22875 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22877 2017-07-31 Martin Liska <mliska@suse.cz>
22879 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22880 (dump_gimple_bb_header): Always dump BB info.
22881 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22883 2017-07-31 Martin Liska <mliska@suse.cz>
22886 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22887 also with current_function_decl non-null equality.
22889 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22892 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22893 change type to the element type, instead add eltype variable and
22894 use it where we are interested in the element type.
22896 PR tree-optimization/81603
22897 * ipa-polymorphic-call.c
22898 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22899 offset arithmetic in offset_int, bail out if the resulting bit offset
22900 doesn't fit into shwi.
22902 2017-07-31 Martin Liska <mliska@suse.cz>
22904 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22905 (gimplify_save_expr): Fix comment.
22907 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22910 * config/i386/i386.c (ix86_function_arg): Update arguments for
22912 (ix86_compute_frame_layout): Set the initial stack offset to
22913 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22914 INCOMING_FRAME_SP_OFFSET.
22915 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22916 stack before exception handler returns.
22917 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22918 the 'ERROR_CODE' for exception handler.
22920 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22922 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22923 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22924 (ASM_OUTPUT_REG_POP): Ditto.
22925 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22926 instead of asm_fprintf to output pure string.
22928 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22930 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22931 to imported_module_or_decl hook.
22932 (debug_nothing_tree_tree_tree_bool): Remove.
22933 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22934 * debug.c (do_nothing_debug_hooks): Use
22935 debug_nothing_tree_tree_tree_bool_bool instead of
22936 debug_nothing_tree_tree_tree_bool.
22937 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22938 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22939 * sdbout.c (sdb_debug_hooks): Likewise.
22940 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22941 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22943 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22944 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22945 attribute, don't add anything.
22947 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22949 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22950 (fold_build2_stat_loc): Likewise.
22951 (fold_build3_stat_loc): Likewise.
22952 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22953 (fold_build1_loc): Remove macro.
22954 (fold_build2_loc): Likewise.
22955 (fold_build3_loc): Likewise.
22957 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22959 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22960 (gimple_build_debug_bind_source_stat): Likewise.
22961 * gimple.h (gimple_build_debug_bind): Remove macro.
22962 (gimple_build_debug_bind_source): Likewise.
22964 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22966 * bitmap.c (bitmap_alloc): Adjust.
22967 (bitmap_gc_alloc): Likewise.
22968 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22970 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22972 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22973 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22974 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22975 (bitmap_gc_alloc_stat): Likewise.
22976 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22978 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22980 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22981 * rtl.h (shallow_copy_rtx): Remove macro.
22983 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22985 * emit-rtl.c (gen_raw_REG): Adjust.
22986 * gengenrtl.c (gendef): Likewise.
22987 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22988 * rtl.h (rtx_alloc): Remove macro.
22990 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22992 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22993 (build_tree_list_stat): Likewise.
22994 * tree.h (build_tree_list): Remove macro.
22995 (build_tree_list_vec): Likewise.
22997 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22999 * tree.c (make_vector_stat): Remove _stat from name.
23000 (build_vector_stat): Likewise.
23001 * tree.h (make_vector_stat): Remove macro.
23002 (build_vector_stat): Likewise.
23004 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23006 * tree.h (build_var_debug_value): Remove prototype.
23008 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23010 * tree.c (tree_cons_stat): Remove _stat from name.
23011 * tree.h (tree_cons): Remove macro.
23013 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23015 * tree.c (build_vl_exp_stat): Remove _stat from name.
23016 * tree.h (build_vl_exp): Remove macro.
23018 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23020 * tree.c (build_decl_stat): Remove _stat from name.
23021 * tree.h (build_decl): Remove macro.
23023 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23025 * gimple.c (gimple_build_with_ops_stat): Adjust.
23026 (gimple_alloc_stat): Remove _stat from name.
23027 * gimple.h (gimple_alloc): Remove macro.
23029 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23031 * tree.c (make_tree_vec_stat): Remove _stat from name.
23032 (grow_tree_vec_stat): Likewise.
23033 * tree.h (make_tree_vec_stat): Adjust prototype.
23034 (grow_tree_vec_stat): Likewise.
23035 (make_tree_vec): Remove macro.
23036 (grow_tree_vec): Likewise.
23038 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23040 * fold-const.c (fold_build1_stat_loc): Adjust.
23041 (fold_build2_stat_loc): Likewise.
23042 (fold_build3_stat_loc): Likewise.
23043 * tree.c (build0_stat): Remove _stat from name.
23044 (build1_stat): Likewise.
23045 (build2_stat): Likewise.
23046 (build3_stat): Likewise.
23047 (build4_stat): Likewise.
23048 (build5_stat): Likewise.
23049 * tree.h (build1_loc): Remove macro, and rename _stat function
23051 (build2_loc): Likewise.
23052 (build3_loc): Likewise.
23053 (build4_loc): Likewise.
23054 (build5_loc): Likewise.
23056 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23058 * tree.c (make_int_cst_stat): Remove _stat from name.
23059 * tree.h (make_int_cst_stat): Adjust prototype.
23060 (make_int_cst): Remove macro.
23062 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23064 * tree.c (make_tre_binfo_stat): Remove _stat from name.
23065 * tree.h (make_tree_binfo_stat): Adjust prototype.
23066 (make_tree_binfo): Remove.
23068 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23070 * tree.c (copy_node_stat): Rename to copy_node.
23071 (build_distinct_type_copy): Adjust.
23072 * tree.h (copy_node_stat): Adjust prototype.
23073 (copy_node): Remove macro.
23075 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23077 * tree.c (make_node_stat): rename to make_node.
23078 (build_tree_list_stat): Adjust.
23079 (build0_stat): Likewise.
23080 (build2_stat): Likewise.
23081 (build3_stat): Likewise.
23082 (build4_stat): Likewise.
23083 (build5_stat): Likewise.
23084 (build_decl_stat): Likewise.
23085 * tree.h (make_node_stat): Adjust prototype.
23086 (make_node): remove macro.
23088 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
23090 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
23091 (PPC_FEATURE2_SCV): Likewise.
23092 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
23094 2017-07-28 Tamar Christina <tamar.christina@arm.com>
23096 * config/aarch64/aarch64.c
23097 (aarch64_internal_mov_immediate): Add new special pattern.
23098 * config/aarch64/aarch64.md (*movdi_aarch64):
23099 Add reg/32bit const mov case.
23101 2017-07-28 Tamar Christina <tamar.christina@arm.com>
23102 Richard Sandiford <richard.sandiford@linaro.org>
23104 * config/aarch64/aarch64.md (mov<mode>): Generalize.
23105 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
23106 Add integer and movi cases.
23107 (movi-split-hf-df-sf split, fp16): New.
23108 (enabled): Added TARGET_FP_F16INST.
23109 * config/aarch64/iterators.md (GPF_HF): New.
23110 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
23112 2017-07-28 Tamar Christina <tamar.christina@arm.com>
23114 * config/aarch64/aarch64.c
23115 (aarch64_simd_container_mode): Add prototype.
23116 (aarch64_expand_mov_immediate): Add HI support.
23117 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
23118 (aarch64_can_const_movi_rtx_p): New.
23119 (aarch64_preferred_reload_class):
23120 Remove restrictions of using FP registers for certain SIMD operations.
23121 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
23122 (aarch64_valid_floating_const): Add integer move validation.
23123 (aarch64_simd_imm_scalar_p): Remove.
23124 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
23125 (aarch64_legitimate_constant_p): Expand list of supported cases.
23126 * config/aarch64/aarch64-protos.h
23127 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
23128 (aarch64_reinterpret_float_as_int): New.
23129 (aarch64_simd_imm_scalar_p): Remove.
23130 * config/aarch64/constraints.md (Uvi): New.
23131 (Dd): Split into Ds and new Dd.
23132 * config/aarch64/aarch64.md (*movsi_aarch64):
23134 (*movdi_aarch64): Add SIMD mov case.
23136 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23138 * tree-predcom.c: (struct chain): Handle store-store chain in which
23139 stores for elimination only store loop invariant values.
23140 (execute_pred_commoning_chain): Ditto.
23141 (prepare_initializers_chain_store_elim): Ditto.
23142 (prepare_finalizers): Ditto.
23143 (is_inv_store_elimination_chain): New function.
23144 (initialize_root_vars_store_elim_1): New function.
23146 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23148 * tree-predcom.c: Revise general description of the pass.
23149 (enum chain_type): New enum type for store elimination.
23150 (struct chain): New field supporting store elimination.
23151 (struct component): Ditto.
23152 (dump_chain): Dump store-stores chain.
23153 (release_chain): Release resources.
23154 (split_data_refs_to_components): Compute and create component
23155 contains only stores for elimination.
23156 (get_chain_last_ref_at): New function.
23157 (make_invariant_chain): Initialization.
23158 (make_rooted_chain): Specify chain type in parameter and record it.
23159 (add_looparound_copies): Skip for store-stores chain.
23160 (determine_roots_comp): Compute type of chain and pass it to
23162 (initialize_root_vars_store_elim_2): New function.
23163 (finalize_eliminated_stores): New function.
23164 (remove_stmt): Handle store for elimination.
23165 (execute_pred_commoning_chain): Execute predictive commoning on
23166 store-store chains.
23167 (determine_unroll_factor): Skip unroll for store-stores chain.
23168 (prepare_initializers_chain_store_elim): New function.
23169 (prepare_initializers_chain): Hanlde store-store chain.
23170 (prepare_finalizers_chain, prepare_finalizers): New function.
23171 (tree_predictive_commoning_loop): Return integer value indicating
23172 if loop is unrolled or lcssa form is corrupted.
23173 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
23175 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23177 * tree-predcom.c (initialize_root): Delete.
23178 (execute_pred_commoning_chain): Initialize root vars and replace
23179 reference of non-combined chain directly, rather than call above
23182 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23184 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
23185 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
23187 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23189 * tree-predcom.c (struct chain): New field init_seq.
23190 (release_chain): Release init_seq.
23191 (prepare_initializers_chain): Record intialization stmts in above
23193 (insert_init_seqs): New function.
23194 (tree_predictive_commoning_loop): Call insert_init_seqs.
23196 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23198 * tree-predcom.c (determine_roots_comp): Skip trivial components.
23200 2017-07-28 Richard Biener <rguenther@suse.de>
23202 * match.pd: Remove superfluous :c.
23203 * genmatch.c (simplify::id): Add member.
23204 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
23206 (current_id): New global.
23207 (dt_node::parent): Move from ...
23208 (dt_operand::parent): ... here. Add for_id member.
23209 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
23210 (decision_tree::find_node): Relax order requirement when
23211 merging DT_TRUE nodes to ones inbetween the current simplify
23212 and the one we try to merge with. Add diagnostic whenever
23213 we need to enforce pattern order by not merging.
23214 (decision_tree::insert): Set current_id.
23215 (decision_tree::print_node): Dump parent node and for_id.
23216 (parser::last_id): Add member.
23217 (parser::push_simplify): Assign unique id.
23218 (parser::parser): Initialize last_id.
23220 2017-07-28 Martin Liska <mliska@suse.cz>
23223 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
23224 gimple_build_debug_bind.
23226 2017-07-28 Richard Biener <rguenther@suse.de>
23228 PR tree-optimization/81502
23229 * match.pd: Add pattern combining BIT_INSERT_EXPR with
23231 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
23233 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
23234 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
23235 for BIT_FIELD_REF args.
23236 * fold-const.c (make_bit_field_ref): Likewise.
23237 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
23239 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23242 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
23243 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
23244 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
23245 Or it into SANITIZER_UNDEFINED.
23246 * ubsan.c: Include gimple-fold.h and varasm.h.
23247 (ubsan_expand_ptr_ifn): New function.
23248 (instrument_pointer_overflow): New function.
23249 (maybe_instrument_pointer_overflow): New function.
23250 (instrument_object_size): Formatting fix.
23251 (pass_ubsan::execute): Call instrument_pointer_overflow
23252 and maybe_instrument_pointer_overflow.
23253 * internal-fn.c (expand_UBSAN_PTR): New function.
23254 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
23255 * sanitizer.def (__ubsan_handle_pointer_overflow,
23256 __ubsan_handle_pointer_overflow_abort): New builtins.
23257 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
23258 * internal-fn.def (UBSAN_PTR): New internal function.
23259 * opts.c (sanitizer_opts): Add pointer-overflow.
23260 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
23261 * fold-const.c (build_range_check): Compute pointer range check in
23262 integral type if pointer arithmetics would be needed. Formatting
23265 2017-07-28 Martin Liska <mliska@suse.cz>
23268 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
23269 parameters that are of a variable-length.
23271 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
23273 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
23275 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
23276 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
23277 (CRT_CALL_STATIC_FUNCTION): Likewise.
23278 (ASM_DEFAULT_SPEC): New define.
23279 (ASM_SPEC32): Likewise.
23280 (ASM_SPEC64): Likewise.
23281 (ASM_SPEC_COMMON): Likewise.
23282 (ASM_SPEC): Likewise.
23283 (INVALID_64BIT): Likewise.
23284 (LINK_OS_DEFAULT_SPEC): Likewise.
23285 (LINK_OS_SPEC32): Likewise.
23286 (LINK_OS_SPEC64): Likewise.
23287 (POWERPC_LINUX): Likewise.
23288 (PTRDIFF_TYPE): Likewise.
23289 (RESTORE_FP_PREFIX): Likewise.
23290 (RESTORE_FP_SUFFIX): Likewise.
23291 (SAVE_FP_PREFIX): Likewise.
23292 (SAVE_FP_SUFFIX): Likewise.
23293 (SIZE_TYPE): Likewise.
23294 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
23295 (TARGET_64BIT): Likewise.
23296 (TARGET_64BIT): Likewise.
23297 (TARGET_AIX): Likewise.
23298 (WCHAR_TYPE_SIZE): Likewise.
23299 (WCHAR_TYPE): Undefine.
23300 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
23301 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
23302 (CPP_OS_RTEMS_SPEC): Delete.
23303 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
23304 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
23306 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
23308 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23310 PR tree-optimization/81578
23311 * tree-parloops.c (build_new_reduction): Bail out if
23312 reduction_code isn't one of the standard OpenMP reductions.
23313 Move the details printing after that decision.
23315 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23317 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
23318 related to reload_in_progress.
23319 (splat_input_operand): Likewise.
23320 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
23322 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
23324 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
23325 (TARGET_INSTANTIATE_DECLS): Likewise.
23326 (legitimate_indexed_address_p): Delete reload_in_progress code.
23327 (rs6000_debug_legitimate_address_p): Likewise.
23328 (rs6000_eliminate_indexed_memrefs): Likewise.
23329 (rs6000_emit_le_vsx_store): Likewise.
23330 (rs6000_emit_move_si_sf_subreg): Likewise.
23331 (rs6000_emit_move): Likewise.
23332 (register_to_reg_type): Likewise.
23333 (rs6000_pre_atomic_barrier): Likewise.
23334 (rs6000_machopic_legitimize_pic_address): Likewise.
23335 (rs6000_allocate_stack_temp): Likewise.
23336 (rs6000_address_for_fpconvert): Likewise.
23337 (rs6000_address_for_altivec): Likewise.
23338 (rs6000_secondary_memory_needed_rtx): Delete function.
23339 (rs6000_check_sdmode): Likewise.
23340 (rs6000_alloc_sdmode_stack_slot): Likewise.
23341 (rs6000_instantiate_decls): Likewise.
23342 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
23343 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
23344 Delete reload_in_progress.
23345 (*vec_reload_and_plus_<mptrsize>): Likewise.
23346 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
23347 (vsx_div_v2di): Likewise.
23348 (vsx_udiv_v2di): Likewise.
23350 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23352 * config/rs6000/rs6000.opt (mlra): Replace with stub.
23353 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
23354 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
23355 (rs6000_debug_reg_global): Delete print of LRA status.
23356 (rs6000_option_override_internal): Delete dead LRA related code.
23357 (rs6000_lra_p): Delete function.
23358 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
23360 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
23362 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
23363 * config/riscv/rtems.h: New file.
23365 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23366 Sudakshina Das <sudi.das@arm.com>
23368 * config/aarch64/aarch64.md
23369 (define_split for and<mode>3nr_compare): Move
23370 non aarch64_logical_operand to a register.
23371 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
23372 register immediate operand to a register.
23373 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
23375 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23377 PR middle-end/81564
23378 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
23380 2017-07-27 Richard Biener <rguenther@suse.de>
23382 PR tree-optimization/81573
23383 PR tree-optimization/81494
23384 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
23385 multi defuse cycle case.
23387 2017-07-27 Richard Biener <rguenther@suse.de>
23389 PR tree-optimization/81571
23390 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
23393 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
23395 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
23396 earlier and only if MASK_FPU is set. Adjust formatting.
23398 2017-07-27 Martin Liska <mliska@suse.cz>
23400 * opt-functions.awk: Add validation of value of Init.
23401 * optc-gen.awk: Pass new argument.
23403 2017-07-27 Martin Liska <mliska@suse.cz>
23405 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
23406 Fix wrong condition.
23408 2017-07-27 Martin Liska <mliska@suse.cz>
23410 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
23411 BBs and edges seen by autoFDO.
23413 2017-07-27 Richard Biener <rguenther@suse.de>
23415 PR tree-optimization/81502
23416 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
23417 with incompatible but same sized type.
23418 (execute_update_addresses_taken): Likewise.
23420 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
23422 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
23423 flag_tree_loop_vectorize rather than flag_tree_vectorize.
23425 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23428 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
23429 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
23430 Change s_operand to memory_operand.
23432 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
23434 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
23435 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
23436 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
23437 Emit instructions rather than returning an expression. Handle TFmode
23438 and KFmode by casting to TImode.
23439 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
23440 (rs6000_emit_le_vsx_store): Likewise.
23441 * config/rs6000/vsx.md (VSX_TI): New iterator.
23442 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
23443 (*vsx_le_undo_permute_<mode>): Likewise.
23444 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
23445 emit the split sequence.
23446 (*vsx_le_perm_store_<mode>): Likewise.
23448 2017-07-27 Jakub Jelinek <jakub@redhat.com>
23450 PR tree-optimization/81555
23451 PR tree-optimization/81556
23452 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
23453 if true, force CHANGED for the recursive invocation.
23454 (reassociate_bb): Remember original length of ops array, pass
23455 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
23457 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
23458 attributes for noipa attribute. For naked attribute use
23459 lookup_attribute first before lookup_attribute_spec.
23460 * final.c (rest_of_handle_final): Disable IPA RA for functions with
23462 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
23463 for functions with noipa attribute.
23464 (cgraph_externally_visible_p): Return true for functions with noipa
23466 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
23467 for functions with noipa attribute.
23468 * doc/extend.texi: Document noipa function attribute.
23469 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
23470 also for functions with noipa attribute.
23471 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
23473 2017-07-26 Andrew Pinski <apinski@cavium.com>
23475 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
23476 vec_unalign_load_cost and vec_unalign_store_cost.
23478 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
23480 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
23481 -mvsx-small-integer option.
23482 (ISA_3_0_MASKS_IEEE): Likewise.
23483 (OTHER_VSX_VECTOR_MASKS): Likewise.
23484 (POWERPC_MASKS): Likewise.
23485 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
23486 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
23487 code, only testing for DImode being allowed in non-VSX floating
23489 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
23490 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
23492 (rs6000_option_override_internal): Delete -mvsx-small-integer.
23493 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
23494 TARGET_P8_VECTOR test.
23495 (rs6000_secondary_reload_simple_move): Likewise.
23496 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
23497 since TARGET_P9_VECTOR was already tested.
23498 (rs6000_opt_masks): Remove -mvsx-small-integer.
23499 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
23500 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23502 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
23503 test for TARGET_VEXTRACTUB was used, and that uses
23505 (p9 extract splitter): Likewise.
23506 (vsx_extract_<mode>_di_p9): Likewise.
23507 (vsx_extract_<mode>_store_p9): Likewise.
23508 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
23509 for TARGET_P9_VECTOR was used. Delete code that is now dead with
23510 the elimination of TARGET_VSX_SMALL_INTEGER.
23511 (vsx_extract_<mode>_p8): Likewise.
23512 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
23513 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
23514 (vsx_set_<mode>_p9): Likewise.
23515 (vsx_set_v4sf_p9): Likewise.
23516 (vsx_set_v4sf_p9_zero): Likewise.
23517 (vsx_insert_extract_v4sf_p9): Likewise.
23518 (vsx_insert_extract_v4sf_p9_2): Likewise.
23519 * config/rs6000/rs6000.md (sign extend splitter): Change
23520 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
23521 (floatsi<mode>2_lfiwax_mem): Likewise.
23522 (floatunssi<mode>2_lfiwzx_mem): Likewise.
23523 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
23524 since a test for TARGET_P9_VECTOR was used.
23525 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23526 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
23527 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23528 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23529 TARGET_P8_VECTOR test.
23530 (fix_trunc<mode>si2_stfiwx): Likewise.
23531 (fix_trunc<mode>si2_internal): Likewise.
23532 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
23533 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23535 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23536 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23537 TARGET_P8_VECTOR test.
23538 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23539 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
23540 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23542 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23543 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
23544 since a test for TARGET_P9_VECTOR was used.
23545 (splitter for loading small constants): Likewise.
23547 2017-07-26 Andrew Pinski <apinski@cavium.com>
23549 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
23552 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
23555 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
23556 (fp_valid_at): Likewise.
23558 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23560 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
23561 (qdf24xx_addrcost_table): Likewise.
23562 (cortexa57_tunings): Update to use generic_branch_cost.
23563 (cortexa72_tunings): Likewise.
23564 (cortexa73_tunings): Likewise.
23565 (qdf24xx_tunings): Likewise.
23567 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23569 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
23570 (thunderx2t99_branch_cost): Likewise.
23571 (cortexa35_tunings): Update to use generic_branch_cost.
23572 (cortexa53_tunings): Likewise.
23573 (cortexa57_tunings): Likewise.
23574 (cortexa72_tunings): Likewise.
23575 (cortexa73_tunings): Likewise.
23576 (thunderx2t99_tunings): Likewise.
23578 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23580 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
23581 (sparc_option_override): Honour MASK_FSMULD.
23582 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
23583 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
23584 * config/sparc/sparc.opt (mfsmuld): New option.
23585 * doc/invoke.texi (mfsmuld): Document option.
23587 2017-07-26 Marek Polacek <polacek@redhat.com>
23589 PR middle-end/70992
23590 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
23592 2017-07-26 Richard Biener <rguenther@suse.de>
23594 * gimple-match-head.c (do_valueize): Return OP if valueize
23596 (get_def): New helper to get at the def stmt of a SSA name
23597 if valueize allows.
23598 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
23599 do_valueize to get at the def stmt.
23600 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
23602 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
23604 PR middle-end/46932
23605 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
23607 2017-07-26 Martin Liska <mliska@suse.cz>
23610 * function.c (expand_function_start): Make expansion of
23611 nonlocal_goto_save_area after parm_birth_insn.
23613 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23615 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
23616 from all CPU target flags enable members.
23618 2017-07-26 Richard Biener <rguenther@suse.de>
23620 * genmatch.c (dt_simplify::gen): Make iterator vars const.
23621 (decision_tree::gen): Make 'type' const.
23622 (write_predicate): Likewise.
23624 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23626 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
23627 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
23628 (rs6000_option_override_internal): Likewise.
23629 (rs6000_expand_vector_set): Likewise.
23630 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
23631 (TARGET_UPPER_REGS_SF): Likewise.
23632 (TARGET_UPPER_REGS_DI): Likewise.
23633 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
23634 (TARGET_DIRECT_MOVE_64BIT): Likewise.
23635 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
23636 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23637 (Splitters for DI constants in Altivec registers): Likewise.
23638 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
23639 (vsx_set_v4sf_p9): Likewise.
23640 (vsx_set_v4sf_p9_zero): Likewise.
23641 (vsx_insert_extract_v4sf_p9): Likewise.
23642 (vsx_insert_extract_v4sf_p9_2): Likewise.
23644 2017-07-25 Carl Love <cel@us.ibm.com>
23646 * doc/extend.texi: Update the built-in documentation file for the
23647 existing built-in functions
23648 vector signed char vec_cnttz (vector signed char);
23649 vector unsigned char vec_cnttz (vector unsigned char);
23650 vector signed short vec_cnttz (vector signed short);
23651 vector unsigned short vec_cnttz (vector unsigned short);
23652 vector signed int vec_cnttz (vector signed int);
23653 vector unsigned int vec_cnttz (vector unsigned int);
23654 vector signed long long vec_cnttz (vector signed long long);
23655 vector unsigned long long vec_cnttz (vector unsigned long long);
23657 2017-07-25 Andrew Pinski <apinski@cavium.com>
23659 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
23660 accesses where the use is for the first operand of a BIT_INSERT.
23662 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23665 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23666 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23668 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23670 * config/i386/gstabs.h: Delete.
23671 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
23673 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
23675 * config/i386/i386.c (ix86_decompose_address): Do not check for
23676 register RTX when looking at index_reg or base_reg.
23677 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
23679 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
23681 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
23682 to update EH info here.
23684 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23686 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
23688 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23690 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
23692 2017-07-25 Torsten Duwe <duwe@suse.de>
23694 * common.opt: Introduce -fpatchable-function-entry
23695 command line option, and its variables function_entry_patch_area_size
23696 and function_entry_patch_area_start.
23697 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
23698 including a two-value parser.
23699 * target.def (print_patchable_function_entry): New target hook.
23700 * targhooks.h (default_print_patchable_function_entry): New function.
23701 * targhooks.c (default_print_patchable_function_entry): Likewise.
23702 * toplev.c (process_options): Switch off IPA-RA if
23703 patchable function entries are being generated.
23704 * varasm.c (assemble_start_function): Look at the
23705 patchable-function-entry command line switch and current
23706 function attributes and maybe generate NOP instructions by
23707 calling the print_patchable_function_entry hook.
23708 * doc/extend.texi: Document patchable_function_entry attribute.
23709 * doc/invoke.texi: Document -fpatchable_function_entry
23710 command line option.
23711 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
23713 * doc/tm.texi: Re-generate.
23715 2017-07-25 Jakub Jelinek <jakub@redhat.com>
23718 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
23719 TARGET_AVX512DQ rather than TARGET_AVX512BW.
23721 2017-07-25 Tamar Christina <tamar.christina@arm.com>
23723 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
23725 2017-07-25 Richard Biener <rguenther@suse.de>
23727 PR tree-optimization/81455
23728 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
23729 not walk in cycles when looking for guards.
23731 2017-07-25 Richard Biener <rguenther@suse.de>
23733 PR tree-optimization/81529
23734 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
23735 when optimizing backedge uses.
23737 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
23739 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
23741 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
23742 to dl_section_ref. On AIX, append an expression to subtract
23743 the size of the section length to dl_section_ref.
23745 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
23747 * configure.ac: If any of the config.* scripts fail, exit 1.
23748 * configure: Regenerate.
23750 2017-07-25 Richard Biener <rguenther@suse.de>
23752 PR middle-end/81546
23753 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
23754 of immediate uses, be more verbose on errors.
23756 2017-07-25 Richard Biener <rguenther@suse.de>
23758 PR tree-optimization/81510
23759 * tree-vect-loop.c (vect_is_simple_reduction): When the
23760 reduction stmt is not inside the loop bail out.
23762 2017-07-25 Richard Biener <rguenther@suse.de>
23764 PR tree-optimization/81303
23765 * tree-vect-loop-manip.c (vect_loop_versioning): Build
23766 profitability check against LOOP_VINFO_NITERSM1.
23768 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23770 * domwalk.c (cmp_bb_postorder): Simplify.
23771 (sort_bbs_postorder): New function. Use it...
23772 (dom_walker::walk): ...here to optimize common cases.
23774 2017-07-25 Martin Liska <mliska@suse.cz>
23777 * ipa-visibility.c (function_and_variable_visibility): Make the
23778 redirection just on target that supports aliasing.
23779 Fix GNU coding style.
23781 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
23784 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
23786 * config/arm/rtems.h: Likewise.
23787 * config/bfin/rtems.h: Likewise.
23788 * config/i386/rtemself.h: Likewise.
23789 * config/lm32/rtems.h: Likewise.
23790 * config/m32c/rtems.h: Likewise.
23791 * config/m68k/rtemself.h: Likewise.
23792 * config/microblaze/rtems.h: Likewise.
23793 * config/mips/rtems.h: Likewise.
23794 * config/moxie/rtems.h: Likewise.
23795 * config/nios2/rtems.h: Likewise.
23796 * config/powerpcspe/rtems.h: Likewise.
23797 * config/rs6000/rtems.h: Likewise.
23798 * config/rtems.h: Likewise.
23799 * config/sh/rtems.h: Likewise.
23800 * config/sh/rtemself.h: Likewise.
23801 * config/sparc/rtemself.h: Likewise.
23803 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23806 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
23807 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
23808 * tree-ssa-structalias.c (alias_get_name): Same.
23810 2017-07-25 Bin Cheng <bin.cheng@arm.com>
23813 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
23814 instructions if no du chain is found.
23816 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23818 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
23820 2017-07-25 Richard Biener <rguenther@suse.de>
23822 PR middle-end/81505
23823 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
23826 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23828 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
23829 upper-regs options.
23830 (ISA_2_7_MASKS_SERVER): Likewise.
23831 (ISA_3_0_MASKS_IEEE): Likewise.
23832 (OTHER_P8_VECTOR_MASKS): Likewise.
23833 (OTHER_VSX_VECTOR_MASKS): Likewise.
23834 (POWERPC_MASKS): Likewise.
23835 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
23836 duplicate list of options.
23837 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
23838 explicit -mupper-regs options.
23839 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
23840 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
23841 alias for -mupper-regs-df.
23842 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
23843 (rs6000_init_hard_regno_mode_ok): Likewise.
23844 (rs6000_option_override_internal): Likewise.
23845 (rs6000_opt_masks): Likewise.
23846 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
23847 options in terms of whether -mvsx or -mpower8-vector was used.
23848 (TARGET_UPPER_REGS_DI): Likewise.
23849 (TARGET_UPPER_REGS_SF): Likewise.
23850 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23851 -mupper-regs-* options.
23853 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23855 * passes.c (emergency_dump_function): Print some empty lines and a
23856 header before the RTL dump.
23858 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23860 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23862 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23865 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23866 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23868 2017-07-24 Carl Love <cel@us.ibm.com>
23870 * config/rs6000/rs6000-c.c: Add support for built-in functions
23871 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23872 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23873 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23874 vec_extract_fp_from_shortl): Add defines for the two builtins.
23875 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23876 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23878 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23879 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23880 * doc/extend.texi: Update the built-in documentation file for the
23881 new built-in function.
23883 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23886 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23888 * doc/generic.texi: Likewise.
23889 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23890 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23892 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23894 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23895 (aarch64_mls_elt_merge<mode>): Likewise.
23897 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23899 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23900 having __cxa_atexit.
23902 2017-07-23 Michael Collison <michael.collison@arm.com>
23904 * config/arm/arm.c (arm_option_override): Deprecate
23905 use of -mstructure-size-boundary.
23906 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23907 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23909 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23912 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23913 Reduce cost estimate for direct moves.
23915 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23918 * config/i386/i386.c (ix86_option_override_internal): Disable
23919 BMI, BMI2 and TBM instructions for -m16.
23921 2017-07-21 Carl Love <cel@us.ibm.com>
23923 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23924 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23925 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23926 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23927 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23928 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23929 VMULOSW): New enum "unspec" values.
23930 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23931 altivec_vmulosw): New patterns.
23932 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23933 VMULOSW): Add definitions.
23935 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23937 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23938 (qdf24xx): Likewise.
23939 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23940 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23941 (AARCH64_FL_V8_1): Renumber.
23942 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23943 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23944 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23945 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23946 rdma to feature modifiers list.
23948 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23950 PR middle-end/56727
23951 * ipa-visibility (function_and_variable_visibility): Convert
23952 recursive PLT call to direct call if appropriate.
23954 2017-07-21 Andrew Pinski <apinski@cavium.com>
23956 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23957 operand 1 to see if the types precision matches.
23958 * fold-const.c (operand_equal_p): Likewise.
23960 2017-07-21 Richard Biener <rguenther@suse.de>
23962 PR tree-optimization/81303
23963 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23964 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23965 (vect_peeling_hash_get_lowest_cost): Adjust.
23966 (vect_enhance_data_refs_alignment): Likewise. Use
23967 vect_get_peeling_costs_all_drs to compute the penalty for no
23968 peeling to match up costs.
23970 2017-07-21 Richard Biener <rguenther@suse.de>
23972 PR tree-optimization/81500
23973 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23974 we didn't identify a reduction path.
23976 2017-07-21 Tom de Vries <tom@codesourcery.com>
23977 Cesar Philippidis <cesar@codesourcery.com>
23979 PR gcov-profile/81442
23980 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23983 2017-07-21 Tom de Vries <tom@codesourcery.com>
23986 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23988 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23989 nvptx_override_options_after_change.
23991 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23993 * dwarf2out.c (output_file_names): Avoid double testing for
23994 dwarf_version >= 5.
23996 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23998 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
24000 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
24002 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
24004 (try_crossjump_to_edge): Do not punt on partitioned functions.
24006 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
24008 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
24009 Put all BBs reachable only via paths crossing cold region to cold
24011 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
24013 2016-07-21 Richard Biener <rguenther@suse.de>
24015 PR tree-optimization/81303
24016 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
24017 into account prologue and epilogue iterations when raising
24018 min_profitable_iters to sth at least covering one vector iteration.
24020 2017-07-21 Tamar Christina <tamar.christina@arm.com>
24022 * config/arm/arm.c (arm_test_cpu_arch_dat):
24025 2017-07-20 Nathan Sidwell <nathan@acm.org>
24027 Remove TYPE_METHODS.
24028 * tree.h (TYPE_METHODS): Delete.
24029 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
24030 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
24031 (dbxout_type_methods): Scan TYPE_FIELDS.
24032 (dbxout_type): Don't check TYPE_METHODS here.
24033 * function.c (use_register_for_decl): Always ignore register for
24034 class types when not optimizing.
24035 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
24036 * tree.c (free_lang_data_in_type): Stitch out member functions and
24037 templates from TYPE_FIELDS.
24038 (build_distinct_type_copy, verify_type_variant,
24039 verify_type): Member fns are on TYPE_FIELDS.
24040 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
24041 * tree-pretty-print.c (dump_generic_node): Likewise.
24043 2017-07-20 Jakub Jelinek <jakub@redhat.com>
24046 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
24047 V2TImode and V4TImode.
24048 (ix86_expand_vector_extract): Likewise.
24049 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
24050 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
24051 (ssescalarmode): Handle V4TImode and V2TImode.
24052 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
24053 (*vec_extractv2ti, *vec_extractv4ti): New insns.
24054 (VEXTRACTI128_MODE): New mode iterator.
24055 (splitter for *vec_extractv?ti first element): New.
24056 (VEC_INIT_MODE): New mode iterator.
24057 (vec_init<mode>): Consolidate 3 expanders into one using
24058 VEC_INIT_MODE mode iterator.
24060 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
24062 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
24063 non_spilled_static_chain_regno_p.
24065 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
24067 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
24069 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
24071 * bb-reorder.c (connect_traces): Allow copying of blocks within
24074 2017-07-20 Richard Biener <rguenther@suse.de>
24076 * gimple.h (gimple_phi_result): Add gphi * overload.
24077 (gimple_phi_result_ptr): Likewise.
24078 (gimple_phi_arg): Likewise. Adjust index assert to only
24079 allow actual argument accesses rather than all slots available
24081 (gimple_phi_arg_def): Add gphi * overload.
24082 * tree-phinodes.c (make_phi_node): Initialize only actual
24084 (resize_phi_node): Clear memory not covered by old node,
24085 do not initialize excess argument slots.
24086 (reserve_phi_args_for_new_edge): Initialize new argument slot
24089 2017-07-20 Bin Cheng <bin.cheng@arm.com>
24091 PR tree-optimization/81388
24093 2016-07-21 Bin Cheng <bin.cheng@arm.com>
24095 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
24096 by removing computation of may_be_zero.
24098 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24099 Tom de Vries <tom@codesourcery.com>
24101 PR middle-end/81030
24102 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
24103 when gimple level profile disagrees with what RTL expander did.
24105 2017-07-20 Richard Biener <rguenther@suse.de>
24107 PR tree-optimization/61171
24108 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
24109 (vect_analyze_stmt): Add slp instance parameter.
24110 (vectorizable_reduction): Likewise.
24111 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
24112 (vect_is_simple_reduction): Deal with chains not detected
24113 as SLP reduction chain, specifically not properly associated
24114 chains containing a mix of plus/minus.
24115 (get_reduction_op): Remove.
24116 (get_initial_defs_for_reduction): Simplify, pass in whether
24117 this is a reduction chain, pass in the SLP node for the PHIs.
24118 (vect_create_epilog_for_reduction): Get the SLP instance as
24120 (vectorizable_reduction): Get the SLP instance as arg.
24121 During analysis remember the SLP node with the PHIs in the
24122 instance. Simplify getting at the vectorized reduction PHIs.
24123 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
24124 through SLP instance.
24125 (vect_slp_analyze_operations): Likewise.
24126 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
24127 (vect_transform_stmt): Likewise.
24129 2017-07-20 Tom de Vries <tom@codesourcery.com>
24131 PR tree-optimization/81489
24132 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
24133 read of phi arg location to before loop that modifies phi.
24135 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24137 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
24140 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24142 PR middle-end/81331
24143 * except.c (execute): Fix ordering issue.
24145 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
24147 PR rtl-optimization/81423
24148 * combine.c (make_compound_operation_int): Don't try to optimize
24149 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
24151 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
24153 PR rtl-optimization/81423
24154 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
24155 with a constant that is -1 in the truncated to mode.
24157 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24159 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
24160 (determine_unlikely_bbs): ... here.
24161 * predict.h (propagate_unlikely_bbs_forward): Declare.
24162 * cfgexpand.c (pass_expand::execute): Use it.
24163 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
24165 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
24166 propagate_unlikely_bbs_forward.
24168 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24170 PR middle-end/81331
24171 * except.c (maybe_add_nop_after_section_switch): New function.
24174 2017-07-19 Tom de Vries <tom@codesourcery.com>
24176 * gimple.h (gimple_phi_set_arg): Make assert more strict.
24178 2017-07-19 Tom de Vries <tom@codesourcery.com>
24180 * gimple.h (gimple_phi_arg): Make assert more strict.
24182 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
24184 * config.gcc (powerpc*-*-*): Add mmintrin.h.
24185 * config/rs6000/mmintrin.h: New file.
24186 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
24188 2017-07-19 Jakub Jelinek <jakub@redhat.com>
24190 PR tree-optimization/81346
24191 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
24193 2017-07-19 Tom de Vries <tom@codesourcery.com>
24195 * config/nvptx/nvptx.md (VECIM): Add V2DI.
24197 2017-07-19 Tom de Vries <tom@codesourcery.com>
24199 * config/nvptx/nvptx-modes.def: Add V2DImode.
24200 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
24201 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
24202 (nvptx_output_mov_insn): Handle lack of mov.b128.
24203 (nvptx_print_operand): Handle 'H' and 'L' codes.
24204 (nvptx_vector_mode_supported): Allow V2DImode.
24205 (nvptx_preferred_simd_mode): New function.
24206 (nvptx_data_alignment): New function.
24207 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
24208 nvptx_preferred_simd_mode.
24209 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
24211 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
24213 2017-07-19 Tom de Vries <tom@codesourcery.com>
24215 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
24216 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
24217 (nvptx_vector_mode_supported): New function. Allow V2SImode.
24218 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
24219 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
24220 (mov<VECIM>_insn): New define_insn.
24221 (define_expand "mov<VECIM>): New define_expand.
24223 2017-07-19 Tom de Vries <tom@codesourcery.com>
24225 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
24227 2017-07-19 Jakub Jelinek <jakub@redhat.com>
24229 PR tree-optimization/81346
24230 * fold-const.h (fold_div_compare, range_check_type): Declare.
24231 * fold-const.c (range_check_type): New function.
24232 (build_range_check): Use range_check_type.
24233 (fold_div_compare): No longer static, rewritten into
24234 a match.pd helper function.
24235 (fold_comparison): Don't call fold_div_compare here.
24236 * match.pd (X / C1 op C2): New optimization using fold_div_compare
24237 as helper function.
24239 2017-07-19 Nathan Sidwell <nathan@acm.org>
24241 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
24242 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
24243 * tree.c (find_decls_types_r, verify_type): Use
24244 TYPE_{MIN,MAX}_VALUE_RAW.
24245 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
24246 (hash_tree): Likewise.
24247 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
24249 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
24252 2017-07-18 Tom de Vries <tom@codesourcery.com>
24254 PR middle-end/81464
24255 * omp-expand.c (expand_omp_for_static_chunk): Handle
24256 equal-argument loop exit phi.
24258 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
24261 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
24262 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
24263 operand 2 predicate.
24264 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
24265 operand 2 predicate.
24266 (ror,rol -> rorx splitters): Use const_int_operand as
24267 operand 2 predicate.
24269 2017-06-18 Richard Biener <rguenther@suse.de>
24271 PR tree-optimization/81410
24272 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
24273 the gap in the ! slp_perm SLP case after each group.
24275 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24277 PR middle-end/81463
24278 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
24281 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24283 PR middle-end/81462
24284 * predict.c (set_even_probabilities): Cleanup; do not affect
24285 probabilities that are already known.
24286 (combine_predictions_for_bb): Call even when count is set.
24288 2017-07-18 Nathan Sidwell <nathan@acm.org>
24290 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
24293 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24296 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
24297 optimization for loop niter analysis.
24299 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
24302 * config/avr/avr.c (avr_optimize_casesi): Don't use
24303 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
24305 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
24307 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
24308 body_cost_vec from _vect_peel_extended_info.
24309 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
24310 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
24313 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24315 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
24317 2017-07-18 Richard Biener <rguenther@suse.de>
24319 PR tree-optimization/80620
24320 PR tree-optimization/81403
24321 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
24322 info when re-using a VN table entry.
24324 2017-07-18 Richard Biener <rguenther@suse.de>
24326 PR tree-optimization/81418
24327 * tree-vect-loop.c (vectorizable_reduction): Properly compute
24328 vectype_in. Verify that with lane-reducing reduction operations
24329 we have a single def-use cycle.
24331 2017-07-17 Carl Love <cel@us.ibm.com>
24333 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
24335 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24336 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24337 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24338 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24339 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24340 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24341 VMULOSW): New enum "unspec" values.
24342 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24343 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24344 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24345 altivec_vmulosw): New patterns.
24346 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24347 VMULOSW): Add definitions.
24349 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
24351 * config/alpha/alpha.c: Include predict.h.
24353 2017-07-17 Yury Gribov <tetra2005@gmail.com>
24355 * tree-vrp.c (compare_assert_loc): Fix comparison function
24356 to return predictable results.
24358 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24360 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
24362 (subdi3): Likewise.
24363 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
24364 * doc/invoke.texi (mexpand-adddi): Update text.
24366 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24368 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
24369 that also clobbers the CC register. The old expand code is moved
24371 (*arc_clzsi2): ... here.
24372 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
24373 the CC register. The old expand code is moved to ...
24374 (arc_ctzsi2): ... here.
24376 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24378 * config/arc/arc.opt (mindexed-loads): Use initial value
24379 TARGET_INDEXED_LOADS_DEFAULT.
24380 (mauto-modify-reg): Use initial value
24381 TARGET_AUTO_MODIFY_REG_DEFAULT.
24382 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24383 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24384 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24385 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24387 2017-07-17 Martin Liska <mliska@suse.cz>
24390 * opts.c (finish_options): Do not allow -fgnu-tm
24391 w/ -fsanitize={kernel-,}address. Say sorry.
24393 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24396 * tree-loop-distribution.c (classify_partition): Only assert on
24397 numer of iterations.
24398 (merge_dep_scc_partitions): Delete prameter. Update function call.
24399 (distribute_loop): Remove code handling loop with unknown niters.
24400 (pass_loop_distribution::execute): Skip loop with unknown niters.
24402 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24405 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
24406 function sort_partitions_by_post_order.
24408 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24410 PR tree-optimization/81374
24411 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
24412 the max index of basic blocks, rather than number of basic blocks.
24414 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24416 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
24418 (arc_legitimate_pic_operand_p): Likewise.
24419 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
24421 (arc_needs_pcl_p): Likewise.
24422 (arc_legitimate_pc_offset_p): Likewise.
24423 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
24424 function is also used in constrains.md.
24425 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
24426 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
24427 PLUS. Only return true/false in known cases, otherwise assert.
24428 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
24429 is already called in arc_legitimate_constant_p.
24430 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
24432 (LEGITIMATE_PIC_OPERAND_P): Use
24433 arc_raw_symbolic_reference_mentioned_p function.
24434 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
24439 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24440 Andrew Burgess <andrew.burgess@embecosm.com>
24442 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
24443 (arc_return_address_register): New function.
24444 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
24445 (arc_handle_fndecl_attribute): Add naked attribute.
24446 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
24447 (TARGET_WARN_FUNC_RETURN): Likewise.
24448 (arc_allocate_stack_slots_for_args): New function.
24449 (arc_warn_func_return): Likewise.
24450 (machine_function): Change type fn_type.
24451 (arc_compute_function_type): Consider new naked function type,
24452 change function return type.
24453 (arc_must_save_register): Adapt to handle new
24454 arc_compute_function_type's return type.
24455 (arc_expand_prologue): Likewise.
24456 (arc_expand_epilogue): Likewise.
24457 (arc_return_address_regs): Delete.
24458 (arc_return_address_register): New function.
24459 (arc_epilogue_uses): Use above function.
24460 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
24461 (arc_function_type): Change encoding, add naked type.
24462 (ARC_INTERRUPT_P): Change to handle the new encoding.
24463 (ARC_FAST_INTERRUPT_P): Likewise.
24464 (ARC_NORMAL_P): Define.
24465 (ARC_NAKED_P): Likewise.
24466 (arc_compute_function_type): Delete prototype.
24467 * config/arc/arc.md (in_ret_delay_slot): Use
24468 arc_return_address_register function.
24469 (simple_return): Likewise.
24470 (p_return_i): Likewise.
24472 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24474 PR tree-optimization/81428
24475 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
24476 can't be built for those types.
24478 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24480 Remove stuff dead since r239246.
24482 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
24483 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
24484 (avr_inform_devices): Remove dead stuff.
24486 2017-07-17 Tamar Christina <tamar.christina@arm.com>
24488 * config/arm/arm_neon.h: Fix softp typo.
24490 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24492 PR tree-optimization/81365
24493 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
24494 aggregate moves onto bb predecessor edges, make sure there are no
24495 loads that could alias the lhs in between the start of bb and the
24498 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24501 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
24502 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
24503 [LSHIFTRT, outer_code = TRUNCATE]: Same.
24505 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24507 PR tree-optimization/81396
24508 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
24509 (init_symbolic_number): Initialize it to 1.
24510 (perform_symbolic_merge): Add n_ops from both operands into the new
24512 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
24513 without base_addr as useless if they need more than one operation.
24514 (bswap_replace): Handle !bswap case for NULL base_addr.
24516 2017-07-17 Tom de Vries <tom@codesourcery.com>
24519 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
24522 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
24524 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
24525 conditional builtin define __FIX_LEON3FT_B2BST.
24527 2017-07-17 Daniel Cederman <cederman@gaisler.com>
24529 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
24530 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
24533 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24535 PR rtl-optimization/81424
24536 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
24537 to remove potential trapping from operands if -fnon-call-exceptions.
24539 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24541 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
24542 profile_proability for scalling.
24543 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
24545 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24547 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
24549 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24551 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
24552 fixpoint arithmetics.
24554 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24556 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
24557 fixpoint arithmetics.
24559 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24561 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
24562 fixpoint arithmetics.
24564 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24566 * profile-count.h (profile_probability::from_reg_br_prob_note,
24567 profile_probability::to_reg_br_prob_note): New functions.
24568 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
24569 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
24570 * predict.c (probability_reliable_p): Update.
24571 (edge_probability_reliable_p): Update.
24572 (br_prob_note_reliable_p): Update.
24573 (invert_br_probabilities): Update.
24574 (add_reg_br_prob_note): New function.
24575 (combine_predictions_for_insn): Update.
24576 * asan.c (asan_clear_shadow): Update.
24577 * cfgbuild.c (compute_outgoing_frequencies): Update.
24578 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24579 (update_br_prob_note): Update.
24580 (rtl_verify_edges): Update.
24581 (purge_dead_edges): Update.
24582 (fixup_reorder_chain): Update.
24583 * emit-rtl.c (try_split): Update.
24584 * ifcvt.c (cond_exec_process_insns): Update.
24585 (cond_exec_process_if_block): Update.
24586 (dead_or_predicable): Update.
24587 * internal-fn.c (expand_addsub_overflow): Update.
24588 (expand_neg_overflow): Update.
24589 (expand_mul_overflow): Update.
24590 * loop-doloop.c (doloop_modify): Update.
24591 * loop-unroll.c (compare_and_jump_seq): Update.
24592 * optabs.c (emit_cmp_and_jump_insn_1): Update.
24593 * predict.h: Update.
24594 * reorg.c (mostly_true_jump): Update.
24596 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
24597 * config/alpha/alpha.c (emit_unlikely_jump): Update.
24598 * config/arc/arc.c: (emit_unlikely_jump): Update.
24599 * config/arm/arm.c: (emit_unlikely_jump): Update.
24600 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
24601 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
24602 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
24603 (ix86_print_operand): Update.
24604 (ix86_split_fp_branch): Update.
24605 (predict_jump): Update.
24606 * config/ia64/ia64.c (ia64_print_operand): Update.
24607 * config/mmix/mmix.c (mmix_print_operand): Update.
24608 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
24609 (rs6000_expand_split_stack_prologue): Update.
24610 * config/rs6000/rs6000.c: Update.
24611 * config/s390/s390.c (s390_expand_vec_strlen): Update.
24612 (s390_expand_vec_movstr): Update.
24613 (s390_expand_cs_tdsi): Update.
24614 (s390_expand_split_stack_prologue): Update.
24615 * config/sh/sh.c (sh_print_operand): Update.
24616 (expand_cbranchsi4): Update.
24617 (expand_cbranchdi4): Update.
24618 * config/sparc/sparc.c (output_v9branch): Update.
24619 * config/spu/spu.c (get_branch_target): Update.
24620 (ea_load_store_inline): Update.
24621 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
24622 * config/tilepro/tilepro.c: Update.
24624 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24626 * gimplify.c (mostly_copy_tree_r): Revert latest change.
24627 (gimplify_save_expr): Likewise.
24629 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24631 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
24633 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24635 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
24637 * timevar.def (TV_IPA_FNSUMMARY): Define.
24639 2017-07-16 Daniel Cederman <cederman@gaisler.com>
24641 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
24642 to back store errata sensitive sequence from being generated.
24643 (sqrtdf2_fix): Likewise.
24645 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24647 * tree-ssa-threadupdate.c (compute_path_counts,
24648 update_joiner_offpath_counts): Use profile_probability.
24650 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
24653 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24655 * config/arm/arm-c.c (arm_cpu_builtins): Define
24656 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24658 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
24660 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24661 array entries to represent __ieee128 versions of the
24662 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
24663 scalar_extract_sig, and scalar_insert_exp built-in functions.
24664 (altivec_resolve_overloaded_builtin): Add special case handling
24665 for the __builtin_scalar_insert_exp function, as represented by
24666 the P9V_BUILTIN_VEC_VSIEDP constant.
24667 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
24668 exponent support for __ieee128 argument.
24669 (VSESQP): Add scalar extract signature support for __ieee128
24671 (VSTDCNQP): Add scalar test negative support for __ieee128
24673 (VSIEQP): Add scalar insert exponent support for __int128 argument
24674 with __ieee128 result.
24675 (VSIEQPF): Add scalar insert exponent support for __ieee128
24676 argument with __ieee128 result.
24677 (VSTDCQP): Add scalar test data class support for __ieee128
24679 (VSTDCNQP): Add overload support for scalar test negative with
24680 __ieee128 argument.
24681 (VSTDCQP): Add overload support for scalar test data class
24682 __ieee128 argument.
24683 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
24684 UNSPEC_VSX_SXSIGDP.
24685 (UNSPEC_VSX_SIEXPQP): New constant.
24686 (xsxexpqp): New insn for VSX scalar extract exponent quad
24688 (xsxsigqp): New insn for VSX scalar extract significand quad
24690 (xsiexpqpf): New insn for VSX scalar insert exponent quad
24691 precision with floating point argument.
24692 (xststdcqp): New expand for VSX scalar test data class quad
24694 (xststdcnegqp): New expand for VSX scalar test negative quad
24696 (xststdcqp): New insn to match expansions for VSX scalar test data
24697 class quad precision and VSX scalar test negative quad precision.
24698 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
24699 special case operand checking to enforce that second operand of
24700 VSX scalar test data class with quad precision argument is a 7-bit
24702 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24703 prototypes and descriptions of __ieee128 versions of
24704 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
24705 scalar_test_data_class, and scalar_test_neg built-in functions.
24707 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24709 PR tree-optimization/81162
24710 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
24711 replace a negate with an add.
24713 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
24715 * doc/invoke.texi (arm/-mcpu): Document +crypto.
24717 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24719 * config/arm/arm-c.c (arm_cpu_builtins): Define
24720 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24722 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24724 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
24725 (armv8-r): Set ARM Cortex-R52 as default CPU.
24726 * config/arm/arm-tables.opt: Regenerate.
24727 * config/arm/arm-tune.md: Regenerate.
24728 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24730 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
24731 extension for -mcpu=cortex-r52.
24733 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24735 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
24736 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
24737 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
24738 (fp-armv8): Define it as FP_ARMv8 only.
24739 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
24740 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
24742 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
24743 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
24744 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
24745 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
24746 than TARGET_FPU_ARMV8.
24747 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
24748 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
24749 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
24751 * config/arm/neon.md (neon_vrint): Likewise.
24752 (neon_vcvt): Likewise.
24753 (neon_<fmaxmin_op><mode>): Likewise.
24754 (<fmaxmin><mode>3): Likewise.
24755 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
24756 * config/arm/predicates.md (arm_cond_move_operator): Check against
24757 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
24759 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
24761 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
24762 to top of function.
24764 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24766 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
24767 loop in comment with memset.
24769 2017-07-14 Martin Liska <mliska@suse.cz>
24771 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
24772 * dwarf2out.c (is_java): Remove the function.
24773 (output_pubname): Remove usage of the function.
24774 (lower_bound_default): Remove usage of DW_LANG_Java.
24775 (gen_compile_unit_die): Likewise.
24776 * gcc.c: Remove compiler defaults for .java and .zip files.
24777 * gimple-expr.c (remove_suffix): Change as there's no longer
24778 extension than 4-letter one.
24779 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
24780 (gimplify_save_expr): Likewise.
24781 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
24782 as it's possible even for other languages than Java.
24783 * langhooks.h (struct lang_hooks): Remove Java from a comment.
24784 * lto-opts.c (lto_write_options): Remove reference to Java.
24785 * opts.c (strip_off_ending): Update file extension handling.
24786 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
24787 * tree-eh.c (lower_resx): Likewise.
24788 * tree.c (free_lang_data_in_type): Remove dead code.
24789 (find_decls_types_r): Likewise.
24790 (build_common_builtin_nodes): Remove Java from a comment.
24791 (verify_type): Remove dead code.
24792 * varasm.c (assemble_external): Remove Java from a comment.
24794 2017-07-14 Martin Liska <mliska@suse.cz>
24796 * opts.c (finish_options): Add quotes.
24797 (common_handle_option): Likewise.
24799 2017-07-14 Martin Liska <mliska@suse.cz>
24801 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
24802 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
24803 Remove N_SO_PASCAL.
24804 * dwarf2out.c (lower_bound_default): Do not handle
24806 (gen_compile_unit_die): Likewise.
24807 * gcc.c: Remove default extension binding for GNU Pascal.
24808 * stmt.c: Remove Pascal language from a comment.
24809 * xcoffout.c: Likewise.
24811 2017-07-13 David Malcolm <dmalcolm@redhat.com>
24814 * diagnostic-show-locus.c (fixit_cmp): New function.
24815 (layout::layout): Sort m_fixit_hints.
24816 (column_range::column_range): Assert that the values are valid.
24817 (struct char_span): New struct.
24818 (correction::overwrite): New method.
24819 (struct source_line): New struct.
24820 (line_corrections::add_hint): Add assertions. Reimplement memcpy
24821 calls in terms of classes source_line and char_span, and
24822 correction::overwrite.
24823 (selftest::test_overlapped_fixit_printing_2): New function.
24824 (selftest::diagnostic_show_locus_c_tests): Call it.
24826 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
24828 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
24829 early if there is no lhs.
24831 2017-07-13 Martin Liska <mliska@suse.cz>
24833 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
24834 (gen_reference_type_die): Likewise.
24835 * stor-layout.c: Remove Pascal-related comment.
24837 2017-07-13 Martin Liska <mliska@suse.cz>
24839 * opts.c (finish_options): Add quotes to error messages.
24840 (parse_sanitizer_options): Likewise.
24842 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
24844 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
24846 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
24848 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24850 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24852 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24854 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24856 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24859 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24860 provides the hardware capability bits, define the macro
24861 __BUILTIN_CPU_SUPPORTS__.
24862 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24863 if GLIBC does not provide the hardware capability bits. Add a
24864 gcc_unreachable call if the built-in cpu function is neither
24865 __builtin_cpu_is nor __builtin_cpu_supports.
24866 (rs6000_get_function_versions_dispatcher): Change the warning
24867 that an old GLIBC is used which does not export the capability
24868 bits to be an error.
24869 * doc/extend.texi (target_clones attribute): Document the
24870 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24871 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24872 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24873 the macros defined by GCC if the newer GLIBC is available.
24875 2017-07-12 Jeff Law <law@redhat.com>
24877 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24878 remaining includes slightly.
24879 * config/riscv/riscv-builtins.c: Include profile-count.h.
24881 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24884 * config/avr/avr.c (avr_set_current_function): In diagnostic
24885 messages: Quote keywords and (parts of) identifiers.
24886 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24889 2017-07-12 Carl Love <cel@us.ibm.com>
24891 * config/rs6000/rs6000-c.c: Add support for built-in functions
24892 vector bool char vec_revb (vector bool char);
24893 vector bool short vec_revb (vector short char);
24894 vector bool int vec_revb (vector bool int);
24895 vector bool long long vec_revb (vector bool long long);
24896 * doc/extend.texi: Update the built-in documentation file for the
24897 new built-in functions.
24899 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24901 * config/s390/s390.md: Remove movcc splitter.
24903 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24905 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24906 load/store on condition.
24908 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24911 * config/avr/avr.c (avr_encode_section_info)
24912 [progmem && !TREE_READONLY]: Error if progmem object needs
24915 2017-07-11 Michael Collison <michael.collison@arm.com>
24917 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24920 2017-07-11 Carl Love <cel@us.ibm.com>
24922 * config/rs6000/rs6000-c.c: Add support for builtins
24923 vector unsigned int vec_parity_lsbb (vector signed int);
24924 vector unsigned int vec_parity_lsbb (vector unsigned int);
24925 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24926 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24927 vector unsigned long long vec_parity_lsbb (vector signed long long);
24928 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24929 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24930 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24931 * doc/extend.texi: Update the built-in documentation file for the
24932 new built-in functions.
24934 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24936 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24937 (layout::m_primary_loc): New field.
24938 (layout::layout): Initialize new field. Move location filtering
24939 logic from here to...
24940 (layout::maybe_add_location_range): ...this new method. Add
24941 support for filtering to just the lines already specified by other
24943 (layout::will_show_line_p): New method.
24944 (gcc_rich_location::add_location_if_nearby): New method.
24945 (selftest::test_add_location_if_nearby): New test function.
24946 (selftest::diagnostic_show_locus_c_tests): Call it.
24947 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24950 2017-07-11 Tom de Vries <tom@codesourcery.com>
24952 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24953 (bb_first_real_insn): New function.
24954 (nvptx_single): Add extra initialization of broadcasted condition
24957 2017-07-11 Nathan Sidwell <nathan@acm.org>
24959 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24961 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24963 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24964 Binutils doc as TEXI will mess them up.
24965 * doc/invoke.texi (AVR Options): Same here.
24967 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24969 * config/sparc/sparc.opt (mfix-ut700): New option.
24970 (mfix-gr712rc): Likewise.
24971 (sparc_fix_b2bst): New variable.
24972 * doc/invoke.texi (SPARC options): Document them.
24973 (ARM options): Fix warnings.
24974 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24975 instructions to prevent sequences that can trigger the store-store
24976 errata for certain LEON3FT processors.
24977 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24978 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24979 * config/sparc/sparc.md (fix_b2bst): New attribute.
24980 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24982 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24985 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24987 (*rsqrtsf2_sse): Ditto.
24989 (div<mode>3): Macroize insn from divdf3 and divsf3
24990 using MODEF mode iterator.
24992 2017-07-10 Martin Sebor <msebor@redhat.com>
24994 PR tree-optimization/80397
24995 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24996 instead of testing for equality to INTEGER_TYPE.
24998 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
25000 * config.gcc: Remove uclibc from arc target spec.
25002 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
25004 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
25006 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
25009 * lto-wrapper.c (remove_option): New function.
25010 (merge_and_complain): Merge PIC/PIE options more realistically.
25012 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
25014 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
25018 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
25019 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
25020 * config.in: Regenerate.
25021 * configure: Regenerate.
25022 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
25023 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
25024 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
25025 (TARGET_GASISR_PROLOGUES): ...target mask.
25026 * common/config/avr/avr-common.c
25027 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
25028 Set -mgas-isr-prologues.
25029 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
25030 INSERT_PASS_BEFORE for it.
25031 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
25032 * config/avr/avr.c (avr_option_override)
25033 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
25034 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
25035 (avr_attribute_table) <no_gccisr>: Add new function attribute.
25036 (avr_set_current_function) <is_no_gccisr>: Init machine field.
25037 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
25039 (make_avr_pass_pre_proep): New function.
25040 (emit_push_sfr) <treg>: Add argument to function and use it
25041 instead of TMP_REG.
25042 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
25043 and set machine->gasisr.yes.
25044 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
25045 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
25046 __gcc_isr.n_pushed to .L__stack_usage.
25047 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
25048 (avr_asm_final_postscan_insn): ...this new static function.
25049 * config/avr/avr.h (machine_function)
25050 <is_no_gccisr, use_L__stack_usage>: New fields.
25051 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
25052 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
25053 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
25054 (gasisr, *gasisr): New expander and insn.
25055 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
25056 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
25057 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
25059 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
25061 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
25064 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
25066 Move jump-tables out of .text again.
25069 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
25070 (ASM_OUTPUT_ADDR_VEC): New function.
25071 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
25072 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
25073 INSN_ADDRESSes as asm comment.
25074 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
25075 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
25076 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
25077 * config/avr/avr.md (*tablejump): Adjust comment.
25078 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
25079 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
25081 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
25082 (avr_output_addr_vec): New proto.
25083 (avr_log_t) <insn_addresses>: New field.
25085 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
25088 * config/i386/i386.c (ix86_function_arg_advance): Set
25089 outgoing_args_on_stack to true if there are outgoing arguments
25091 (ix86_function_arg): Likewise.
25092 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
25093 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
25094 * config/i386/i386.h (machine_function): Add
25095 outgoing_args_on_stack.
25097 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
25099 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
25100 supporting pthreds.
25101 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
25103 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
25105 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
25106 (REAL_H): Remove $(MACHMODE_H).
25107 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
25109 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
25110 $(MACHMODE_H) and double-int.h.
25111 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
25113 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
25116 2017-07-07 Andrew Pinski <apinski@cavium.com>
25118 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
25119 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
25121 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
25123 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
25124 Add warning if GCC was not configured to link against a GLIBC that
25125 exports the hardware capability bits.
25126 (make_resolver_func): Make resolver function private and not a
25127 COMDAT function. Create the name with clone_function_name instead
25128 of make_unique_name.
25131 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
25132 correct operand in doing the split.
25134 2017-07-07 Carl Love <cel@us.ibm.com>
25136 * config/rs6000/rs6000-c: Add support for built-in function
25137 vector unsigned short vec_pack_to_short_fp32 (vector float,
25139 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
25140 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
25141 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
25142 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
25143 (convert_4f32_8i16): Add define_expand.
25144 * doc/extend.texi: Update the built-in documentation file for the
25145 new built-in function.
25147 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25149 * config/sparc/m8.md: New file.
25150 * config/sparc/sparc.md: Include m8.md.
25152 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25154 * config/sparc/sparc.opt: New option -mvis4b.
25155 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
25156 (sparc_option_override): Handle VIS4B.
25157 (enum sparc_builtins): Define
25158 SPARC_BUILTIN_DICTUNPACK{8,16,32},
25159 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
25160 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
25161 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
25162 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
25163 (check_constant_argument): New function.
25164 (sparc_vis_init_builtins): Define builtins
25165 __builtin_vis_dictunpack{8,16,32},
25166 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
25167 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
25168 __builtin_vis_fpcmpde{8,16,32}shl and
25169 __builtin_vis_fpcmpur{8,16,32}shl.
25170 (sparc_expand_builtin): Check that the constant operands to
25171 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
25172 constant and in range.
25173 * config/sparc/sparc-c.c (sparc_target_macros): Handle
25175 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
25176 (SPARC_IMM5_P): Likewise.
25177 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
25178 (enabled): Handle vis4b.
25179 (UNSPEC_DICTUNPACK): New unspec.
25180 (UNSPEC_FPCMPSHL): Likewise.
25181 (UNSPEC_FPUCMPSHL): Likewise.
25182 (UNSPEC_FPCMPDESHL): Likewise.
25183 (UNSPEC_FPCMPURSHL): Likewise.
25184 (cpu_feature): New CPU feature `vis4b'.
25185 (dictunpack{8,16,32}): New insns.
25186 (FPCSMODE): New mode iterator.
25187 (fpcscond): New code iterator.
25188 (fpcsucond): Likewise.
25189 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
25190 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
25191 (fpcmpde{8,16,32}{si,di}shl): Likewise.
25192 (fpcmpur{8,16,32}{si,di}shl): Likewise.
25193 * config/sparc/constraints.md: Define constraints `q' for unsigned
25194 2-bit integer constants and `t' for unsigned 5-bit integer
25196 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
25198 (imm5_operand_dictunpack16): Likewise.
25199 (imm5_operand_dictunpack32): Likewise.
25200 (imm2_operand): Likewise.
25201 * doc/invoke.texi (SPARC Options): Document -mvis4b.
25202 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
25203 ditunpack* and fpcmp*shl builtins.
25205 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25207 * config.gcc: Handle m8 in --with-{cpu,tune} options.
25208 * config.in: Add HAVE_AS_SPARC6 define.
25209 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
25211 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
25213 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
25214 (CPP_CPU_SPEC): Handle m8.
25215 (ASM_CPU_SPEC): Likewise.
25216 * config/sparc/sparc-opts.h (enum processor_type): Add
25218 * config/sparc/sparc.c (m8_costs): New struct.
25219 (sparc_option_override): Handle TARGET_CPU_m8.
25220 (sparc32_initialize_trampoline): Likewise.
25221 (sparc64_initialize_trampoline): Likewise.
25222 (sparc_issue_rate): Likewise.
25223 (sparc_register_move_cost): Likewise.
25224 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
25225 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
25226 (ASM_CPU64_DEFAULT_SPEC): Likewise.
25227 (CPP_CPU_SPEC): Handle M8.
25228 (ASM_CPU_SPEC): Likewise.
25229 (AS_M8_FLAG): Define.
25230 * config/sparc/sparc.md: Add m8 to the cpu attribute.
25231 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
25232 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
25234 * configure: Regenerate.
25235 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
25238 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25240 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
25242 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
25243 ("*movdi_insn_sp32"): Do not set v3pipe.
25244 ("*movsi_insn"): Likewise.
25245 ("*movdi_insn_sp64"): Likewise.
25246 ("*movsf_insn"): Likewise.
25247 ("*movdf_insn_sp32"): Likewise.
25248 ("*movdf_insn_sp64"): Likewise.
25249 ("*zero_extendsidi2_insn_sp64"): Likewise.
25250 ("*sign_extendsidi2_insn"): Likewise.
25251 ("*mov<VM32:mode>_insn"): Likewise.
25252 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25253 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25254 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25255 ("<vlop:code><VL:mode>3"): Likewise.
25256 ("*not_<vlop:code><VL:mode>3"): Likewise.
25257 ("*nand<VL:mode>_vis"): Likewise.
25258 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
25259 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
25260 ("one_cmpl<VL:mode>2"): Likewise.
25261 ("faligndata<VM64:mode>_vis"): Likewise.
25262 ("alignaddrsi_vis"): Likewise.
25263 ("alignaddrdi_vis"): Likweise.
25264 ("alignaddrlsi_vis"): Likewise.
25265 ("alignaddrldi_vis"): Likewise.
25266 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25267 ("bmaskdi_vis"): Likewise.
25268 ("bmasksi_vis"): Likewise.
25269 ("bshuffle<VM64:mode>_vis"): Likewise.
25270 ("cmask8<P:mode>_vis"): Likewise.
25271 ("cmask16<P:mode>_vis"): Likewise.
25272 ("cmask32<P:mode>_vis"): Likewise.
25273 ("pdistn<P:mode>_vis"): Likewise.
25274 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25276 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25278 * config/sparc/sparc.md ("subtype"): New insn attribute.
25279 ("*wrgsr_sp64"): Set insn subtype.
25280 ("*rdgsr_sp64"): Likewise.
25281 ("alignaddrsi_vis"): Likewise.
25282 ("alignaddrdi_vis"): Likewise.
25283 ("alignaddrlsi_vis"): Likewise.
25284 ("alignaddrldi_vis"): Likewise.
25285 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25286 ("fexpand_vis"): Likewise.
25287 ("fpmerge_vis"): Likewise.
25288 ("faligndata<VM64:mode>_vis"): Likewise.
25289 ("bshuffle<VM64:mode>_vis"): Likewise.
25290 ("cmask8<P:mode>_vis"): Likewise.
25291 ("cmask16<P:mode>_vis"): Likewise.
25292 ("cmask32<P:mode>_vis"): Likewise.
25293 ("fchksm16_vis"): Likewise.
25294 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
25295 ("fmean16_vis"): Likewise.
25296 ("fp<plusminus_insn>64_vis"): Likewise.
25297 ("<plusminus_insn>v8qi3"): Likewise.
25298 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25299 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
25300 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
25301 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
25302 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
25303 ("*movqi_insn"): Likewise.
25304 ("*movhi_insn"): Likewise.
25305 ("*movsi_insn"): Likewise.
25306 ("movsi_pic_gotdata_op"): Likewise.
25307 ("*movdi_insn_sp32"): Likewise.
25308 ("*movdi_insn_sp64"): Likewise.
25309 ("movdi_pic_gotdata_op"): Likewise.
25310 ("*movsf_insn"): Likewise.
25311 ("*movdf_insn_sp32"): Likewise.
25312 ("*movdf_insn_sp64"): Likewise.
25313 ("*zero_extendhisi2_insn"): Likewise.
25314 ("*zero_extendqihi2_insn"): Likewise.
25315 ("*zero_extendqisi2_insn"): Likewise.
25316 ("*zero_extendqidi2_insn"): Likewise.
25317 ("*zero_extendhidi2_insn"): Likewise.
25318 ("*zero_extendsidi2_insn_sp64"): Likewise.
25319 ("ldfsr"): Likewise.
25320 ("prefetch_64"): Likewise.
25321 ("prefetch_32"): Likewise.
25322 ("tie_ld32"): Likewise.
25323 ("tie_ld64"): Likewise.
25324 ("*tldo_ldub_sp32"): Likewise.
25325 ("*tldo_ldub1_sp32"): Likewise.
25326 ("*tldo_ldub2_sp32"): Likewise.
25327 ("*tldo_ldub_sp64"): Likewise.
25328 ("*tldo_ldub1_sp64"): Likewise.
25329 ("*tldo_ldub2_sp64"): Likewise.
25330 ("*tldo_ldub3_sp64"): Likewise.
25331 ("*tldo_lduh_sp32"): Likewise.
25332 ("*tldo_lduh1_sp32"): Likewise.
25333 ("*tldo_lduh_sp64"): Likewise.
25334 ("*tldo_lduh1_sp64"): Likewise.
25335 ("*tldo_lduh2_sp64"): Likewise.
25336 ("*tldo_lduw_sp32"): Likewise.
25337 ("*tldo_lduw_sp64"): Likewise.
25338 ("*tldo_lduw1_sp64"): Likewise.
25339 ("*tldo_ldx_sp64"): Likewise.
25340 ("*mov<VM32:mode>_insn"): Likewise.
25341 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25342 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25344 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25346 * config/sparc/sparc.md ("type"): New insn type viscmp.
25347 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
25349 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
25350 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
25351 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25352 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
25354 ("n7_vis_logical_11cycle"): Likewise.
25355 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
25356 * config/sparc/niagara2.md ("niag3_vis": Likewise.
25357 * config/sparc/niagara.md ("niag_vis"): Likewise.
25358 * config/sparc/ultra3.md ("us3_fga"): Likewise.
25359 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
25361 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25363 * config/sparc/sparc.md: New instruction type `bmask'.
25364 (bmaskdi_vis): Use the `bmask' type.
25365 (bmasksi_vis): Likewise.
25366 * config/sparc/ultra3.md (us3_array): Likewise.
25367 * config/sparc/niagara7.md (n7_array): Likewise.
25368 * config/sparc/niagara4.md (n4_array): Likewise.
25369 * config/sparc/niagara2.md (niag2_vis): Likewise.
25370 (niag3_vis): Likewise.
25371 * config/sparc/niagara.md (niag_vis): Likewise.
25373 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25375 * ipa-comdats.c: Remove optimize check from gate.
25376 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
25377 for functions not optimized.
25378 (ipa_fn_summary_read): Skip optimize check.
25379 (ipa_fn_summary_write): Likewise.
25380 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
25382 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
25384 (can_inline_edge_p): Check flag_pcc_struct_return for match.
25385 (check_callers): Give up on caller which is not optimized.
25386 (inline_small_functions): Likewise.
25387 (ipa_inline): Do not give up when not optimizing.
25388 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
25389 away unoptimizes cdtors.
25390 (whole_program_function_and_variable_visibility): Do
25391 ipa_discover_readonly_nonaddressable_vars in LTO mode.
25392 * ipa.c (process_references): Do not check optimize.
25393 (symbol_table::remove_unreachable_nodes): Update optimize check.
25394 (set_writeonly_bit): Update optimize check.
25395 (pass_ipa_cdtor_merge::gate): Do not check optimize.
25396 (pass_ipa_single_use::gate): Remove.
25398 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25400 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
25401 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
25402 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
25403 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
25404 permute_load, permute_store, adjust_extract, adjust_splat,
25405 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
25406 replace_swap_with_copy, dump_swap_insn_table,
25407 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
25408 recombine_lvx_pattern, recombine_stvx_pattern,
25409 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
25410 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
25411 to file rs6000-p8swap.c.
25412 * config/rs6000/rs6000-p8swap.c: New file.
25413 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
25414 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
25415 and rs6000*-*-* targets.
25417 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25419 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
25421 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25423 * lto-wrapper.c (merge_and_complain): Do not merge
25424 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
25425 fsigned_zeros, ftrapping_math, fwrapv.
25426 (append_compiler_options): Do not track these options.
25427 (append_linker_options): Likewie
25429 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25431 * cgraphunit.c (cgraph_node::finalize_function): When
25432 !flag_toplevel_reorde set no_reorder flag.
25433 (varpool_node::finalize_decl): Likewise.
25434 (symbol_table::compile): Drop no toplevel reorder path.
25436 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25438 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
25439 edges; zero probability is not better than uninitialized.
25441 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25443 * asan.h (asan_sanitize_allocas_p): Declare.
25444 * asan.c (asan_sanitize_allocas_p): New function.
25445 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
25446 (handle_builtin_alloca): Likewise.
25447 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
25448 if !asan_sanitize_allocas_p.
25449 * params.def (asan-instrument-allocas): Add new option.
25450 * params.h (ASAN_PROTECT_ALLOCAS): Define.
25451 * opts.c (common_handle_option): Disable allocas sanitization for
25454 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25456 * asan.c: Include gimple-fold.h.
25457 (get_last_alloca_addr): New function.
25458 (handle_builtin_stackrestore): Likewise.
25459 (handle_builtin_alloca): Likewise.
25460 (asan_emit_allocas_unpoison): Likewise.
25461 (get_mem_refs_of_builtin_call): Add new parameter, remove const
25462 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
25463 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
25464 (instrument_builtin_call): Pass gimple iterator to
25465 get_mem_refs_of_builtin_call.
25466 (last_alloca_addr): New global.
25467 * asan.h (asan_emit_allocas_unpoison): Declare.
25468 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
25469 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
25470 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
25471 if function calls alloca.
25472 * gimple-fold.c (replace_call_with_value): Remove static keyword.
25473 * gimple-fold.h (replace_call_with_value): Declare.
25474 * internal-fn.c: Include asan.h.
25475 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
25476 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
25478 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25480 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
25481 (C_SELFTEST_FLAGS): New.
25482 (CPP_SELFTEST_FLAGS): New.
25483 (SELFTEST_DEPS): New, from deps of s-selftest.
25484 (C_SELFTEST_DEPS): New, from deps of s-selftest.
25485 (CPP_SELFTEST_DEPS): New.
25486 (selftest): Add dependency on s-selftest-c++.
25487 (s-selftest): Rename to...
25488 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
25489 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
25490 than SELFTEST_FLAGS.
25491 (selftest-gdb): Rename to...
25492 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
25494 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
25495 (selftest-valgrind): Rename to...
25496 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
25498 (selftest-valgrind): Reintroduce as an alias for
25499 selftest-c-valgrind.
25500 (s-selftest-c++): New.
25501 (selftest-c++-gdb): New.
25502 (selftest-c++-valgrind): New.
25504 2017-07-06 Olivier Hainque <hainque@adacore.com>
25506 * gcc.c (process_command): When deciding if undefined variables
25507 should be ignored when processing specs, accept "gcc -v" as well.
25509 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25511 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
25512 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
25514 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25516 * config/arm/arm-cpus.in (armv8-r): Add new entry.
25517 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
25518 * config/arm/arm-tables.opt: Regenerate.
25519 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
25521 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
25523 2017-07-06 Carl Love <cel@us.ibm.com>
25525 * ChangeLog: Clean up from mid air collision
25527 2017-07-06 Carl Love <cel@us.ibm.com>
25529 * config/rs6000/rs6000-c.c: Add support for built-in functions
25530 vector signed int vec_subc (vector signed int, vector signed int);
25531 vector signed __int128 vec_subc (vector signed __int128,
25532 vector signed __int128);
25533 vector unsigned __int128 vec_subc (vector unsigned __int128,
25534 vector unsigned __int128);
25535 vector signed int vec_sube (vector signed int, vector signed int,
25536 vector signed int);
25537 vector unsigned int vec_sube (vector unsigned int,
25538 vector unsigned int,
25539 vector unsigned int);
25540 vector signed __int128 vec_sube (vector signed __int128,
25541 vector signed __int128,
25542 vector signed__int128);
25543 vector unsigned __int128 vec_sube (vector unsigned __int128,
25544 vector unsigned __int128,
25545 vector unsigned __int128);
25546 vector signed int vec_subec (vector signed int, vector signed int,
25547 vector signed int);
25548 vector unsigned int vec_subec (vector unsigned int,
25549 vector unsigned int,
25550 vector unsigned int);
25551 vector signed __int128 vec_subec (vector signed __int128,
25552 vector signed __int128,
25553 vector signed__int128);
25554 vector unsigned __int128 vec_subec (vector unsigned __int128,
25555 vector unsigned __int128,
25556 vector unsigned __int128);
25557 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
25558 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
25559 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
25560 BU_ALTIVEC_OVERLOAD_X definitions.
25561 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
25562 * doc/extend.texi: Update the built-in documentation file for the new
25563 built-in functions.
25565 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25568 * diagnostic-show-locus.c (layout::layout): Use start and finish
25569 spelling location for the start and finish of each range.
25570 * genmatch.c (linemap_client_expand_location_to_spelling_point):
25571 Add unused aspect param.
25572 * input.c (expand_location_1): Add "aspect" param, and use it
25573 to access the correct part of the location.
25574 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
25576 (expand_location_to_spelling_point): Likewise.
25577 (linemap_client_expand_location_to_spelling_point): Add "aspect"
25578 param, and pass it to expand_location_1.
25580 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
25582 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
25583 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
25584 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
25585 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
25586 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
25587 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
25588 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
25589 _mm_maskz_getmant_ss): New intrinsics.
25590 (__builtin_ia32_getexpss128_mask): Changed to ...
25591 __builtin_ia32_getexpss128_round ... this.
25592 (__builtin_ia32_getexpsd128_mask): Changed to ...
25593 __builtin_ia32_getexpsd128_round ... this.
25594 * config/i386/i386-builtin-types.def
25595 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
25596 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
25597 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
25598 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
25599 __builtin_ia32_getmantss_mask_round): New builtins.
25600 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
25601 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
25602 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
25603 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
25604 * config/i386/sse.md
25605 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
25606 avx512f_sgetexp<mode><mask_scalar_name>
25607 <round_saeonly_scalar_name> ... this.
25608 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
25609 %0, %1, %2<round_saeonly_op3>}): Changed to ...
25610 vgetexp<ssescalarmodesuffix>
25611 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25612 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
25613 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
25614 avx512f_vgetmant<mode><mask_scalar_name>
25615 <round_saeonly_scalar_name> ... this.
25616 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
25617 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
25618 vgetmant<ssescalarmodesuffix>
25619 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
25620 %0<mask_scalar_operand4>, %1, %2
25621 <round_saeonly_scalar_mask_op4>, %3} ... this.
25622 * config/i386/subst.md (mask_scalar_operand4,
25623 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
25624 round_saeonly_scalar_nimm_predicate): New subst attributes.
25626 2017-07-06 Julia Koval <julia.koval@intel.com>
25628 * config/i386/i386.c (ix86_erase_embedded_rounding):
25629 Remove code for old rounding pattern.
25631 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
25633 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
25635 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
25637 * doc/sourcebuild.texi (Test Directives, Variants of
25638 dg-require-support): Add documentation for dg-require-stack-check.
25640 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
25642 * config/i386/subst.md (mask_scalar, round_scalar,
25643 round_saeonly_scalar): New meta-templates.
25644 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
25645 round_scalar_mask_operand3, round_scalar_mask_op3,
25646 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
25647 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
25648 round_saeonly_scalar_constraint,
25649 round_saeonly_scalar_prefix): New subst attribute.
25650 * config/i386/sse.md
25651 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
25652 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
25653 <round_scalar_name> ... this.
25654 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
25655 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
25656 <round_scalar_name> ... this.
25657 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
25658 <sse>_vm<code><mode>3<mask_scalar_name>
25659 <round_saeonly_scalar_name> ... this.
25660 (v<plusminus_mnemonic><ssescalarmodesuffix>
25661 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25662 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25663 v<plusminus_mnemonic><ssescalarmodesuffix>
25664 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25665 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25666 (v<multdiv_mnemonic><ssescalarmodesuffix>
25667 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25668 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25669 v<multdiv_mnemonic><ssescalarmodesuffix>
25670 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25671 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25672 (v<maxmin_float><ssescalarmodesuffix>
25673 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
25674 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
25675 v<maxmin_float><ssescalarmodesuffix>
25676 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25677 %0<mask_scalar_operand3>, %1, %<iptr>2
25678 <round_saeonly_scalar_mask_op3>} ... this.
25680 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
25682 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
25683 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
25685 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
25686 Alan Hayward <alan.hayward@arm.com>
25687 David Sherwood <david.sherwood@arm.com>
25689 * combine.c (simplify_if_then_else): Remove "enum" before
25691 * compare-elim.c (can_eliminate_compare): Likewise.
25692 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
25694 (aarch64_lookup_simd_builtin_type): Likewise.
25695 (aarch64_simd_builtin_type): Likewise.
25696 (aarch64_init_simd_builtin_types): Likewise.
25697 (aarch64_simd_expand_args): Likewise.
25698 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
25700 (aarch64_reverse_mask): Likewise.
25701 (aarch64_simd_emit_reg_reg_move): Likewise.
25702 (aarch64_gen_adjusted_ldpstp): Likewise.
25703 (aarch64_ccmp_mode_to_code): Likewise.
25704 (aarch64_operands_ok_for_ldpstp): Likewise.
25705 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25706 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
25708 (aarch64_min_divisions_for_recip_mul): Likewise.
25709 (aarch64_reassociation_width): Likewise.
25710 (aarch64_get_condition_code_1): Likewise.
25711 (aarch64_simd_emit_reg_reg_move): Likewise.
25712 (aarch64_simd_attr_length_rglist): Likewise.
25713 (aarch64_reverse_mask): Likewise.
25714 (aarch64_operands_ok_for_ldpstp): Likewise.
25715 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25716 (aarch64_gen_adjusted_ldpstp): Likewise.
25717 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
25719 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
25720 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
25721 (arm_lookup_simd_builtin_type): Likewise.
25722 (arm_simd_builtin_type): Likewise.
25723 (arm_init_simd_builtin_types): Likewise.
25724 (arm_expand_builtin_args): Likewise.
25725 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
25726 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
25727 (ft32_setup_incoming_varargs): Likewise.
25728 (ft32_function_arg): Likewise.
25729 (ft32_function_arg_advance): Likewise.
25730 (ft32_pass_by_reference): Likewise.
25731 (ft32_arg_partial_bytes): Likewise.
25732 (ft32_valid_pointer_mode): Likewise.
25733 (ft32_addr_space_pointer_mode): Likewise.
25734 (ft32_addr_space_legitimate_address_p): Likewise.
25735 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
25737 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
25738 (ix86_emit_outlined_ms2sysv_restore): Likewise.
25739 (iamcu_alignment): Likewise.
25740 (canonicalize_vector_int_perm): Likewise.
25741 (ix86_noce_conversion_profitable_p): Likewise.
25742 (ix86_mpx_bound_mode): Likewise.
25743 (ix86_operands_ok_for_move_multiple): Likewise.
25744 * config/microblaze/microblaze-protos.h
25745 (microblaze_expand_conditional_branch_reg): Likewise.
25746 * config/microblaze/microblaze.c
25747 (microblaze_expand_conditional_branch_reg): Likewise.
25748 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
25750 (rs6000_reassociation_width): Likewise.
25751 (rs6000_invalid_binary_op): Likewise.
25752 (fusion_p9_p): Likewise.
25753 (emit_fusion_p9_load): Likewise.
25754 (emit_fusion_p9_store): Likewise.
25755 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
25757 (riscv_hard_regno_mode_ok_p): Likewise.
25758 (riscv_address_insns): Likewise.
25759 (riscv_split_symbol): Likewise.
25760 (riscv_legitimize_move): Likewise.
25761 (riscv_function_value): Likewise.
25762 (riscv_hard_regno_nregs): Likewise.
25763 (riscv_expand_builtin): Likewise.
25764 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
25765 (riscv_build_integer): Likewise.
25766 (riscv_split_integer): Likewise.
25767 (riscv_legitimate_constant_p): Likewise.
25768 (riscv_cannot_force_const_mem): Likewise.
25769 (riscv_regno_mode_ok_for_base_p): Likewise.
25770 (riscv_valid_base_register_p): Likewise.
25771 (riscv_valid_offset_p): Likewise.
25772 (riscv_valid_lo_sum_p): Likewise.
25773 (riscv_classify_address): Likewise.
25774 (riscv_legitimate_address_p): Likewise.
25775 (riscv_address_insns): Likewise.
25776 (riscv_load_store_insns): Likewise.
25777 (riscv_force_binary): Likewise.
25778 (riscv_split_symbol): Likewise.
25779 (riscv_force_address): Likewise.
25780 (riscv_legitimize_address): Likewise.
25781 (riscv_move_integer): Likewise.
25782 (riscv_legitimize_const_move): Likewise.
25783 (riscv_legitimize_move): Likewise.
25784 (riscv_address_cost): Likewise.
25785 (riscv_subword): Likewise.
25786 (riscv_output_move): Likewise.
25787 (riscv_canonicalize_int_order_test): Likewise.
25788 (riscv_emit_int_order_test): Likewise.
25789 (riscv_function_arg_boundary): Likewise.
25790 (riscv_pass_mode_in_fpr_p): Likewise.
25791 (riscv_pass_fpr_single): Likewise.
25792 (riscv_pass_fpr_pair): Likewise.
25793 (riscv_get_arg_info): Likewise.
25794 (riscv_function_arg): Likewise.
25795 (riscv_function_arg_advance): Likewise.
25796 (riscv_arg_partial_bytes): Likewise.
25797 (riscv_function_value): Likewise.
25798 (riscv_pass_by_reference): Likewise.
25799 (riscv_setup_incoming_varargs): Likewise.
25800 (riscv_print_operand): Likewise.
25801 (riscv_elf_select_rtx_section): Likewise.
25802 (riscv_save_restore_reg): Likewise.
25803 (riscv_for_each_saved_reg): Likewise.
25804 (riscv_register_move_cost): Likewise.
25805 (riscv_hard_regno_mode_ok_p): Likewise.
25806 (riscv_hard_regno_nregs): Likewise.
25807 (riscv_class_max_nregs): Likewise.
25808 (riscv_memory_move_cost): Likewise.
25809 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
25810 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
25811 (rl78_addr_space_address_mode): Likewise.
25812 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
25814 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
25815 (rs6000_reassociation_width): Likewise.
25816 (rs6000_invalid_binary_op): Likewise.
25817 (fusion_p9_p): Likewise.
25818 (emit_fusion_p9_load): Likewise.
25819 (emit_fusion_p9_store): Likewise.
25820 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
25821 (ok_for_simple_move_operands): Likewise.
25822 (ok_for_simple_move_strict_operands): Likewise.
25823 (ok_for_simple_arith_logic_operands): Likewise.
25824 (visium_legitimize_reload_address): Likewise.
25825 (visium_select_cc_mode): Likewise.
25826 (output_cbranch): Likewise.
25827 (visium_split_double_move): Likewise.
25828 (visium_expand_copysign): Likewise.
25829 (visium_expand_int_cstore): Likewise.
25830 (visium_expand_fp_cstore): Likewise.
25831 * config/visium/visium.c (visium_pass_by_reference): Likewise.
25832 (visium_function_arg): Likewise.
25833 (visium_function_arg_advance): Likewise.
25834 (visium_libcall_value): Likewise.
25835 (visium_setup_incoming_varargs): Likewise.
25836 (visium_legitimate_constant_p): Likewise.
25837 (visium_legitimate_address_p): Likewise.
25838 (visium_legitimize_address): Likewise.
25839 (visium_secondary_reload): Likewise.
25840 (visium_register_move_cost): Likewise.
25841 (visium_memory_move_cost): Likewise.
25842 (prepare_move_operands): Likewise.
25843 (ok_for_simple_move_operands): Likewise.
25844 (ok_for_simple_move_strict_operands): Likewise.
25845 (ok_for_simple_arith_logic_operands): Likewise.
25846 (visium_function_value_1): Likewise.
25847 (rtx_ok_for_offset_p): Likewise.
25848 (visium_legitimize_reload_address): Likewise.
25849 (visium_split_double_move): Likewise.
25850 (visium_expand_copysign): Likewise.
25851 (visium_expand_int_cstore): Likewise.
25852 (visium_expand_fp_cstore): Likewise.
25853 (visium_split_cstore): Likewise.
25854 (visium_select_cc_mode): Likewise.
25855 (visium_split_cbranch): Likewise.
25856 (output_cbranch): Likewise.
25857 (visium_print_operand_address): Likewise.
25858 * expmed.c (flip_storage_order): Likewise.
25859 * expmed.h (emit_cstore): Likewise.
25860 (flip_storage_order): Likewise.
25861 * genrecog.c (validate_pattern): Likewise.
25862 * hsa-gen.c (gen_hsa_addr): Likewise.
25863 * internal-fn.c (expand_arith_overflow): Likewise.
25864 * ira-color.c (allocno_copy_cost_saving): Likewise.
25865 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25866 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25867 (process_invariant_for_inheritance): Likewise.
25868 * lra-eliminations.c (move_plus_up): Likewise.
25869 * omp-low.c (lower_oacc_reductions): Likewise.
25870 * simplify-rtx.c (simplify_subreg): Likewise.
25871 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25872 (TARGET_CHKP_BOUND_MODE): Likewise..
25873 * targhooks.c (default_chkp_bound_mode): Likewise.
25874 (default_setup_incoming_vararg_bounds): Likewise.
25875 * targhooks.h (default_chkp_bound_mode): Likewise.
25876 (default_setup_incoming_vararg_bounds): Likewise.
25877 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25878 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25879 (have_whole_vector_shift): Likewise.
25880 * tree-vect-stmts.c (vectorizable_load): Likewise.
25881 * doc/tm.texi: Regenerate.
25883 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25885 Graceful degrade if Binutils PR21472 is not available.
25888 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25889 .rodata in flash test fails.
25890 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25891 * confgure: Regenerate.
25892 * config.in: Regenerate.
25893 * config/avr/avr.c (avr_asm_named_section)
25894 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25895 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25896 (avr_asm_init_sections): Same.
25898 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25900 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25901 (fma<VH:mode>4_intrinsic): Likewise.
25902 (*fmsub<VCVTF:mode>4): Likewise.
25903 (*fmsub<VH:mode>4_intrinsic): Likewise.
25905 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25908 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25909 Don't depend on "optimize > 0".
25910 (out_movhi_r_mr, out_movqi_mr_r): Same.
25911 (out_movhi_mr_r, out_movqi_r_mr): Same.
25912 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25913 io_address_operand on "optimize > 0".
25915 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25917 * tree-loop-distribution.c: Add general explanantion on the pass.
25918 (generate_loops_for_partition): Mark distributed loop.
25919 (pg_add_dependence_edges): New parameter. Handle alias data
25920 dependence specially and record it in the parameter if asked.
25921 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25922 (init_partition_graph_vertices, add_partition_graph_edge): New.
25923 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25924 (free_partition_graph_vdata, build_partition_graph): New.
25925 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25926 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25927 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25928 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25929 (version_for_distribution_p, finalize_partitions): New.
25930 (distribute_loop): Handle alias data dependence specially. Factor
25931 out loop fusion code as functions and call these functions.
25933 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25935 * tree-loop-distribution.c (classify_partition): New parameter and
25936 better handle reduction statement.
25937 (rdg_build_partitions): Revise comment.
25938 (distribute_loop): Compute statements in all partitions and pass it
25939 to classify_partition.
25941 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25943 * tree-loop-distribution.c (enum partition_type): New.
25944 (struct partition): New field type.
25945 (partition_merge_into): Add parameter. Update partition type.
25946 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25947 (build_rdg_partition_for_vertex): Compute partition type.
25948 (rdg_build_partitions): Dump partition type.
25949 (distribute_loop): Update calls to partition_merge_into.
25951 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25953 * tree-loop-distribution.c (struct ddr_hasher): New.
25954 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25956 (classify_partition): Call get_data_dependence.
25957 (pg_add_dependence_edges): Ditto.
25958 (distribute_loop): Release data dependence hash table.
25960 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25962 * tree-loop-distribution.c (ref_base_address): Delete.
25963 (similar_memory_accesses): Rename ...
25964 (share_memory_accesses): ... to this. Check if partitions access
25965 the same memory reference.
25966 (distribute_loop): Call share_memory_accesses.
25968 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25970 * tree-loop-distribution.c (struct partition): New field recording
25971 its data reference.
25972 (partition_alloc, partition_free): Init and release data refs.
25973 (partition_merge_into): Merge data refs.
25974 (build_rdg_partition_for_vertex): Collect data refs for partition.
25975 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25977 (distribute_loop): Remve data refs from vertice data of partition
25980 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25982 * tree-loop-distribution.c (params.h): Include header file.
25983 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25984 (datarefs_vec): New global var.
25985 (create_rdg_vertices): Use datarefs_vec directly.
25986 (free_rdg): Don't free data references.
25987 (build_rdg): Update use. Don't free data references.
25988 (distribute_loop): Compute global variable for data references.
25989 Bail out if there are too many data references.
25991 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25993 * tree-loop-distribution.c (loop_nest): New global var.
25994 (build_rdg): Use loop directly, rather than loop nest.
25995 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25997 (distribute_loop): Compute global variable loop nest. Update use.
25999 2017-07-05 Bin Cheng <bin.cheng@arm.com>
26001 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
26002 (partition_merge_into): New parameter. Dump reason for fusion.
26003 (distribute_loop): Update use of partition_merge_into.
26005 2017-07-05 Bin Cheng <bin.cheng@arm.com>
26007 * tree-loop-distribution.c (bb_top_order_index): New.
26008 (bb_top_order_index_size, bb_top_order_cmp): New.
26009 (stmts_from_loop): Use topological order.
26010 (pass_loop_distribution::execute): Compute and release topological
26011 order for basic blocks.
26013 2017-07-05 Bin Cheng <bin.cheng@arm.com>
26015 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
26018 2017-07-05 Bin Cheng <bin.cheng@arm.com>
26020 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
26021 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
26022 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
26023 * internal-fn.def (LOOP_DIST_ALIAS): New.
26024 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
26025 (fold_loop_internal_call): ... this.
26026 (vect_loop_dist_alias_call): New function.
26027 (set_uid_loop_bbs): Call fold_loop_internal_call.
26028 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
26031 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
26034 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
26035 Require dead FLAGS_REG at the beginning of a peephole.
26037 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
26040 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
26041 arguments in the call to __builtin_ia32_sbb_u32.
26042 (_subborrow_u64): Swap _X and _Y arguments in the call to
26043 __builtin_ia32_sbb_u64.
26045 2017-07-04 Jakub Jelinek <jakub@redhat.com>
26048 * tree-vrp.c (compare_assert_loc): Turn into a function template
26049 with stable template parameter. Only test if a->e is NULL,
26050 !a->e == !b->e has been verified already. Use e == NULL or
26051 e != NULL instead of e or ! e tests. If stable is true, don't use
26052 iterative_hash_expr, on the other side allow a or b or both NULL
26053 and sort the NULLs last.
26054 (process_assert_insertions): Sort using compare_assert_loc<false>
26055 instead of compare_assert_loc, later sort using
26056 compare_assert_loc<true> before calling process_assert_insertions_for
26057 in a loop. Use break instead of continue once seen NULL pointer.
26059 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26061 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
26062 Cortex-R7 and Cortex-R8 processors.
26064 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
26066 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
26067 uninitialized while src is not.
26069 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
26071 * common/config/arm/arm-common.c: Adjust include path for
26073 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
26074 (arm-cpu.h): Create in build directory. Adjust dependency rules.
26075 (arm-cpu-data.h): Likewise.
26076 (arm-cpu-cdata.h): Likewise.
26077 * config/arm/arm-cpu.h: Delete.
26078 * config/arm/arm-cpu-cdata.h: Delete.
26079 * config/arm/arm-cpu-data.h: Delete.
26081 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
26083 * config/arm/arm-cpus.in (cortex-a55): New.
26084 (cortex-a75): Likewise.
26085 (cortex-a75.cortex-a55): Likewise.
26086 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
26088 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
26089 * config/arm/arm-cpu-cdata.h: Regenerate.
26090 * config/arm/arm-cpu-data.h: Regenerate.
26091 * config/arm/arm-cpu.h: Regenerate.
26092 * config/arm/arm-tables.opt: Regenerate.
26093 * config/arm/arm-tune.md: Regenerate.
26095 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
26097 * haifa-sched.c (sched_create_recovery_edges): Update profile.
26099 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
26101 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
26104 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
26106 PR tree-optimization/81292
26107 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
26108 full_string_p, also call adjust_related_strinfos if the adjustment
26109 is simple, otherwise invalidate related strinfos.
26111 2017-07-04 Martin Liska <mliska@suse.cz>
26114 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
26115 newly created variable as DECL_IGNORED_P.
26117 2017-07-04 Martin Liska <mliska@suse.cz>
26120 * ipa-inline.c (inline_small_functions):
26121 Use xstrdup_for_dump.
26123 2017-07-04 Tom de Vries <tom@codesourcery.com>
26125 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
26127 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
26130 * config/darwin.c (darwin_function_switched_text_sections):
26133 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
26135 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
26137 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
26139 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
26141 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26143 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
26144 min_profitable_iters, and th as inclusive lower bounds.
26145 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
26146 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
26147 for min_profitable_iters and min_profitable_estimate.
26148 (vect_transform_loop): Treat th as an inclusive lower bound.
26149 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
26151 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
26154 * config/darwin.c (darwin_function_switched_text_sections):
26155 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
26156 in two pieces, and suppress the use of buf.
26158 2017-07-03 Nathan Sidwell <nathan@acm.org>
26160 * hash-table.h (hash_table_mod1): Fix indentation.
26162 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26164 PR middle-end/81290
26165 * predict.c (force_edge_cold): Be more careful about propagation
26167 * profile-count.h (profile_probability::guessed,
26168 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
26170 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
26172 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
26174 * doc/invoke.texi (rcpc architecture extension): Document it.
26176 2017-07-03 Richard Biener <rguenther@suse.de>
26178 PR tree-optimization/60510
26179 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
26180 the scalar reduction PHI and use it.
26181 (vectorizable_reduction): Properly guard the single_defuse_cycle
26182 path for non-SLP reduction chains where we cannot use it.
26183 Rework reduc_def/index and vector type deduction. Rework
26184 vector operand gathering during reduction op code-gen.
26185 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
26186 chains dissolve the chain and leave it to non-SLP reduction
26189 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26191 * tree-data-ref.h (dr_alignment): Declare.
26192 * tree-data-ref.c (dr_alignment): New function.
26193 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
26194 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
26196 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
26198 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26200 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
26201 and base_misalignment fields.
26202 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
26203 * tree-data-ref.c: Include builtins.h.
26204 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
26205 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
26206 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
26207 * tree-vect-data-refs.c: Include tree-cfg.h.
26208 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
26209 fields instead of calculating an alignment here.
26210 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
26211 innermost_loop_behavior fields.
26213 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26215 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
26217 (DR_STEP_ALIGNMENT): New macro.
26218 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
26219 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
26220 (create_data_ref): Print it.
26221 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
26222 to tell whether the step preserves vector (mis)alignment.
26223 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26224 Move the check for an integer step and generalise to all INTEGER_CST.
26225 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
26226 Print the outer step alignment.
26228 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26230 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
26231 with offset_alignment.
26232 (DR_ALIGNED_TO): Delete.
26233 (DR_OFFSET_ALIGNMENT): New macro.
26234 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
26235 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
26236 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
26237 (create_data_ref): Likewise.
26238 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26239 (vect_analyze_data_refs): Likewise.
26240 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
26241 creating dummy innermost behavior.
26243 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26245 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
26246 with a "innermost_loop_behavior *" and refeence tree.
26247 * tree-data-ref.c (dr_analyze_innermost): Likewise.
26248 (create_data_ref): Update call accordingly.
26249 * tree-predcom.c (find_looparound_phi): Likewise.
26251 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26253 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
26254 fields with dr_wrt_vec_loop.
26255 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
26256 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
26257 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
26258 (vect_dr_behavior): New function.
26259 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26260 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
26261 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
26262 track whether the step preserves the misalignment.
26263 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26264 Use vect_dr_behavior.
26265 (vect_setup_realignment): Update call accordingly.
26266 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
26267 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
26268 call to vect_create_addr_base_for_vector_ref.
26269 (vect_create_cond_for_align_checks): Likewise.
26270 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
26271 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
26272 (vect_recog_mask_conversion_pattern): Likewise.
26273 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
26274 (new_stmt_vec_info): Remove redundant zeroing.
26276 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
26278 * common/config/arm/arm-common.c (arm_be8_option): New function.
26279 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
26280 (ISA_ARMv6): Add isa_bit_be8.
26281 * config/arm/arm.h (arm_be8_option): Add prototype.
26282 (BE8_SPEC_FUNCTION): New define.
26283 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
26284 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
26285 (mlittle-endian): Similarly.
26286 (mbe8, mbe32): New options.
26287 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
26288 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
26290 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26292 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
26294 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26296 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
26297 (cleanup_tree_cfg_bb): Use it.
26298 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
26300 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
26302 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26305 * loop-doloop.c (add_test): Update profile.
26307 2017-07-03 Martin Liska <mliska@suse.cz>
26310 * sanopt.c (rewrite_usage_of_param): New function.
26311 (sanitize_rewrite_addressable_params): Likewise.
26312 (pass_sanopt::execute): Call rewrite_usage_of_param.
26314 2017-07-03 Richard Biener <rguenther@suse.de>
26316 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
26317 back to using VIEW_CONVERT_EXPR.
26319 2017-07-03 Martin Liska <mliska@suse.cz>
26322 * doc/extend.texi: Document when a resolver function is
26323 generated for target_clones.
26325 2017-07-03 Martin Liska <mliska@suse.cz>
26327 * asan.c (asan_emit_stack_protection): Unpoison just red zones
26328 and shadow memory of auto variables which are subject of
26329 use-after-scope sanitization.
26330 (asan_expand_mark_ifn): Add do set only when is_poison.
26332 2016-07-03 Richard Biener <rguenther@suse.de>
26334 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
26336 (vect_force_simple_reduction): Record reduction def -> phi mapping.
26337 (vectorizable_reduction): Perform reduction PHI creation when
26338 visiting a reduction PHI and adjust and simplify code generation
26339 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
26340 (vect_transform_loop): Visit reduction PHIs.
26341 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
26342 defs into the SLP tree.
26343 (vect_build_slp_tree): Reduction defs terminate the recursion.
26344 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
26346 (vect_get_vec_defs_for_stmt_copy): Export.
26347 (vect_get_vec_defs): Likewise.
26348 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
26350 (vect_get_vec_defs_for_stmt_copy): Declare.
26351 (vect_get_vec_defs): Likewise.
26353 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26355 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
26356 parameter with a "loop" parameter and use it instead of the
26357 loop containing DR_STMT. Don't check simple_iv when doing
26358 BB analysis. Describe the two analysis modes in the comment.
26360 2017-07-03 Tom de Vries <tom@codesourcery.com>
26362 PR tree-optimization/69468
26363 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
26364 (find_same_succ_bb): Handle ignore_edge_flags.
26366 2017-07-03 Tom de Vries <tom@codesourcery.com>
26368 PR tree-optimization/81192
26369 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
26371 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
26373 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
26375 2017-07-03 Tom de Vries <tom@codesourcery.com>
26377 PR tree-optimization/81192
26378 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
26379 BB_SAME_SUCC (bb) == NULL.
26381 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26383 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
26386 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26388 * dumpfile.c: Include profile-count.h
26389 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
26391 (insert_cond_bb): Update profile.
26392 * tree-cfg.h (insert_cond_bb): Update prototype.
26393 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
26394 * tree-dump.c: Do not include tree-cfg.
26396 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26398 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
26400 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26402 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
26405 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26407 * tree-complex.c (expand_complex_div_wide): update profile.
26409 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26410 Alan Hayward <alan.hayward@arm.com>
26411 David Sherwood <david.sherwood@arm.com>
26413 * Makefile.in (MACHMODE_H): Remove insn-modes.h
26414 (CORETYPES_H): New define.
26415 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
26416 (insn-modes-inline.h, s-modes-inline-h): New rules.
26417 (generated_files): Add insn-modes-inline.h.
26418 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
26419 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
26420 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
26421 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
26422 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
26423 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
26424 (build/gencodes.o, build/genconditions.o): Likewise.
26425 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
26426 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
26427 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
26428 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
26429 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
26430 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
26431 * coretypes.h: Include everything up to real.h for generators.
26432 Include insn-modes.h first. Include wide-int-print.h after
26433 wide-int.h. Include insn-modes-inline.h and then machmode.h.
26434 * machmode.h: Don't include insn-modes.h here.
26435 * function-tests.c: Remove includes of signop.h, machmode.h,
26436 double-int.h and wide-int.h.
26438 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
26440 * optc-save-gen.awk: Likewise.
26441 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
26442 * godump.c: Remove include of wide-int-print.h.
26443 * pretty-print.h: Likewise.
26444 * wide-int-print.cc: Likewise.
26445 * wide-int.cc: Likewise.
26446 * hash-map-tests.c: Remove include of signop.h.
26447 * hash-set-tests.c: Likewise.
26448 * rtl-tests.c: Likewise.
26449 * mkconfig.sh: Remove include of machmode.h.
26450 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
26452 (emit_insn_modes_inline_h): ...this new function. Emit the code
26453 into an insn-modes-inline.h header file, adding appropriate
26454 include guards and end comments.
26455 (emit_insn_modes_c_header): Remove include of machmode.h.
26456 (emit_min_insn_modes_c_header): Include coretypes.h rather than
26458 (main): Handle -i flag and call emit_insn_modes_inline_h when
26461 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26463 * tree-ssa-strlen.c (strinfo): Rename the length field to
26464 nonzero_chars. Add a full_string_p field.
26465 (compare_nonzero_chars, zero_length_string_p): New functions.
26466 (get_addr_stridx): Add an offset_out parameter.
26467 Use compare_nonzero_chars.
26468 (get_stridx): Update accordingly. Use compare_nonzero_chars.
26469 (new_strinfo): Update after above changes to strinfo.
26470 (set_endptr_and_length): Set full_string_p.
26471 (get_string_length): Update after above changes to strinfo.
26472 (unshare_strinfo): Update call to new_strinfo.
26473 (maybe_invalidate): Likewise.
26474 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
26475 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
26476 as a uhwi instead of an shwi. Update after above changes to
26477 strinfo and new_strinfo.
26478 (zero_length_string): Assert that chainsi contains full strings.
26479 Use zero_length_string_p. Update call to new_strinfo.
26480 (adjust_related_strinfos): Update after above changes to strinfo.
26481 Copy full_string_p from origsi.
26482 (adjust_last_stmt): Use zero_length_string_p.
26483 (handle_builtin_strlen): Update after above changes to strinfo and
26484 new_strinfo. Install the lhs as the string length if the previous
26485 entry didn't describe a full string.
26486 (handle_builtin_strchr): Update after above changes to strinfo
26488 (handle_builtin_strcpy): Likewise.
26489 (handle_builtin_strcat): Likewise.
26490 (handle_builtin_malloc): Likewise.
26491 (handle_pointer_plus): Likewise.
26492 (handle_builtin_memcpy): Likewise. Track nonzero characters
26493 that aren't necessarily followed by a nul terminator.
26494 (handle_char_store): Likewise.
26496 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26498 PR tree-optimization/80769
26499 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
26500 for malloc and calloc. Document the new invariant that all related
26501 strinfos have delayed lengths or none do.
26502 (verify_related_strinfos): Move earlier in file.
26503 (set_endptr_and_length): New function, split out from...
26504 (get_string_length): ...here. Also set the lengths of related
26506 (zero_length_string): Assert that chainsi has known (rather than
26508 (adjust_related_strinfos): Likewise.
26510 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26512 PR tree-optimization/81136
26513 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
26514 assert that two references with the same misalignment have the same
26515 compile-time misalignment if those compile-time misalignments
26518 2017-07-01 Andi Kleen <ak@linux.intel.com>
26520 * print-tree.c (print_node): Print all attributes.
26522 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26524 * cfg.c (scale_bbs_frequencies): New function.
26525 * cfg.h (scale_bbs_frequencies): Declare it.
26526 * cfgloopanal.c (single_likely_exit): Cleanup.
26527 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
26529 (scale_loop_profile): Likewise.
26530 (loop_version): Likewise.
26531 (create_empty_loop_on_edge): Update.
26532 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
26533 scale_loop_frequencies, scale_loop_profile, loopify,
26534 loop_version): Update prototypes.
26535 * modulo-sched.c (sms_schedule): Update.
26536 * predict.c (unlikely_executed_edge_p): Also check probability.
26537 (probably_never_executed_edge_p): Fix typo.
26538 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26539 * tree-parloops.c (gen_parallel_loop): Update.
26540 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
26541 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26542 * tree-ssa-loop-split.c (split_loop): Update.
26543 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26544 * tree-vect-loop-manip.c (vect_do_peeling): Update.
26545 (vect_loop_versioning): Update.
26546 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26548 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26550 * trans-mem.c (split_bb_make_tm_edge): Update profile.
26552 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26554 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
26555 to keep profile consistent.
26557 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26559 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
26560 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
26561 * profile-count.h (max_safe_multiplier): Make unsigned.
26562 (profile_count::guessed_zero): New.
26564 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26566 * bb-reorder.c (fix_up_crossing_landing_pad,
26567 fix_crossing_conditional_branches): Use make_single_succ_edge
26568 to keep profile consistent.
26570 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26572 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
26575 2017-07-01 Jakub Jelinek <jakub@redhat.com>
26578 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
26579 the right scopes, make sure cond_jump isn't preserved between multiple
26580 iterations. Search for fallthru edge whenever there are 3+ edges and
26581 use find_fallthru_edge for it.
26583 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26585 Patch by Alexander Monakov <amonakov@ispras.ru>
26586 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
26587 probabilities consistently.
26589 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26591 * pa.c (pa_expand_compare_and_swap_loop): Update call of
26592 emit_cmp_and_jump_insns.
26594 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26597 * tree-inline.c (expand_call_inline): Combine profile statuses.
26599 2017-06-30 Andrew Pinski <apinski@cavium.com>
26601 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
26602 fold_stmt returned true.
26604 2017-06-30 Nathan Sidwell <nathan@acm.org>
26606 * ggc.h (empty_string): Delete.
26607 * cfgexpand.c (expand_asm_stmt): Use plain "".
26608 * optabs.c (expand_asm_memory_barrier): Likewise.
26609 * stringpool.c (empty_string): Delete.
26610 (digit_vector, digit_string): Delete.
26611 (ggc_alloc_string): Use plain "", don't optimize single digit
26612 strings. Use ggc_alloc_atomic.
26614 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
26616 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
26617 comparison set and one other set, use the cost of the non-comparison
26620 2017-06-30 Nathan Sidwell <nathan@acm.org>
26622 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
26625 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
26627 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
26628 loops. Remove now unneeded calls to gimple_switch_set_label() that
26629 just set removed labels to NULL_TREE.
26631 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
26633 * tree-ssanames.c (set_range_info_raw): Abstract from ...
26634 (set_range_info): ...here. Only call set_range_info_raw if domain
26636 (set_nonzero_bits): Call set_range_info_raw.
26637 * tree-ssanames.h (set_range_info_raw): New.
26639 2017-06-30 Jakub Jelinek <jakub@redhat.com>
26642 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
26643 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
26644 of nonimmediate_operand and <store_mask_constraint> instead of m
26645 for the input operand. For V8FI iterator, always split if input
26646 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
26647 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
26648 <store_mask_predicate> instead of register_operand and
26649 <store_mask_constraint> instead of v for the input operand. Make
26650 sure both operands aren't MEMs for if not <mask_applied>.
26652 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
26654 * lto-wrapper.c (copy_file) Close both file descriptors before
26657 2017-06-30 Martin Liska <mliska@suse.cz>
26660 * multiple_target.c (create_dispatcher_calls): Make ifunc
26661 also for function that don't have calls or are not referenced.
26663 2017-06-30 Richard Biener <rguenther@suse.de>
26665 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
26666 analyze the first scalar stmt. Move vector type computation
26667 for the BB case here from ...
26668 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
26669 live operation processing in the SLP case properly.
26671 2017-06-30 Richard Biener <rguenther@suse.de>
26673 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
26675 2017-06-30 Martin Liska <mliska@suse.cz>
26678 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
26679 before BUILT_IN_UNWIND_RESUME when ASAN is used.
26681 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
26683 * doc/invoke.texi (AArch64): Add missing options and remove redundant
26686 2017-06-30 Richard Biener <rguenther@suse.de>
26688 PR tree-optimization/81249
26689 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
26690 condition reduction result to original scalar type.
26692 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26694 * profile-count.h (enum profile_quality): Fix typos and whitespace
26697 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26699 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
26700 type for branch probabilities.
26702 2017-06-29 Julian Brown <julian@codesourcery.com>
26703 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26705 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
26706 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
26707 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
26708 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
26710 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26712 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
26713 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
26714 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
26715 CC usage from generic code to here.
26716 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
26717 CC usage into the target macros.
26719 2017-06-29 Maya Rashish <coypu@sdf.org>
26721 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
26724 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26726 * arm/arm-builtins.c: Include profile-count.h
26727 * except.c (sjlj_emit_function_enter): Use
26728 profile_probability::unlikely.
26730 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26732 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
26733 and tocrel_offset be pointer args rather than implicitly using
26735 (legitimate_constant_pool_address_p, rs6000_emit_move,
26736 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
26737 tocrel_offset and use in toc_relative_expr_p call.
26738 (print_operand, print_operand_address): Use static tocrel_base_oac
26739 and tocrel_offset_oac.
26740 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
26743 2017-06-29 Maya Rashish <coypu@sdf.org>
26745 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
26747 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
26749 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
26750 objects, take into account only the alignment of 'op0' and 'mode1' if
26753 2017-06-29 Steve Ellcey <sellcey@cavium.com>
26755 * ccmp.c (ccmp_tree_comparison_p): New function.
26756 (ccmp_candidate_p): Update to use above function.
26757 (get_compare_parts): New function.
26758 (expand_ccmp_next): Update to use new functions.
26759 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
26761 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
26762 take mode as argument.
26763 * ccmp.h (expand_ccmp_expr): Add mode as argument.
26764 * expr.c (expand_expr_real_1): Pass mode as argument.
26766 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
26768 * combine.c (combine_instructions): Print insns to dump_file, together
26771 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26773 * asan.c (asan_emit_stack_protection): Update.
26774 (create_cond_insert_point): Update.
26775 * auto-profile.c (afdo_propagate_circuit): Update.
26776 * basic-block.h (struct edge_def): Turn probability to
26777 profile_probability.
26778 (EDGE_FREQUENCY): Update.
26779 * bb-reorder.c (find_traces_1_round): Update.
26780 (better_edge_p): Update.
26781 (sanitize_hot_paths): Update.
26782 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
26783 (make_single_succ_edge): Update.
26784 (check_bb_profile): Update.
26785 (dump_edge_info): Update.
26786 (update_bb_profile_for_threading): Update.
26787 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
26788 probabilitycount to 0.
26789 * cfgbuild.c (compute_outgoing_frequencies): Update.
26790 * cfgcleanup.c (try_forward_edges): Update.
26791 (outgoing_edges_match): Update.
26792 (try_crossjump_to_edge): Update.
26793 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
26794 (expand_gimple_tailcall): Update.
26795 (construct_init_block): Use make_single_succ_edge.
26796 (construct_exit_block): Use make_single_succ_edge.
26797 * cfghooks.c (verify_flow_info): Update.
26798 (redirect_edge_succ_nodup): Update.
26799 (split_edge): Update.
26800 (account_profile_record): Update.
26801 * cfgloopanal.c (single_likely_exit): Update.
26802 * cfgloopmanip.c (scale_loop_profile): Update.
26803 (set_zero_probability): Remove.
26804 (duplicate_loop_to_header_edge): Update.
26805 * cfgloopmanip.h (loop_version): Update prototype.
26806 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
26807 (force_nonfallthru_and_redirect): Update.
26808 (update_br_prob_note): Update.
26809 (rtl_verify_edges): Update.
26810 (purge_dead_edges): Update.
26811 (rtl_lv_add_condition_to_bb): Update.
26812 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
26813 * cgraphunit.c (init_lowered_empty_function): Update.
26814 (cgraph_node::expand_thunk): Update.
26815 * cilk-common.c: Include profile-count.h
26816 * dojump.c (inv): Remove.
26817 (jumpifnot): Update.
26818 (jumpifnot_1): Update.
26819 (do_jump_1): Update.
26821 (do_jump_by_parts_greater_rtx): Update.
26822 (do_compare_rtx_and_jump): Update.
26823 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
26824 do_jump_1. do_compare_rtx_and_jump): Update prototype.
26825 * dwarf2cfi.c: Include profile-count.h
26826 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
26827 (sjlj_emit_dispatch_table): Likewise.
26828 * explow.c: Include profile-count.h
26829 * expmed.c (emit_store_flag_force): Update.
26830 (do_cmp_and_jump): Update.
26831 * expr.c (compare_by_pieces_d::generate): Update.
26832 (compare_by_pieces_d::finish_mode): Update.
26833 (emit_block_move_via_loop): Update.
26834 (store_expr_with_bounds): Update.
26835 (store_constructor): Update.
26836 (expand_expr_real_2): Update.
26837 (expand_expr_real_1): Update.
26838 * expr.h (try_casesi, try_tablejump): Update prototypes.
26839 * gimple-pretty-print.c (dump_probability): Update.
26840 (dump_profile): New.
26841 (dump_gimple_label): Update.
26842 (dump_gimple_bb_header): Update.
26843 * graph.c (draw_cfg_node_succ_edges): Update.
26844 * hsa-gen.c (convert_switch_statements): Update.
26845 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
26846 (find_if_case_1): Update.
26847 (find_if_case_2): Update.
26848 * internal-fn.c (expand_arith_overflow_result_store): Update.
26849 (expand_addsub_overflow): Update.
26850 (expand_neg_overflow): Update.
26851 (expand_mul_overflow): Update.
26852 (expand_vector_ubsan_overflow): Update.
26853 * ipa-cp.c (good_cloning_opportunity_p): Update.
26854 * ipa-split.c (split_function): Use make_single_succ_edge.
26855 * ipa-utils.c (ipa_merge_profiles): Update.
26856 * loop-doloop.c (add_test): Update.
26857 (doloop_modify): Update.
26858 * loop-unroll.c (compare_and_jump_seq): Update.
26859 (unroll_loop_runtime_iterations): Update.
26860 * lra-constraints.c (lra_inheritance): Update.
26861 * lto-streamer-in.c (input_cfg): Update.
26862 * lto-streamer-out.c (output_cfg): Update.
26863 * mcf.c (adjust_cfg_counts): Update.
26864 * modulo-sched.c (sms_schedule): Update.
26865 * omp-expand.c (expand_omp_for_init_counts): Update.
26866 (extract_omp_for_update_vars): Update.
26867 (expand_omp_ordered_sink): Update.
26868 (expand_omp_for_ordered_loops): Update.
26869 (expand_omp_for_generic): Update.
26870 (expand_omp_for_static_nochunk): Update.
26871 (expand_omp_for_static_chunk): Update.
26872 (expand_cilk_for): Update.
26873 (expand_omp_simd): Update.
26874 (expand_omp_taskloop_for_outer): Update.
26875 (expand_omp_taskloop_for_inner): Update.
26876 * omp-simd-clone.c (simd_clone_adjust): Update.
26877 * optabs.c (expand_doubleword_shift): Update.
26878 (expand_abs): Update.
26879 (emit_cmp_and_jump_insn_1): Update.
26880 (expand_compare_and_swap_loop): Update.
26881 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26882 * predict.c (predictable_edge_p): Update.
26883 (edge_probability_reliable_p): Update.
26884 (set_even_probabilities): Update.
26885 (combine_predictions_for_insn): Update.
26886 (combine_predictions_for_bb): Update.
26887 (propagate_freq): Update.
26888 (estimate_bb_frequencies): Update.
26889 (force_edge_cold): Update.
26890 * profile-count.c (profile_count::dump): Add missing space into dump.
26891 (profile_count::debug): Add newline.
26892 (profile_count::differs_from_p): Explicitly convert to unsigned.
26893 (profile_count::stream_in): Update.
26894 (profile_probability::dump): New member function.
26895 (profile_probability::debug): New member function.
26896 (profile_probability::differs_from_p): New member function.
26897 (profile_probability::differs_lot_from_p): New member function.
26898 (profile_probability::stream_in): New member function.
26899 (profile_probability::stream_out): New member function.
26900 * profile-count.h (profile_count_quality): Rename to ...
26901 (profile_quality): ... this one.
26902 (profile_probability): New.
26903 (profile_count): Update.
26904 * profile.c (compute_branch_probabilities): Update.
26905 * recog.c (peep2_attempt): Update.
26906 * sched-ebb.c (schedule_ebbs): Update.
26907 * sched-rgn.c (find_single_block_region): Update.
26908 (compute_dom_prob_ps): Update.
26909 (schedule_region): Update.
26910 * sel-sched-ir.c (compute_succs_info): Update.
26911 * stmt.c (struct case_node): Update.
26912 (do_jump_if_equal): Update.
26913 (get_outgoing_edge_probs): Update.
26914 (conditional_probability): Update.
26915 (emit_case_dispatch_table): Update.
26916 (expand_case): Update.
26917 (expand_sjlj_dispatch_table): Update.
26918 (emit_case_nodes): Update.
26919 * targhooks.c: Update.
26920 * tracer.c (better_p): Update.
26921 (find_best_successor): Update.
26922 * trans-mem.c (expand_transaction): Update.
26923 * tree-call-cdce.c: Update.
26924 * tree-cfg.c (gimple_split_edge): Upate.
26925 (move_sese_region_to_fn): Upate.
26926 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26927 * tree-eh.c (lower_resx): Upate.
26928 (cleanup_empty_eh_move_lp): Upate.
26929 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26930 * tree-inline.c (copy_edges_for_bb): Update.
26931 (copy_cfg_body): Update.
26932 * tree-parloops.c (gen_parallel_loop): Update.
26933 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26934 (gimple_gen_time_profiler): Update.
26935 * tree-ssa-dce.c (remove_dead_stmt): Update.
26936 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26937 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26938 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26939 (unloop_loops): Update.
26940 (try_peel_loop): Update.
26941 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26942 * tree-ssa-loop-split.c (connect_loops): Update.
26943 (split_loop): Update.
26944 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26945 (hoist_guard): Update.
26946 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26947 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26948 (value_replacement): Update.
26949 * tree-ssa-reassoc.c (branch_fixup): Update.
26950 * tree-ssa-tail-merge.c (replace_block_by): Update.
26951 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26952 (create_edge_and_update_destination_phis): Update.
26953 (compute_path_counts): Update.
26954 (recompute_probabilities): Update.
26955 (update_joiner_offpath_counts): Update.
26956 (freqs_to_counts_path): Update.
26957 (duplicate_thread_path): Update.
26958 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26959 (struct switch_conv_info): Update.
26960 (gen_inbound_check): Update.
26961 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26962 (vect_do_peeling): Update.
26963 (vect_loop_versioning): Update.
26964 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26965 (optimize_mask_stores): Update.
26966 * ubsan.c (ubsan_expand_null_ifn): Update.
26967 * value-prof.c (gimple_divmod_fixed_value): Update.
26968 (gimple_divmod_fixed_value_transform): Update.
26969 (gimple_mod_pow2): Update.
26970 (gimple_mod_pow2_value_transform): Update.
26971 (gimple_mod_subtract): Update.
26972 (gimple_mod_subtract_transform): Update.
26973 (gimple_ic): Update.
26974 (gimple_stringop_fixed_value): Update.
26975 (gimple_stringops_transform): Update.
26976 * value-prof.h: Update.
26978 2017-06-29 Carl Love <cel@us.ibm.com>
26980 * config/rs6000/rs6000-c.c: Add support for built-in functions
26981 vector signed int vec_signed (vector float);
26982 vector signed long long vec_signed (vector double);
26983 vector signed int vec_signed2 (vector double, vector double);
26984 vector signed int vec_signede (vector double);
26985 vector signed int vec_signedo (vector double);
26986 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26987 instruction generator.
26988 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26989 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26990 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26992 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26993 vunsignede_v2df): Add define_expands.
26994 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26995 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26996 VEC_UNSIGNEDO): Add definitions.
26997 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26998 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26999 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
27000 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
27001 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
27002 * config/rs6000/altivec.h (vec_signed, vec_signed2,
27003 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
27004 vec_unsignede, vec_unsignedo): Add builtin defines.
27005 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
27007 * doc/extend.texi: Update the built-in documentation file for the
27008 new built-in functions.
27010 2017-06-29 Richard Biener <rguenther@suse.de>
27012 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
27013 reduction chains to LOOP_VINFO_REDUCTIONS.
27014 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
27015 SLP reductions after processing reduction chains.
27017 2017-06-29 Nathan Sidwell <nathan@acm.org>
27019 * builtins.c (fold_builtin_FUNCTION): Use
27020 lang_hooks.decl_printable_name.
27022 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
27024 PR middle-end/81194
27025 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
27026 with only one label.
27027 * stmt.c (expand_case): Assert NCASES is greater than one.
27029 2017-06-29 Richard Biener <rguenther@suse.de>
27031 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
27033 (group_case_labels): Likewise.
27034 (find_taken_edge): Push sanity checking on val to workers...
27035 (find_taken_edge_cond_expr): ... here
27036 (find_taken_edge_switch_expr): ... and here, handle cases
27037 with just a default label.
27038 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
27039 (group_case_labels): Likewise.
27040 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
27041 group_case_labels does anything cleanup the CFG again.
27043 2017-06-29 Bin Cheng <bin.cheng@arm.com>
27045 PR tree-optimization/81196
27046 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
27047 exit condition comparing two IVs.
27049 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
27051 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
27052 profile to the dummy entry at the end of the list of architectures.
27053 * config/arm/arm-cpu-cdata.h: Regenerated.
27055 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27056 Michael Collison <michael.collison@arm.com>
27059 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
27061 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
27062 (*aarch64_reg_<mode>3_minus_mask): New pattern.
27063 (*aarch64_<optab>_reg_di3_mask2): New pattern.
27064 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
27065 of shift when the shift amount is masked with constant equal to
27066 the size of the mode.
27067 * config/aarch64/predicates.md (subreg_lowpart_operator): New
27070 2017-06-29 Martin Liska <mliska@suse.cz>
27072 * config/i386/i386.opt: Change range from [1,5] to [0,5].
27074 2017-06-29 Yury Gribov <tetra2005@gmail.com>
27077 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
27080 (edge_growth_cache_entry::edge_growth_cache_entry): New
27082 (reset_edge_growth_cache): Update to use constructor.
27084 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27086 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
27087 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
27088 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
27090 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
27092 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
27093 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
27095 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
27097 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
27098 (*-linux-uclibc*): Add t-uclibc tmake_file.
27099 * config/t-musl: New.
27100 * config/t-uclibc: New.
27102 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
27104 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
27106 (gen_comm_data): Emit architectural setting of arch_prof.
27107 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
27109 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
27110 (armv8-m.base, armv8-m.main): Likewise.
27111 * arm-protos.h (arm_build_target): Add profile field.
27112 (arch_option): Likewise.
27113 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
27115 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
27116 arm_active_target.profile.
27118 2017-06-28 Richard Biener <rguenther@suse.de>
27120 PR middle-end/81227
27121 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
27122 TYPE_OVERFLOW_WRAPS.
27123 * match.pd (negate_expr_p): Likewise.
27124 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
27125 fold_build2, not fold_binary.
27127 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
27129 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
27130 Convert memory address to Pmode.
27131 (aarch64_print_operand): Assert MEM operands are always Pmode.
27133 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
27136 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
27137 Remove redundant if.
27138 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
27139 * config/arm/aarch-common-protos.h
27140 (aarch_forward_to_shift_is_not_shifted_re): Remove.
27141 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
27143 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
27146 * multiple_target.c (create_dispatcher_calls): Set the default
27147 clone to be static, not public.
27149 2017-06-28 Richard Biener <rguenther@suse.de>
27151 * tree-vect-loop.c (vectorizable_reduction): Move special
27152 cond reduction IV var creation ...
27153 (vect_create_epilog_for_reduction): ... here. Remove induction_index
27154 parameter. Use STMT_VINFO_VECTYPE.
27155 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
27158 2017-06-28 Martin Liska <mliska@suse.cz>
27161 * ipa-visibility.c (non_local_p): Handle visibility.
27163 2017-06-28 Martin Liska <mliska@suse.cz>
27166 * common.opt: Add IntegerRange to various options.
27167 * opt-functions.awk (integer_range_info): New function.
27168 * optc-gen.awk: Add integer_range_info to cl_options struct.
27169 * opts-common.c (decode_cmdline_option): Handle
27170 CL_ERR_INT_RANGE_ARG.
27171 (cmdline_handle_error): Likewise.
27172 * opts.c (print_filtered_help): Show valid interval in
27173 when --help is provided.
27174 * opts.h (struct cl_option): Add range_min and range_max fields.
27175 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
27177 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
27179 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
27180 (x * C EQ/NE y * C): New transformation.
27182 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
27184 * genmultilib (combination_space): Accept '+' in option names.
27186 2017-06-28 Martin Liska <mliska@suse.cz>
27189 * asan.c (instrument_derefs): Bail out inner references
27190 that are hard register variables.
27192 2017-06-28 Jakub Jelinek <jakub@redhat.com>
27195 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
27196 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
27198 2017-06-28 Richard Biener <rguenther@suse.de>
27200 * tree-vectorizer.h (vect_get_vec_defs): Remove.
27201 (vect_get_slp_defs): Adjust.
27202 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
27204 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
27206 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
27207 get_initial_defs_for_reduction instead of vect_get_vec_defs.
27208 (vectorizable_reduction): Adjust.
27209 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
27211 (vect_get_slp_defs): Likewise.
27212 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
27213 (vectorizable_bswap): Adjust.
27214 (vectorizable_call): Likewise.
27215 (vectorizable_conversion): Likewise.
27216 (vectorizable_assignment): Likewise.
27217 (vectorizable_shift): Likewise.
27218 (vectorizable_operation): Likewise.
27219 (vectorizable_store): Likewise.
27220 (vectorizable_condition): Likewise.
27221 (vectorizable_comparison): Likewise.
27223 2017-06-28 Michael Collison <michael.collison@arm.com>
27226 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
27228 (arm_gen_movmemqi): Removed unused variable 'i'.
27229 Convert 'for' loop into 'while' loop.
27230 (arm_expand_prologue): Remove last unnecessary set of insn.
27231 (thumb_pop): Remove unused variable 'pushed_words'.
27232 (thumb_exit): Remove last unnecessary set of regs_to_pop.
27234 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27236 * config/s390/predicates.md: Use s390_rel_address_ok_p.
27237 * config/s390/s390-protos.h: Add prototype of
27238 s390_rel_address_ok_p.
27239 * config/s390/s390.c (s390_got_symbol): New function.
27240 (s390_rel_address_ok_p): New function.
27241 (legitimize_pic_address): Use s390_rel_address_ok_p.
27242 (s390_load_got): Use s390_got_symbol.
27243 (s390_option_override): Issue error if
27244 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
27245 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
27247 * config/s390/s390.opt: New option mpic-data-is-text-relative.
27249 2017-06-27 Andrew Pinski <apinski@cavium.com>
27251 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
27252 (X * copysign (1.0, X)): New pattern.
27253 (X * copysign (1.0, -X)): New pattern.
27254 (copysign (-1.0, CST)): New pattern.
27256 2017-06-27 Joseph Myers <joseph@codesourcery.com>
27258 * genmultilib (combination_space): Remove variable.
27259 Validate reuse rules against regular expression for any sequence
27260 of multilib options in any order.
27262 2017-06-27 Michael Collison <michael.collison@arm.com>
27264 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
27265 call aarch64_split_simd_combine.
27266 * (aarch64_combine_internal<mode>): Delete pattern.
27267 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
27268 Allow register and subreg operands.
27270 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27272 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
27273 specific need, just fallback on defaults.
27274 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
27276 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27277 Olivier Hainque <hainque@adacore.com>
27279 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
27281 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
27282 targets. Pick a default if no particular attempt applied.
27283 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
27286 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27288 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
27289 (x86_64-wrs-vxworks7): Likewise.
27291 2017-06-27 Marek Polacek <polacek@redhat.com>
27294 * ubsan.c (instrument_null): Check get_base_address's result for null.
27296 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27298 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
27300 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27302 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
27303 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
27304 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
27305 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
27306 New function types.
27307 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
27308 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
27309 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
27310 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
27311 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
27312 BUILT_IN_FEUPDATEENV): New builtins.
27313 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
27314 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
27315 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
27316 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
27318 (builtin_structptr_types): Adjust size.
27319 * tree.c (builtin_structptr_types): Add four entries.
27321 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27322 Olivier Hainque <hainque@adacore.com>
27324 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
27325 (TLS_SYM): New local macro, forcing reference to __tls__ on
27326 link command lines for VxWorks 7 RTPs, triggering initialization
27328 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
27329 OS features TLS support, true for RTPs on VxWorks 7.
27330 * config/vxworks.c (vxworks_override_options): Setup emutls
27333 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27335 * predict.c (test_prediction_value_range): Use -1U instead of -1
27336 to avoid narrowing conversion warning.
27337 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
27338 to avoid narrowing conversion warning.
27339 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
27341 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
27343 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27345 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
27346 64bit configurations.
27347 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
27348 (SIZE_TYPE): Likewise.
27349 * config/vxworks.c (vxworks_emutls_var_fields): Use
27350 long_unsigned_type_node instead of unsigned_type_node as the offset
27351 field type, which is "pointer" mode in emutls.c.
27353 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27356 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
27358 PR middle-end/81207
27359 * gimple-fold.c (replace_call_with_call_and_fold): Handle
27360 gimple_vuse copying separately from gimple_vdef copying.
27362 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27364 * value-prof.c (free_hist): Remove call to memset and the enclosing if
27367 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27368 Olivier Hainque <hainque@adacore.com>
27370 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
27371 for all vxworks7 targets.
27372 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
27373 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
27374 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
27375 variations for VX6/VX7 and 32/64bits later on in ...
27376 (VXWORKS_LIB_SPEC): Leverage new macros.
27377 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
27378 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
27380 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27382 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
27383 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
27385 2017-06-26 Carl Love <cel@us.ibm.com>
27387 * config/rs6000/rs6000-c.c: Add support for built-in functions
27388 vector bool char vec_reve (vector bool char);
27389 vector signed char vec_reve (vector signed char);
27390 vector unsigned char vec_reve (vector unsigned char);
27391 vector bool int vec_reve (vector bool int);
27392 vector signed int vec_reve (vector signed int);
27393 vector unsigned int vec_reve (vector unsigned int);
27394 vector bool long long vec_reve (vector bool long long);
27395 vector signed long long vec_reve (vector signed long long);
27396 vector unsigned long long vec_reve (vector unsigned long long);
27397 vector bool short vec_reve (vector bool short);
27398 vector signed short vec_reve (vector signed short);
27399 vector double vec_reve (vector double);
27400 vector float vec_reve (vector float);
27401 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
27402 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
27403 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
27404 (altivec_vreve): New pattern.
27405 * config/rs6000/altivec.h (vec_reve): New define.
27406 * doc/extend.texi (vec_rev): Update the built-in documentation file
27407 for the new built-in functions.
27409 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27411 PR tree-optimization/71815
27412 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
27414 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
27416 (slsr_process_phi): Likewise.
27417 (replace_uncond_cands_and_profitable_phis): Don't replace a
27418 multiply candidate with a stride of 1 (copy or cast).
27419 (phi_incr_cost): Call uses_consumed_by_stmt rather than
27421 (lowest_cost_path): Likewise.
27422 (total_savings): Likewise.
27424 2017-06-26 Richard Biener <rguenther@suse.de>
27427 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
27428 Use def_builtin_pure for all gather builtins.
27430 2017-06-26 Richard Biener <rguenther@suse.de>
27432 PR tree-optimization/81203
27433 * tree-tailcall.c (find_tail_calls): Do not move stmts into
27434 non-dominating BBs.
27436 2017-06-26 Marek Polacek <polacek@redhat.com>
27439 * doc/invoke.texi: Document -Wmultistatement-macros.
27441 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
27443 * doc/sourcebuild.texi (ARM-specific attributes): Document new
27444 arm_neon_ok_no_float_abi effective target.
27446 2017-06-26 Richard Biener <rguenther@suse.de>
27448 PR tree-optimization/80928
27449 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
27450 (copy_bbs): Set BB_DUPLICATED flag early.
27451 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
27453 (execute_on_shrinking_pred): Likewise.
27454 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
27455 BB_DUPLICATED blocks.
27456 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
27457 iterate over all PHIs considering removal of *gsi.
27459 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
27461 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
27464 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27466 * config/rs6000/rs6000-string.c: (expand_block_clear,
27467 do_load_for_compare, select_block_compare_mode,
27468 compute_current_alignment, expand_block_compare,
27469 expand_strncmp_align_check, expand_strn_compare,
27470 expand_block_move, rs6000_output_load_multiple)
27471 Move functions related to string/block move/compare
27472 to a separate file.
27473 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
27474 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
27475 for this function which is now used in two files.
27476 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
27477 * config.gcc: Add rs6000-string.o to extra_objs for
27478 targets powerpc*-*-* and rs6000*-*-*.
27480 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
27483 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
27484 32-bit, since indexed is not valid for DImode.
27485 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
27486 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
27487 (define_peephole2 for Altivec d-form load): Add 32-bit support.
27488 (define_peephole2 for Altivec d-form store): Likewise.
27491 * multiple_target.c (create_dispatcher_calls): Only create the
27492 dispatcher call if the function is the default clone of a
27493 versioned function.
27495 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
27497 PR middle-end/80902
27498 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
27499 a call, force the call to not be a tail call.
27501 2017-06-23 Jeff Law <law@redhat.com>
27503 * doc/contrib.texi: Add entry for Steven Pemberton's work on
27506 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
27508 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
27509 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
27510 handling for early expansion of vector shifts (sl,sr,sra,rl).
27511 (builtin_function_type): Add vector shift right instructions
27512 to the unsigned argument list.
27514 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
27516 rtl-optimizatoin/79286
27517 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
27518 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
27519 trap. PIC register plus a const unspec without offset can never trap.
27521 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
27523 * tree.h (builtin_structptr_type): New type.
27524 (builtin_structptr_types): Declare new array.
27525 * tree.c (builtin_structptr_types): New array.
27526 (free_lang_data, build_common_tree_nodes): Use it.
27528 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
27531 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
27534 2017-06-22 Matt Turner <mattst88@gmail.com>
27536 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
27537 Lake models to skylake case. Assume skylake for unknown
27538 models with clflushopt.
27540 2017-06-22 Jeff Law <law@redhat.com>
27542 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
27543 frame sizes that do not satisfy aarch64_uimm12_shift.
27545 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
27547 * profile-count.h (apply_probability,
27548 apply_scale, probability_in): Fix checks for zero.
27550 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27552 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
27553 * doc/cppdiropts.texi (-I @var{dir}): Document it.
27555 2016-06-22 Richard Biener <rguenther@suse.de>
27557 * tree-vect-loop.c (vect_model_reduction_cost): Handle
27558 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
27559 REDUC_MAX_EXPR support.
27560 (vectorizable_reduction): Likewise.
27561 (vect_create_epilog_for_reduction): Likewise.
27563 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
27565 * match.pd (A / (1 << B) -> A >> B): New.
27566 * generic-match-head.c: Include optabs-tree.h.
27567 * gimple-match-head.c: Likewise.
27568 * optabs-tree.h (target_supports_op_p): New.
27569 * optabs-tree.c (target_supports_op_p): New.
27571 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27573 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
27574 $gcc_cv_ld --help output.
27575 (gcc_cv_ld_demangle): Likewise.
27576 (gcc_cv_ld_eh_frame_hdr): Likewise.
27577 (gcc_cv_ld_pie): Likewise.
27578 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
27579 (gcc_cv_ld_buildid): Likewise.
27580 (gcc_cv_ld_sysroot): Likewise.
27581 (ld_bndplt_support): Likewise.
27582 (ld_pushpopstate_support): Likewise.
27583 * configure: Regenerate.
27584 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
27586 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27589 * config/i386/sse.md (round<mode>2): Renumber match_dup and
27590 operands indexes to avoid gap between operands and match_dups.
27592 2017-06-21 Andrew Pinski <apinski@cavium.com>
27594 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
27595 Increment Arith_shift and Arith_shift_reg by 1.
27596 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
27598 * config/aarch64/aarch64.c (thunderx_tunings): Enable
27599 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
27600 (aarch64_strip_extend): Add new argument and test for it.
27601 (aarch64_cheap_mult_shift_p): New function.
27602 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
27603 add a cost if it is true.
27604 Update calls to aarch64_strip_extend.
27605 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
27607 2017-06-21 Andrew Pinski <apinski@cavium.com>
27609 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
27611 (thunderxt88): Likewise.
27612 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
27613 (thunderx_prefetch_tune): New variable.
27614 (thunderx2t99_prefetch_tune): Update for the correct values.
27615 (thunderxt88_tunings): New variable.
27616 (thunderx_tunings): Use thunderx_prefetch_tune instead of
27617 generic_prefetch_tune.
27618 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
27620 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27622 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
27623 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
27624 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
27625 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
27626 (aarch64_atomic_cas<mode>, GPI): Likewise.
27628 2017-06-21 Martin Liska <mliska@suse.cz>
27630 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
27631 statements on cold and hot labels.
27632 * predict.c (tree_estimate_probability_bb): Remove the
27633 prediction from this place.
27635 2017-06-21 Martin Liska <mliska@suse.cz>
27637 PR tree-optimization/79489
27638 * gimplify.c (maybe_add_early_return_predict_stmt): New
27640 (gimplify_return_expr): Call the function.
27641 * predict.c (tree_estimate_probability_bb): Remove handling
27643 * predict.def: Update comment about early return predictor.
27644 * gimple-predict.h (is_gimple_predict): New function.
27645 * predict.def: Change default value of early return to 66.
27646 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
27648 * passes.def: Put pass_strip_predict_hints to the beginning of
27651 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
27653 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
27654 FUNCTION_DECL declarations.
27655 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
27657 (dwaf2out_decl): Likewise.
27658 * godump.c (go_early_global_decl): Skip call to the real debug hook
27659 for FUNCTION_DECL declarations.
27660 * passes.c (rest_of_decl_compilation): Skip call to the
27661 early_global_decl debug hook for FUNCTION_DECL declarations, unless
27662 -fdump-go-spec is passed.
27664 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27666 * config/i386/i386.c (struct builtin_isa): New field pure_p.
27667 Reorder for compactness.
27668 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
27669 (def_builtin_pure, def_builtin_pure2): New functions.
27670 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
27672 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27674 * match.pd (nop_convert): New predicate.
27675 ((A +- CST1) +- CST2): Allow some NOP conversions.
27677 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27680 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
27681 with ctors/dtors if GOVD_SHARED is set.
27683 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27685 * config/aarch64/aarch64.md (movti_aarch64):
27686 Emit mov rather than orr.
27687 (movtf_aarch64): Likewise.
27688 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
27689 Emit mov rather than orr.
27691 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27693 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
27694 Swap alternatives, make integer dup more expensive.
27696 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27698 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
27699 Return true for non-tls symbols.
27701 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
27703 * config/aarch64/aarch64-cores.def (cortex-a55): New.
27704 (cortex-a75): Likewise.
27705 (cortex-a75.cortex-a55): Likewise.
27706 * config/aarch64/aarch64-tune.md: Regenerate.
27707 * doc/invoke.texi (-mtune): Document new values for -mtune.
27709 2017-06-21 Tom de Vries <tom@codesourcery.com>
27711 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
27712 stack_size feature.
27713 (Effective-Target Keywords, Other attributes): Suggest using
27714 dg-add-options stack_size feature to get stack limit in stack_size
27715 effective target documentation.
27717 2017-06-21 Julian Brown <julian@codesourcery.com>
27718 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27720 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
27721 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
27722 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
27724 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
27725 attribute type list for neon_multiply.
27726 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
27727 attribute type list for neon_multiply.
27728 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
27729 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
27730 attribute type list for neon_multiply.
27731 * config/arm/types.md (crypto_pmull): Add.
27732 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
27733 attribute type list.
27735 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
27737 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
27740 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27742 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
27743 to make sure not to dereference a NULL cost_classes_ptr pointer.
27745 2017-06-20 Carl Love <cel@us.ibm.com>
27747 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27748 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
27749 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
27750 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
27751 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
27752 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
27753 VMULOSW): New enum "unspec" values.
27754 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
27755 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
27756 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
27757 altivec_vmulosw): New patterns.
27758 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
27759 VMULOSW): Add definitions.
27761 2017-06-20 Julia Koval <julia.koval@intel.com>
27763 * config/i386/i386.c: Fix rounding expand for new pattern.
27764 * config/i386/subst.md: Fix pattern (parallel -> unspec).
27766 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27768 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
27769 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
27771 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27773 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
27776 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27778 * config/aarch64/aarch64-cores.def: Rearrange to sort by
27779 architecture, then by implementer ID.
27780 * config/aarch64/aarch64-tune.md: Regenerate.
27782 2017-06-20 Richard Biener <rguenther@suse.de>
27784 PR middle-end/81097
27785 * fold-const.c (split_tree): Fold to type before negating.
27787 2017-06-20 David Malcolm <dmalcolm@redhat.com>
27789 * diagnostic-show-locus.c
27790 (selftest::test_fixit_deletion_affecting_newline): New function.
27791 (selftest::diagnostic_show_locus_c_tests): Call it.
27793 2017-06-20 Andreas Schwab <schwab@suse.de>
27796 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
27799 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27801 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
27802 __ARM_FEATURE_COPROC according to support.
27804 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27806 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27807 Rewritten to avoid overflow for > 32-bit pointers.
27810 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
27811 by removing enum keyword.
27812 (ubsan_type_descriptor): Likewise. Formatting fix.
27815 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
27816 splitter): Require TARGET_SSE2 in the condition.
27818 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
27821 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
27822 for doing vector set of SFmode on ISA 3.0.
27823 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
27824 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
27826 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
27827 SFmode value into a V4SF variable that was extracted from another
27828 V4SF variable without converting the element to double precision
27829 and back to single precision vector format.
27830 (vsx_insert_extract_v4sf_p9_2): Likewise.
27832 2017-06-19 Jakub Jelinek <jakub@redhat.com>
27834 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
27835 in UWHI to avoid undefined overflow.
27838 * ubsan.h (enum ubsan_encode_value_phase): New.
27839 (ubsan_encode_value): Change second argument to
27840 enum ubsan_encode_value_phase with default value of
27841 UBSAN_ENCODE_VALUE_GENERIC.
27842 * ubsan.c (ubsan_encode_value): Change second argument to
27843 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
27844 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
27845 create_tmp_var_raw instead of create_tmp_var and use a
27847 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
27848 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27849 ubsan_encode_value callers.
27852 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27853 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27854 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27856 2017-06-19 Richard Biener <rguenther@suse.de>
27858 PR middle-end/81118
27859 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27860 estimates if we changed anything.
27862 2017-06-19 Richard Biener <rguenther@suse.de>
27864 PR tree-optimization/80887
27865 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27866 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27867 simplified lookups, then reset mprts_hook.
27868 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27870 (try_to_simplify): Likewise.
27872 2017-06-19 Martin Liska <mliska@suse.cz>
27875 * gimplify.c (gimplify_switch_expr):
27876 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27878 2017-06-19 Martin Liska <mliska@suse.cz>
27880 * doc/install.texi: Document that PGO runs in 4 stages.
27882 2017-06-19 Martin Liska <mliska@suse.cz>
27885 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27886 to dispatcher function name.
27887 * multiple_target.c (replace_function_decl): New function.
27888 (create_dispatcher_calls): Redirect both edges and references.
27890 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27892 * profile-count.c (profile_count::dump): Dump quality.
27893 (profile_count::differs_from_p): Update for unsigned val.
27894 * profile-count.h (profile_count_quality): New enum.
27895 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27897 2017-06-19 Richard Biener <rguenther@suse.de>
27899 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27900 struct function as arg.
27901 (estimate_numbers_of_iterations): Export overload with loop arg.
27902 (free_numbers_of_iterations_estimates_loop): Use an overload of
27903 free_numbers_of_iterations_estimates instead.
27904 * tree-cfg.c (remove_bb): Adjust.
27905 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27906 * tree-parloops.c (gen_parallel_loop): Likewise.
27907 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27909 (tree_unroll_loops_completely): Likewise.
27910 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27911 Use an overload instead and export.
27912 (estimated_loop_iterations): Adjust.
27913 (max_loop_iterations): Likewise.
27914 (likely_max_loop_iterations): Likewise.
27915 (estimate_numbers_of_iterations): Take struct function as arg
27917 (loop_exits_before_overflow): Adjust.
27918 (free_numbers_of_iterations_estimates_loop): Use an overload.
27919 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27920 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27922 2017-06-19 Richard Biener <rguenther@suse.de>
27925 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27926 RANGE_EXPR conservatively.
27928 2017-06-16 Carl Love <cel@us.ibm.com>
27930 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27931 definitions for vec_float, vec_float2, vec_floato,
27932 vec_floate built-ins.
27933 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27934 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27936 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27937 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27938 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27939 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27940 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27941 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27942 vec_floato): Add builtin defines.
27943 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27944 Update the built-in documentation file for the new built-in
27947 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27949 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27950 (mthumb): Mark as the negative of -marm.
27952 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27954 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27956 (ARM Options, -mtune): Document that this accepts the same
27957 extension options as -mcpu.
27958 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27960 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27962 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27963 permitted extensions.
27965 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27967 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27968 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27969 (armv8-m.main): Add option +nodsp.
27970 * config/arm/arm-cpu-cdata.h: Regenerated.
27972 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27974 * config/arm/t-fuchsia: New file.
27975 * config.gcc (arm*-*-fuchsia*): Use it.
27977 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27979 * config/arm/t-symbian: Rewrite for new option infrastructure.
27981 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27983 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27984 (MULTILIB_REQUIRED): Likewise.
27986 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27988 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27989 (MULTILIB_RESUE): Likewise.
27990 (MULTILIB_MATCHES): Likewise.
27991 (MULTLIB_REQUIRED): Likewise.
27993 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27995 * config/arm/t-rtems: Rewrite for new option framework.
27997 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27999 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
28000 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
28001 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
28002 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
28003 * config/arm/t-multilib: ... here.
28004 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
28005 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
28006 armv7-a and armv8*-a when A-profile libraries have not been built.
28007 * config/arm/t-rmprofile: Rewrite.
28009 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28011 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
28012 with a backslash. Remove the backslash after substituting unescaped
28014 * doc/fragments.texi (MULTILIB_REUSE): Document it.
28016 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28018 * config.gcc: (arm*-*-*): When building a-profile libraries, force
28019 the driver to pass through the default setting of -mfloat-abi.
28020 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
28022 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
28023 (all_feat_combs): New rule.
28024 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
28026 * config/arm/t-aprofile: Rewrite.
28028 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28030 * config/arm/arm.h (FPUTYPE_AUTO): Define.
28031 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
28032 fpu is not specified by the user/command-line.
28033 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
28034 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
28035 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
28036 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
28037 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
28038 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
28040 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28042 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
28043 * config/arm/t-arm-elf: Rewritten.
28045 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28047 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
28048 have some floating-point instructions.
28049 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
28050 (TARGET_MAYBE_HARD_FLOAT): New macro.
28051 * config/arm/arm-builtins.c (arm_init_builtins): Use
28052 TARGET_MAYBE_HARD_FLOAT.
28053 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
28055 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28057 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
28058 (configargs.h): Include it.
28059 (arm_print_hint_for_fpu_option): New function.
28060 (arm_parse_fpu_option): New function.
28061 (candidate_extension): New class.
28062 (arm_canon_for_multilib): New function.
28063 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
28064 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
28065 (ARCH_CANONICAL_SPECS): New macro.
28066 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
28068 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28070 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
28071 are set after handling multilib fragments. Set target_cpu_default2
28074 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28076 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
28078 (arm*-*-*): Set target_cpu_default2 to a quoted string.
28079 * config/arm/parsecpu.awk (check_cpu): Validate any extension
28081 (check_arch): Likewise.
28082 * config/arm/arm.c (arm_configure_build_target): Handle
28083 TARGET_CPU_DEFAULT being a string constant. Scan any feature
28084 options in the default.
28086 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28088 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
28089 when an option is an alias of another.
28090 * config/arm/parsecpu.awk (optalias): New parser token.
28091 (gen_comm_data): Mark non-alias options as such. Emit entries
28092 for extension aliases.
28093 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
28094 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
28095 (armv6kz, armv6zk, armv6t2): Likewise.
28096 (armv7): Make vfpv3-d16 an alias.
28097 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
28099 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
28100 Sort in canonical order.
28101 (armv8-a): Sort in canonical order.
28102 (armv8.1-a, armv8.2-a): Likewise.
28103 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
28105 (cortex-a9): Sort in canonical order.
28106 * config/arm/arm.c (selftests.h): Include it.
28107 (arm_test_cpu_arch_data): New function.
28108 (arm_run_self_tests): New function.
28109 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
28110 (targetm): Move declaration to the end of the file.
28111 * arm-cpu-cdata.h: Regenerated.
28113 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28115 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
28116 call to target_mode_check describing the type of option passed.
28117 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
28118 (arm_target_thumb_only): Use arm_parse_arch_option_name or
28119 arm_parse_cpu_option_name to match parameters against list of
28121 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
28122 arm_arch_core_flags data structure.
28123 * config/arm/arm-cpu_cdata.h: Regenerated.
28125 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28127 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
28129 (arm_print_hint_for_cpu_option): Likewise.
28130 (arm_print_hint_for_arch_option): Likewise.
28131 (arm_parse_cpu_option_name): Likewise.
28132 (arm_parse_arch_option_name): Likewise.
28133 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
28134 of entries in the all_fpus list.
28135 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
28136 (arm_parse_cpu_option_name): Declare.
28137 (arm_parse_arch_option_name): Declare.
28138 (arm_parse_option_features): Declare.
28139 (arm_intialize_isa): Declare.
28140 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
28142 (gen_comm_data): ... here. Make definitions non-static.
28143 * config/arm/arm-cpu-data.h: Regenerated.
28144 * config/arm/arm-cpu-cdata.h: Regenerated.
28146 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28148 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
28149 (cpu_arch_extension): New structure.
28150 (cpu_arch_option, arch_option, cpu_option): New structures.
28151 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
28152 architecture types.
28153 (gen_data): Generate new format data tables.
28154 * config/arm/arm.c (cpu_tune): New structure.
28155 (cpu_option, processors): Delete.
28156 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
28157 (arm_print_hint_for_cpu_option): ... this and ...
28158 (arm_print_hint_for_arch_option): ... this.
28159 (arm_parse_arch_cpu_name): Delete. Replace with ...
28160 (arm_parse_cpu_option_name): ... this and ...
28161 (arm_parse_arch_option_name): ... this.
28162 (arm_unrecognized_feature): Change type of target parameter to
28164 (arm_parse_arch_cpu_features): Delete. Replace with ...
28165 (arm_parse_option_features): ... this.
28166 (arm_configure_build_target): Rework to use new configuration data
28168 (arm_print_tune_info): Rework for new configuration data tables.
28169 * config/arm/arm-cpu-data.h: Regenerated.
28170 * config/arm/arm-cpu.h: Regenerated.
28172 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28174 * Makefile.in (OBJS): Move sbitmap.o from here ...
28175 (OBJS-libcommon): ... to here.
28177 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28179 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
28180 (ISA_ALL_CRYPTO): New macro.
28181 (ISA_ALL_SIMD): New macro
28182 (ISA_ALL_FP): New macro.
28183 * config/arm/arm.c (fpu_bitlist): Update initializer.
28184 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
28186 (arm9e): Add fpu. Add option for nofp
28187 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
28188 (arm926ej-s, arm1026ej-s): Likewise.
28189 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
28190 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
28191 neon-fp16, neon-vfpv4, nofp and nosimd.
28192 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
28193 (cortex-a8): Add fpu. Add option for nofp.
28194 (cortex-a9): Add fpu. Add options for nosimd and nofp.
28195 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
28196 (cortex-r4f): Add fpu.
28197 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
28198 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
28200 (cortex-r8): Likewise.
28201 (cortex-m4): Add fpu. Add option for nofp.
28202 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
28203 (cortex-a17.cortex-a7): Likewise.
28204 (cortex-a32): Add fpu. Add options for crypto and nofp.
28205 (cortex-a35, cortex-a53): Likewise.
28206 (cortex-a57): Add fpu. Add option for crypto.
28207 (cortex-a72, cortex-a73): Likewise.
28208 (exynos-m1): Likewise.
28209 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
28210 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
28211 (cortex-m33): Add fpu. Add option for nofp.
28212 * config/arm/arm-cpu-cdata.h: Regenerated
28213 * config/arm/arm-cpu-data.h: Regenerated.
28215 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28217 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
28218 (armv5te, armv5tej): Likewise.
28219 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
28220 (armv7): Add options fp and vfpv3-d16.
28221 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
28222 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
28224 (armv7ve): Likewise.
28225 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
28226 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
28227 (armv8-a): Add nocrypto option.
28228 (armv8.1-a, armv8.2-a): Likewise.
28229 (armv8-m.main): add options fp, fp.dp and nofp.
28231 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28233 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
28235 (armv8-a+crc): Delete.
28236 (armv8.1-a): Add options simd, crypto and nofp.
28237 (armv8.2-a): Add options fp16, simd, crypto and nofp.
28238 (armv8.2-a+fp16): Delete.
28239 (armv8-m.main): Add option dsp.
28240 (armv8-m.main+dsp): Delete.
28241 (cortex-a8): Add fpu. Add nofp option.
28242 (cortex-a9): Add fpu. Add nofp and nosimd options.
28243 * config/arm/parsecpu.awk (gen_data): Generate option tables and
28244 link to main cpu and architecture data structures.
28245 (gen_comm_data): Only put isa attributes from the main architecture
28247 (option): New statement for architecture and CPU entries.
28248 * arm.c (struct cpu_option): New structure.
28249 (struct processors): Add entry for options.
28250 (arm_unrecognized_feature): New function.
28251 (arm_parse_arch_cpu_name): Ignore any characters after the first
28253 (arm_parse_arch_cpu_feature): New function.
28254 (arm_configure_build_target): Separate out any CPU and architecture
28255 features and parse separately. Don't error out if -mfpu=auto is
28256 used with only an architecture string.
28257 (arm_print_asm_arch_directives): New function.
28258 (arm_file_start): Call it.
28259 * config/arm/arm-cpu-cdata.h: Regenerated.
28260 * config/arm/arm-cpu-data.h: Likewise.
28261 * config/arm/arm-tables.opt: Likewise.
28263 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28265 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
28266 assembler when it is not -mfpu=auto.
28268 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28270 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
28271 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
28272 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
28273 (ASM_CPU_SPEC): Rewrite.
28274 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
28275 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
28276 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
28277 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
28278 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
28279 copied string is NUL-terminated. Also strip any characters prefixed
28281 (arm_rewrite_selected_arch): New function.
28282 (arm_rewrite_march): New function.
28284 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28286 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
28287 (x_arm_cpu_string, x_arm_tune_string): Likewise.
28288 (march, mcpu, mtune): Convert to string-based options.
28289 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
28290 (arm_parse_arch_cpu_name): New function.
28291 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
28292 identify selected architecture or CPU.
28293 (arm_option_save): New function.
28294 (TARGET_OPTION_SAVE): Redefine.
28295 (arm_option_restore): Restore string options.
28296 (arm_option_print): Print string options.
28298 2017-06-16 Martin Sebor <msebor@redhat.com>
28300 PR tree-optimization/80933
28301 PR tree-optimization/80934
28302 * builtins.c (fold_builtin_3): Do not handle bcmp here.
28303 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
28304 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
28305 (gimple_fold_builtin): Call them.
28307 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28309 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
28310 as unlikely; update profile.
28312 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28314 * predict.c (force_edge_cold): Handle declaring edges impossible
28317 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28319 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
28321 (try_unroll_loop_completely): Fix reporting.
28323 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28325 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
28327 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
28330 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
28331 if given a non-constant argument for an intrinsic which requires a
28334 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28336 * profile.c (compare_freqs): New function.
28337 (branch_prob): Sort edge list.
28338 (find_spanning_tree): Assume that the list is priority sorted.
28340 2017-06-16 Richard Biener <rguenther@suse.de>
28342 PR tree-optimization/81090
28343 * passes.def (pass_record_bounds): Remove.
28344 * tree-pass.h (make_pass_record_bounds): Likewise.
28345 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
28346 make_pass_record_bounds): Likewise.
28347 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
28348 not free niter estimates at the beginning but at the end.
28349 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
28351 2017-06-16 Richard Biener <rguenther@suse.de>
28353 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
28354 initializer to workaround ICE in host GCC 4.8.
28356 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28358 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
28360 (clone_inlined_nodes): Update.
28362 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28364 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
28365 prefetch settings, and enable prefetching by default at -O3.
28367 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28369 * config/aarch64/aarch64.c (aarch64_override_options_internal):
28370 Set flag_prefetch_loop_arrays according to tuning data.
28372 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28374 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
28375 New tune structure.
28376 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
28377 [Unrelated to main purpose of the patch] Place the pointer field last
28378 to enable type checking errors when tune structure are wrongly merged.
28379 * config/aarch64/aarch64.c (generic_prefetch_tune,)
28380 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
28381 (thunderx2t99_prefetch_tune): New tune constants.
28382 (tune_params *_tunings): Update all tunings (no functional change).
28383 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
28384 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
28385 from tunings structures.
28387 2017-06-16 Jakub Jelinek <jakub@redhat.com>
28390 * ubsan.c (instrument_null): Add T argument, use it instead
28391 of computing it based on IS_LHS.
28392 (instrument_object_size): Likewise.
28393 (pass_ubsan::execute): Adjust instrument_null and
28394 instrument_object_size callers to pass gimple_get_lhs or
28395 gimple_assign_rhs1 result to it. Use instrument_null instead of
28396 calling get_base_address and instrument_mem_ref. Handle
28397 aggregate call arguments for object-size sanitization.
28399 2017-06-16 Yury Gribov <tetra2005@gmail.com>
28401 PR tree-optimization/81089
28402 * tree-vrp.c (is_masked_range_test): Validate operands of
28405 2017-06-15 Martin Sebor <msebor@redhat.com>
28408 * dumpfile.c (dump_register): Avoid calling memset to initialize
28409 a class with a default ctor.
28410 * gcc.c (struct compiler): Remove const qualification.
28411 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
28412 * hash-table.h: Ditto.
28413 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
28415 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
28416 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
28418 * params.h (struct param_info): Make struct members non-const.
28419 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
28420 with default initialization.
28421 * vec.h (vec_copy_construct, vec_default_construct): New helper
28423 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
28424 with vec_copy_construct.
28425 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
28426 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
28427 * doc/invoke.texi (-Wclass-memaccess): Document.
28429 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28431 * emit-rtl.h (is_leaf): Update comment about local
28432 register allocator.
28434 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28437 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
28438 for a variable to have a section before checking if the section has a
28440 Set section to.persistent if persistent attribute is set.
28441 Warn if .persistent attribute is used on an automatic variable.
28443 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
28445 PR rtl-optimization/80474
28446 * reorg.c (update_block): Do not ignore instructions in a delay slot.
28448 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
28450 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
28453 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
28455 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
28456 (casesi): Emit bounds checking as RTL.
28457 (casesi_internal_mips16_<mode>): Remove bounds checking.
28459 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
28461 * config/xtensa/xtensa.c (xtensa_option_override): Append
28462 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
28463 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
28464 xtensa_doloop_hooks): Define unconditionally.
28465 (xtensa_reorg_loops): Only call reorg_loops in the presence of
28467 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
28468 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
28469 for it in xtensa_option_override.
28470 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
28471 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
28473 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
28475 * doc/cppopts.texi: Document '-' special value to -MF.
28477 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
28479 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
28480 (cortex_a53_fconst): Likewise.
28481 (cortex_a53_fpmul): Likewise.
28482 (cortex_a53_f_load_64): Likewise.
28483 (cortex_a53_f_load_many): Likewise.
28484 (cortex_a53_advsimd_alu): Likewise.
28485 (cortex_a53_advsimd_alu_q): Likewise.
28486 (cortex_a53_advsimd_mul): Likewise.
28487 (cortex_a53_advsimd_mul_q): Likewise.
28488 (fpmac bypass): Add new bypass for fpmac-fpmac case.
28489 Add missing fmul, r2f_cvt and fconst cases.
28491 2017-06-14 Richard Biener <rguenther@suse.de>
28493 PR middle-end/81088
28494 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
28496 (fold_binary_loc): When associating do not treat pre-existing
28497 TREE_OVERFLOW on literal constants as a reason to allow
28498 TREE_OVERFLOW on associated literal constants.
28500 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28502 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
28503 (MASK_FEATURES): New macro.
28504 * config/sparc/sparc.c (sparc_option_override): Remove the special
28505 handling of -mfpu and generalize it to all MASK_FEATURES switches.
28507 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28509 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
28510 a division of 0 if non-call exceptions are enabled.
28512 2017-06-14 Andrew Pinski <apinski@cavium.com>
28513 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28516 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
28517 Improve vector initialization code gen for only variable case.
28519 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28521 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
28523 2017-06-14 Richard Biener <rguenther@suse.de>
28525 PR tree-optimization/81083
28526 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
28529 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28531 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
28532 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
28533 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
28534 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
28535 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
28536 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
28538 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28540 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
28541 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
28543 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28545 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
28547 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28549 * config/rs6000/t-rtems: Don't handle SPE.
28551 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28553 * config/rs6000/t-linux: Don't handle SPE.
28555 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28557 * config/rs6000/eabispe.h: Delete file.
28559 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28561 * config/rs6000/t-spe: Delete file.
28563 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28565 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
28566 (rs6000_legitimate_offset_address_p): Return false for anything in
28567 V2SImode or V2SFmode.
28569 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28571 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
28572 except V2SF and V2SI. Rearrange the vector modes, and add comments.
28573 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
28575 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
28576 (rs6000_legitimate_offset_address_p): Ditto.
28577 (rs6000_emit_move): Ditto.
28578 (rs6000_init_builtins): Remove V4HI_type_node.
28580 2017-06-13 Martin Liska <mliska@suse.cz>
28583 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
28584 (gate_asan): Likewise.
28585 * asan.h (asan_no_sanitize_address_p): Remove the function.
28586 (sanitize_flags_p): New function.
28587 * builtins.def: Fix coding style.
28588 * common.opt: Use renamed enum value.
28589 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
28590 * doc/extend.texi: Document no_sanitize attribute.
28591 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
28592 to SANITIZE_UNDEFINED_NONDEFAULT.
28593 * gcc.c (sanitize_spec_function): Use the renamed enum value.
28594 * gimple-fold.c (optimize_atomic_compare_exchange_p):
28595 Use sanitize_flags_p.
28596 * gimplify.c (gimplify_function_tree): Likewise.
28597 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
28598 * opts.c (parse_no_sanitize_attribute): New function.
28599 (common_handle_option): Use renamed enum value.
28600 * opts.h (parse_no_sanitize_attribute): Declare.
28601 * tree.c (sanitize_flags_p): New function.
28602 * tree.h: Declared here.
28603 * tsan.c: Use sanitize_flags_p.
28604 * ubsan.c (ubsan_expand_null_ifn): Likewise.
28605 (instrument_mem_ref): Likewise.
28606 (instrument_bool_enum_load): Likewise.
28607 (do_ubsan_in_current_function): Remove the function.
28608 (pass_ubsan::execute): Use sanitize_flags_p.
28609 * ubsan.h: Remove do_ubsan_in_current_function
28610 * tree-cfg.c (print_no_sanitize_attr_value): New function.
28611 (dump_function_to_file): Use it here.
28613 2017-06-13 Martin Jambor <mjambor@suse.cz>
28615 PR tree-optimization/80803
28616 PR tree-optimization/81063
28617 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
28618 (propagate_subaccesses_across_link): Enqueue subtree whenever
28619 necessary instead of relying on the caller.
28621 2017-06-13 Martin Jambor <mjambor@suse.cz>
28623 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
28624 that have a first_link.
28625 (sort_and_splice_var_accesses): Do not check first_link before
28627 (subtree_mark_written_and_enqueue): Likewise.
28628 (propagate_all_subaccesses): Likewise and do not stop at first
28629 parent with a first_link.
28631 2017-06-13 Martin Jambor <mjambor@suse.cz>
28633 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
28636 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28638 * match.pd: New pattern.
28640 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28642 * tree-vrp.c (is_masked_range_test): New function.
28643 (register_edge_assert_for): Determine ranges for
28646 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28648 PR tree-optimization/67328
28649 * fold-const.c (maskable_range_p): New function.
28650 (build_range_check): Generate bittests if possible.
28652 2017-06-13 Martin Liska <mliska@suse.cz>
28654 * gimple-pretty-print.c (dump_probability): Add new argument.
28655 (dump_edge_probability): Dump both probability and count.
28656 (dump_gimple_label): Likewise.
28657 (dump_gimple_bb_header): Likewise.
28659 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
28662 * config/avr/avr-devices.c: Fix indentation.
28663 * config/avr/gen-avr-mmcu-specs.c: Dito.
28665 2017-06-13 Richard Biener <rguenther@suse.de>
28667 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
28668 instead get vector type from stmt_info.
28669 (vectorizable_reduction): Adjust. Remove dead code.
28671 2017-06-13 Richard Biener <rguenther@suse.de>
28673 PR middle-end/81065
28674 * fold-const.c (extract_muldiv_1): Remove bogus distribution
28675 case of C * (x * C2 + C3).
28676 (fold_addr_of_array_ref_difference): Properly fold index difference.
28678 2017-06-12 David S. Miller <davem@davemloft.net>
28681 * config/sparc/sparc.md (return expander): Emit frame blockage if
28682 function uses alloca.
28684 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
28686 * combine.c (make_field_assignment): Check len rather than the mode
28687 precision when calling force_to_mode.
28689 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
28691 Support multilibs and devices that see flash in RAM address range.
28694 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
28695 (avr_mcu_t) <flash_pm_offset>: New field.
28696 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
28697 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
28698 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
28699 (AVR_TINY_PM_OFFSET): Remove macro.
28700 * config/avr/avr.opt (-mshort-calls): New option.
28701 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
28702 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
28703 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
28704 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
28705 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
28706 instead of avr_arch->have_jmp_call.
28707 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
28708 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
28709 avr_arch->flash_pm_offset to define.
28710 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
28711 new field flash_pm_offset. Add entry for avrxmega3.
28712 (avr_texinfo): Add entry for avrxmega3.
28713 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
28714 attiny212, attiny214,
28715 attiny412, attiny414, attiny416, attiny417,
28716 attiny814, attiny816, attiny817,
28717 attiny1614, attiny1616, attiny1617,
28718 attiny3214, attiny3216, attiny3217.
28719 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
28720 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
28721 (avr_print_operand_address) [AVR_TINY]: Same.
28722 (avr_asm_init_sections) <readonly_data_section>: Only patch
28723 callback if avr_arch->flash_pm_offset = 0.
28724 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
28725 for rodata if avr_arch->flash_pm_offset != 0.
28726 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
28727 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
28728 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
28729 (m_options): Append opt_rcall.
28730 (m_dirnames): Append dir_rcall.
28731 * config/avr/t-multilib: Regenerate.
28733 * configure.ac [target=avr]: Check whether avrxmega3 default
28734 linker description file works as needed.
28735 * configure: Regenerate.
28736 * doc/avr-mmcu.texi: Regenerate.
28737 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
28738 <__AVR_ARCH__>: Document avrxmega3 and 103.
28739 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
28740 <__AVR_SHORT_CALLS__>: Document it.
28741 <__AVR_PM_BASE_ADDRESS__>: Document it.
28742 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
28743 (AVR Variable Attributes) <progmem>: Document this is
28744 not needed for avrxmega3.
28745 (AVR Named Address Spaces) <__flash>: Dito.
28747 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
28749 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
28751 2017-06-12 Doug Rupp <rupp@adacore.com>
28753 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
28754 Append vxworks-stdint.h to the tm_file list.
28755 * config/vxworks-stdint.h: New file.
28757 2017-06-12 Martin Liska <mliska@suse.cz>
28759 PR tree-optimization/81041
28760 * tree-profile.c (gimple_gen_ic_func_profiler):
28761 Create an extra BB in profile-generate
28762 (gimple_gen_time_profiler): Likewise.
28764 2017-06-12 Jakub Jelinek <jakub@redhat.com>
28766 PR tree-optimization/81003
28767 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
28768 (update_range_test): Use it instead of force_gimple_operand_gsi.
28770 2017-06-12 Richard Biener <rguenther@suse.de>
28772 PR tree-optimization/81053
28773 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
28774 with backedge value not defined in loop. Simplify def stmt
28777 2017-06-11 Tom de Vries <tom@codesourcery.com>
28780 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
28782 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
28783 nvptx_cannot_force_const_mem.
28785 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28787 * opts.c (finish_options): Move test for flag_split_stack after
28788 it has been initialized.
28790 2017-06-11 Jason Merrill <jason@redhat.com>
28792 * tree.h (id_equal): New.
28793 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
28794 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
28795 instead of strcmp of IDENTIFIER_POINTER.
28797 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28799 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
28800 (mark_all_inlined_calls_cdtor): Fix formating.
28801 (inline_transform): Rescale profile before inlining.
28803 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28805 * cgraph.h (cgraph_edge::clone): Update prototype.
28806 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
28807 (cgraph_node::create_clone): Update.
28808 (cgraph_node::create_version_clone): Update.
28809 * tree-inline.c (copy_bb): Update.
28810 (expand_call_inline): Update.
28812 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28814 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
28815 factored out from ...
28816 (rs6000_emit_prologue): ... here.
28818 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28820 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
28821 factored out from ...
28822 (rs6000_emit_prologue): ... here.
28824 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28826 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
28828 (handle_missing_profiles): Fix computation of tp_first_run.
28829 (counts_to_freqs): Do not touch freqs when count is 0.
28831 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28833 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
28836 2017-06-10 Tom de Vries <tom@codesourcery.com>
28838 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28839 attributes): Document signal effective target.
28841 2017-06-10 Tom de Vries <tom@codesourcery.com>
28843 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28844 Document effective target stack_size.
28846 2017-06-09 David Malcolm <dmalcolm@redhat.com>
28848 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28849 to the edit_context if they can be auto-applied.
28851 2017-06-9 Ian Lance Taylor <iant@golang.org>
28853 * opts.c (finish_options): If -fsplit-stack, disable implicit
28854 -forder-blocks-and-partition.
28855 * doc/invoke.texi (Optimize Options): Document that when using
28856 -fsplit-stack -forder-blocks-and-partition is not implicitly
28859 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28861 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28862 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28863 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28864 * builtins.def (abort, trap, unreachable): Declare cold.
28865 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28866 * tree-core.h (ECF_COLD): New.
28867 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28868 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28870 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28872 * predict.c (unlikely_executed_stmt_p): Cleanup.
28874 2017-06-09 Richard Biener <rguenther@suse.de>
28876 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28877 model if the ref is always written to.
28879 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28881 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28883 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28885 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28888 2017-06-09 Tom de Vries <tom@codesourcery.com>
28891 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28892 "target cannot support label values" when encountering LABEL_REF.
28894 2017-06-09 Martin Liska <mliska@suse.cz>
28896 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28897 (gimple_gen_ic_func_profiler): Emit direct comparison
28898 of __gcov_indirect_call_callee with NULL.
28899 (gimple_gen_time_profiler): Change probability from
28900 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28902 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28904 * profile.c (edge_gcov_counts): Turn to pointer.
28905 (compute_branch_probabilities, compute_branch_probabilities): Update.
28906 (branch_prob): Do not clear edge_gcov_count.
28907 * profile.h (edge_gcov_counts): Turn to pointer.
28908 (edge_gcov_count): Update.
28910 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28912 * gimple.h (gimple_check_failed): Mark cold.
28914 2017-06-09 Richard Biener <rguenther@suse.de>
28916 PR tree-optimization/66623
28917 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28918 refactor check_reduction into two parts, properly computing
28919 whether we have to check reduction validity for outer loop
28922 2017-06-09 Richard Biener <rguenther@suse.de>
28924 PR tree-optimization/79483
28925 * graphite-scop-detection.c (order): New global.
28926 (get_order): Compute bb to order mapping that satisfies code
28927 generation constraints.
28928 (cmp_pbbs): New helper.
28929 (build_scops): Start domwalk at entry block, sort generated
28932 2017-06-09 Richard Biener <rguenther@suse.de>
28934 PR middle-end/81007
28935 * ipa-polymorphic-call.c
28936 (ipa_polymorphic_call_context::restrict_to_inner_class):
28937 Skip FIELD_DECLs with error_mark_node type.
28938 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28941 2017-06-09 Martin Liska <mliska@suse.cz>
28943 * predict.c (struct branch_predictor): New struct.
28944 (test_prediction_value_range): New test.
28945 (predict_c_tests): New function.
28946 * selftest-run-tests.c (selftest::run_tests): Run the function.
28947 * selftest.h: Declare new tests.
28949 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28952 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28953 gen_add3_insn did not fail.
28954 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28955 r0, construct that number in a temporary reg and add that reg to r0.
28956 If asked to put the result in r0 as well, fail.
28958 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28960 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28961 for early expansion of vec_eqv.
28963 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28965 PR middle-end/81005
28966 * ubsan.c (instrument_null): Avoid pointless code temporary.
28967 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28969 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28973 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28976 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28979 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28981 * predict.c (unlikely_executed_edge_p): Move ahead.
28982 (probably_never_executed_edge_p): Use it.
28984 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28986 PR middle-end/79988
28987 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28988 gimple_call_builtin_p call.
28990 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28992 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28993 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28994 rtl_check_failed_type2, rtl_check_failed_code1,
28995 rtl_check_failed_code2, rtl_check_failed_code_mode,
28996 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28997 rtvec_check_failed_bounds, rtl_check_failed_flag,
28998 _fatal_insn_not_found, _fatal_insn): Likewise.
28999 * tree.h (tree_contains_struct_check_failed,
29000 tree_check_failed, tree_not_check_failed,
29001 tree_class_check_failed, tree_range_check_failed,
29002 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
29003 tree_vec_elt_check_failed, phi_node_elt_check_failed,
29004 tree_operand_check_failed, omp_clause_check_failed,
29005 omp_clause_operand_check_failed, omp_clause_range_check_failed):
29008 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
29010 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
29011 flag_branch_probabilities.
29012 * ipa-inline.c (edge_badness): Likewise.
29013 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
29014 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
29015 * predict.c (maybe_hot_frequency_p): Likewise.
29016 (probably_never_executed): Likewise.
29017 * sched-ebb.c (schedule_ebbs): Likewise.
29018 * sched-rgn.c (find_single_block_region): Likewise.
29019 * tracer.c (tail_duplicate): Likewise.
29021 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
29023 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
29024 longer requires x_flag_profile_use.
29026 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
29028 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
29029 instead of flag_reorder_blocks_and_partition.
29030 * dbxout.c (dbxout_function_end): Likewise.
29031 * dwarf2out.c (gen_subprogram_die): Likewise.
29032 * haifa-sched.c (sched_create_recovery_edges): Likewise.
29033 * hw-doloop.c (reorg_loops): Likewise.
29034 * varasm.c (assemble_start_function,
29035 assemble_end_function): Likewise.
29036 (decide_function_section): Do not check for
29037 flag_reorder_blocks_and_partition.
29039 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
29041 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
29043 (chkp_get_hard_register_fake_addr_expr): Ditto.
29044 (chkp_build_addr_expr): Add check for hard reg case.
29045 (chkp_parse_array_and_component_ref): Ditto.
29046 (chkp_find_bounds_1): Ditto.
29047 (chkp_process_stmt): Don't generate bounds store for
29050 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
29052 * predict.c (maybe_hot_bb_p): Do not check profile status.
29053 (maybe_hot_edge_p): Likewise.
29054 (probably_never_executed): Check for zero counts even if profile
29056 (unlikely_executed_edge_p): New function.
29057 (unlikely_executed_stmt_p): New function.
29058 (unlikely_executed_bb_p): New function.
29059 (set_even_probabilities): Use unlikely predicates.
29060 (combine_predictions_for_bb): Likewise.
29061 (predict_paths_for_bb): Likewise.
29062 (predict_paths_leading_to_edge): Likewise.
29063 (determine_unlikely_bbs): New function.
29064 (estimate_bb_frequencies): Use it.
29065 (compute_function_frequency): Use zero counts even if profile is
29067 * profile-count.h: Fix typo.
29069 2017-08-08 Julia Koval <julia.koval@intel.com>
29071 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
29072 _mm512_mask_cvtsepi16_storeu_epi8,
29073 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
29074 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
29075 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
29076 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
29077 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
29078 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
29079 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
29080 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
29081 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
29082 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
29083 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
29084 __builtin_ia32_pmovuswb256mem_mask,
29085 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
29086 __builtin_ia32_pmovwb512mem_mask): New builtins.
29088 2017-08-08 Julia Koval <julia.koval@intel.com>
29090 PR target/73350,80862
29091 * config/i386/subst.md (round): Fix round pattern.
29092 * config/i386/i386.c (ix86_erase_embedded_rounding):
29093 Fix erasing rounding for the fixed pattern.
29095 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
29097 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
29099 2017-06-08 Martin Liska <mliska@suse.cz>
29101 PR gcov-profile/80911
29102 * gcov.c (block_info::block_info): New constructor.
29104 2017-06-07 Carl Love <cel@us.ibm.com>
29106 * config/rs6000/rs6000-c: The return type of the following
29107 built-in functions was implemented as int not long long. Fix sign
29108 of return value for the unsigned version of vec_mulo and vec_mule.
29109 vector unsigned long long vec_bperm (vector unsigned long long,
29110 vector unsigned char)
29111 vector signed long long vec_mule (vector signed int,
29113 vector unsigned long long vec_mule (vector unsigned int,
29114 vector unsigned int)
29115 vector signed long long vec_mulo (vector signed int,
29117 vector unsigned long long vec_mulo (vector unsigned int,
29118 vector unsigned int)
29119 * doc/extend.texi: Fix the documentation for the built-in
29122 2017-06-07 Carl Love <cel@us.ibm.com>
29125 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
29128 2017-06-07 Carl Love <cel@us.ibm.com>
29130 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
29131 support, Generate doublehv for signed int/float for BE case only.
29133 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
29135 * doc/invoke.texi (mcx16): Rewrite.
29137 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29139 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
29140 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
29141 *mov<mode>_softfloat, and an anonymous splitter): Use
29142 nonimmediate_operand instead of rs6000_nonimmediate_operand.
29144 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29146 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
29148 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
29149 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
29150 (rs6000_debug_reg_global): Adjust.
29151 (rs6000_init_hard_regno_mode_ok): Adjust.
29152 (rs6000_dbx_register_number): Adjust.
29153 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
29154 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
29155 Remove SPE_ACC and SPEFSCR.
29156 (REG_ALLOC_ORDER): Ditto.
29157 (FRAME_POINTER_REGNUM): Change to 111.
29158 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
29159 (REG_CLASS_NAMES): Ditto.
29160 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
29161 (REGISTER_NAMES): Ditto.
29162 (ADDITIONAL_REG_NAMES): Ditto.
29163 (rs6000_reg_names): Ditto.
29164 * config/rs6000/rs6000.md: Renumber some register number
29167 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29169 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
29171 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
29172 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
29174 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
29175 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
29176 Delete the SPE high registers.
29177 (REG_ALLOC_ORDER): Ditto.
29178 (enum reg_class): Remove SPE_HIGH_REGS.
29179 (REG_CLASS_NAMES): Ditto.
29180 (REG_CLASS_CONTENTS): Delete the SPE high registers.
29181 (REGISTER_NAMES): Ditto.
29182 (rs6000_reg_names): Ditto.
29183 * doc/tm.texi.in: Remove SPE as example.
29184 * doc/tm.texi: Regenerate.
29186 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29188 * config/rs6000/8540.md (ppc8540_brinc): Delete.
29189 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
29190 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
29191 * config/rs6000/rs6000.md (type): Remove "brinc".
29193 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29195 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
29196 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
29197 * config/rs6000/linuxspe.h: Delete file.
29198 * config/rs6000/rs6000.md: Don't include spe.md.
29199 * config/rs6000/spe.h: Delete file.
29200 * config/rs6000/spe.md: Delete file.
29201 * config/rs6000/t-rs6000: Remove spe.md.
29203 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29205 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
29206 (reg_or_none500mem_operand): Delete.
29207 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
29208 instead of reg_or_none500mem_operand.
29210 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29212 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29213 handling of SPE flags.
29214 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
29216 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29218 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
29220 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
29221 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
29222 paired_divv2sf3): Similar.
29223 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
29224 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
29225 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
29227 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
29228 Rename the paired_* instruction patterns.
29229 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
29231 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
29232 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
29233 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
29234 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
29235 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
29236 PAIRED_VECTOR_MODE.
29237 (struct machine_function): Delete field spe_insn_chain_scanned_p.
29238 (spe_func_has_64bit_regs_p): Delete.
29239 (spe_expand_predicate_builtin): Delete.
29240 (spe_expand_evsel_builtin): Delete.
29241 (TARGET_DWARF_REGISTER_SPAN): Do not define.
29242 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
29243 (invalid_e500_subreg): Delete.
29244 (rs6000_legitimize_address): Always force_reg op2 as well, for
29245 paired single memory accesses.
29246 (rs6000_member_type_forces_blk): Delete.
29247 (rs6000_spe_function_arg): Delete.
29248 (rs6000_expand_unop_builtin): Delete SPE handling.
29249 (rs6000_expand_binop_builtin): Ditto.
29250 (spe_expand_stv_builtin): Delete.
29251 (bdesc_2arg_spe): Delete.
29252 (spe_expand_builtin): Delete.
29253 (spe_expand_predicate_builtin): Delete.
29254 (spe_expand_evsel_builtin): Delete.
29255 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
29256 (spe_init_builtins): Delete.
29257 (spe_func_has_64bit_regs_p): Delete.
29258 (savres_routine_name): Delete "info" parameter. Adjust callers.
29259 (rs6000_emit_stack_reset): Ditto.
29260 (rs6000_dwarf_register_span): Delete.
29261 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
29262 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
29263 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
29265 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
29267 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
29268 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
29269 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
29270 mulv2sf3, divv2sf3): Delete expanders.
29272 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29274 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
29276 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29278 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
29279 * config/rs6000/rs6000.c: Ditto.
29281 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29283 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
29284 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
29285 comparison_operator.
29287 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29289 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
29290 * config/rs6000/rs6000.opt: Ditto.
29291 * config/rs6000/t-rtems: Ditto.
29293 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29295 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
29296 TARGET_E500_SINGLE by 0, simplify.
29297 * config/rs6000/rs6000.c: Ditto.
29298 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
29299 (spe_build_register_parallel): Delete.
29300 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
29301 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
29302 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
29303 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
29304 (E500_CONVERT): Delete.
29305 * config/rs6000/spe.md: Remove many patterns and all define_constants.
29307 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29309 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
29310 * config/rs6000/dfp.md: Ditto.
29311 (negdd2, *negdd2_fpr): Merge.
29312 (absdd2, *absdd2_fpr): Merge.
29313 (negtd2, *negtd2_fpr): Merge.
29314 (abstd2, *abstd2_fpr): Merge.
29315 * config/rs6000/e500.h: Delete file.
29316 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
29317 TARGET_FPRS by 1 and simplify.
29318 * config/rs6000/rs6000-c.c: Ditto.
29319 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
29320 TARGET_DF_SPE by 0.
29321 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
29323 * config/rs6000/rs6000.md: Ditto.
29324 (floatdidf2, *floatdidf2_fpr): Merge.
29325 (move_from_CR_gt_bit): Delete.
29326 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
29327 (E500_CR_IOR_COMPARE): Delete.
29328 (All patterns that require !TARGET_FPRS): Delete.
29329 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
29331 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29333 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
29335 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29337 * graphds.c (add_edge): Intitialize edge's attached data.
29338 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
29339 pointer parameter. Call pointed function on each edge during
29340 graph traversing. Skip traversing the edge when the function
29342 (graphds_dfs, graphds_scc): Ditto.
29343 (for_each_edge): New parameter. Pass the new parameter to callback
29345 * graphds.h (skip_edge_callback): New function pointer type.
29346 (graphds_dfs, graphds_scc): New function pointer parameter.
29347 (graphds_edge_callback, for_each_edge): New parameter.
29349 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29351 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
29352 out code checking if runtime alias check is possible to below ...
29353 Call the new function.
29354 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
29355 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
29357 2017-06-07 Marek Polacek <polacek@redhat.com>
29360 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
29361 TYPE_OVERFLOW_WRAPS check.
29363 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29365 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
29366 if versioning is required.
29367 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
29368 peeling with the check for versioning.
29370 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29372 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
29373 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
29374 Set true to new parameter if new ssa variable is defined.
29375 (vect_gen_vector_loop_niters): Refactor. Set range information
29376 for the new vector loop bound variable.
29377 (vect_do_peeling): Ditto.
29379 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29381 * tree-affine.c (ssa.h): Include header file.
29382 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
29383 has wrapping overflow behavior.
29385 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29387 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
29389 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29391 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
29392 (tree_to_aff_combination): ... here.
29394 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29396 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
29397 reg_pressure model function.
29398 (ivopts_global_cost_for_size): Delete.
29399 (determine_set_costs, iv_ca_recount_cost): Call new model function
29400 ivopts_estimate_reg_pressure.
29402 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29404 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
29405 expensive than udiv. Remove floating point cases from mod.
29407 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29409 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
29410 Increase idiv cost.
29412 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29414 * config/aarch64/aarch64.md
29415 (copysignsf3): Fix mask generation.
29417 2017-06-07 Jakub Jelinek <jakub@redhat.com>
29419 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
29421 (class dump_manager): Add register_dumps method.
29422 * dumpfile.c: Include langhooks.h.
29423 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
29424 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
29425 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
29426 (dump_manager::dump_register): Start with 512 entries instead of 32.
29427 (dump_manager::register_dumps): New method.
29428 * toplev.c (general_init): Instead of invoking register_dumps
29429 langhook, invoke register_dumps method on the dump manager.
29430 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
29433 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
29435 * doc/md.texi: Clarify the restrictions on a define_insn condition.
29436 Say that # requires an associated define_split to exist, and that
29437 the define_split must be suitable for use after register allocation.
29439 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29441 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
29442 (compute_outgoing_frequencies): Also initialize zero counts.
29443 (find_many_sub_basic_blocks): Do not produce uninitialized profile
29444 around loops; preserve more of profile when nothing changes.
29446 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
29448 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
29450 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
29451 * config/arm/arm-cpu-cdata.h: Regenerate.
29452 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
29453 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
29454 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
29455 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
29456 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
29458 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
29460 * config/arm/t-rmprofile: Likewise.
29461 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
29463 2017-06-06 David S. Miller <davem@davemloft.net>
29466 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
29467 blockage if function uses alloca.
29469 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29471 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
29472 New "uid" fields to hold pretty-print IDs of group and ref.
29473 Memory references are now identified as <group_id>:<ref_id>
29474 instead of using [random] addresses.
29475 (dump_mem_details): Simplify, no functional change.
29476 (dump_mem_ref): Simplify and make output more concise.
29477 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
29478 (find_or_create_group): Initialize group uid.
29479 (record_ref): Initialize ref uid. Improve debug output.
29480 (prune_group_by_reuse, should_issue_prefetch_p,)
29481 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
29482 (mark_nontemporal_store, determine_loop_nest_reuse):
29483 Improve debug output.
29485 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29487 * dbgcnt.def (prefetch): New debug counter.
29488 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
29489 (schedule_prefetches): Stop issueing prefetches if debug counter
29492 2017-06-06 Tom de Vries <tom@codesourcery.com>
29494 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
29495 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
29497 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29499 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
29500 Use aarch64_reg_or_zero predicate for operand 4.
29501 (aarch64_compare_and_swap<mode> define_insn_and_split):
29502 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
29503 (aarch64_store_exclusive<mode>): Likewise for operand 2.
29505 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
29507 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
29508 (arm_compute_save_core_reg_mask): This.
29509 (thumb1_compute_save_reg_mask): Rename into ...
29510 (thumb1_compute_save_core_reg_mask): This.
29511 (arm_compute_save_reg0_reg12_mask): Adapt comment.
29512 (arm_compute_frame_layout): Likewise.
29514 2017-06-06 Richard Biener <rguenther@suse.de>
29516 PR tree-optimization/80974
29517 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
29518 keep or clear leaders SSA info.
29520 2017-06-06 Tom de Vries <tom@codesourcery.com>
29522 * config/nvptx/nvptx.c (split_mode_p): New function.
29523 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
29525 2017-06-06 Tom de Vries <tom@codesourcery.com>
29527 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
29529 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29532 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
29535 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29537 * shrink-wrap.c (handle_simple_exit): Update profile.
29538 (try_shrink_wrapping): Upate profile.
29540 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29542 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
29543 (tree_guess_outgoing_edge_probabilities): New.
29544 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
29545 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
29547 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29549 * ipa-split.c (split_function): Initialize return bb profile.
29551 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29553 * profile.c (compute_branch_probabilities): Also initialize
29554 EXIT_BLOCK profile.
29556 2017-06-06 Richard Biener <rguenther@suse.de>
29558 PR tree-optimization/80928
29559 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
29560 (vect_analyze_loop_operations): Properly guard analysis for
29562 (vect_transform_loop): Likewise.
29563 (vect_analyze_loop_2): Also reset SLP type on PHIs.
29564 (vect_model_induction_cost): Do not cost for pure SLP.
29565 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
29566 of induction in inner loop vectorization.
29567 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
29568 (vect_get_and_check_slp_defs): Handle vect_induction_def.
29569 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
29571 (vect_analyze_slp_cost_1): Cost induction.
29572 (vect_detect_hybrid_slp_stmts): Handle PHIs.
29573 (vect_get_slp_vect_defs): Likewise.
29574 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
29575 (vect_transform_stmt): Handle SLP reductions.
29576 * tree-vectorizer.h (vectorizable_induction): Adjust.
29578 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29580 * config/rs6000/rs6000.c (make_resolver_func): Update
29581 init_lowered_empty_function call.
29583 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
29585 * doc/invoke.texi: Document the -fprofile-abs-path option.
29586 * common.opt (fprofile-abs-path): New option.
29587 * gcov-io.h (gcov_write_filename): Declare.
29588 * gcov-io.c (gcov_write_filename): New function.
29589 * coverage.c (coverage_begin_function): Use gcov_write_filename.
29590 * profile.c (output_location): Likewise.
29592 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29594 * shring-wrap.c: Revert accidental commit.
29596 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
29598 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
29600 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29602 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
29604 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
29605 profile in callgraph edge.
29606 * profile-count.h (apply_probability): If THIS is 0, then result is 0
29607 (apply_scale): Likewise.
29608 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
29609 Also scale profile when inlining function with zero profile.
29610 (initialize_cfun): Update exit block profile even when it is zero.
29611 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
29612 when profile is read.
29614 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29616 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
29617 (CLONE_*): New constants to define the processors we can generate
29618 code for with the target_clone attribute.
29619 (rs6000_clone_map): New array to identify which clone processors
29620 the current program is running on.
29621 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
29622 target_clone attribute.
29623 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
29624 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
29625 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
29626 (cpu_expand_builtin): Add support for target_clone attribute.
29627 (rs6000_valid_attribute_p): Allow "default" attribute.
29628 (get_decl_name): New debug function to simplify printing the
29629 current function name in debugging statements.
29630 (rs6000_clone_priority): New functions to support the target_clone
29631 attribute, and be able to generate code to switch between ISA 2.05
29632 through ISA 3.0 (power6 through power9).
29633 (rs6000_compare_version_priority): Likewise.
29634 (rs6000_get_function_versions_dispatcher): Likewise.
29635 (make_resolver_func): Likewise.
29636 (add_condition_to_bb): Likewise.
29637 (dispatch_function_versions): Likewise.
29638 (rs6000_generate_version_dispatcher_body): Likewise.
29639 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
29640 (fusion_gpr_load_p): Fix a spacing issue.
29641 * doc/extend.texi (Common Function Attributes): Document that the
29642 PowerPC supports the target_clone attribute.
29644 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29646 * config/arm/arm.h: explain F symbol found in description of ARM
29647 register allocation in its legend.
29649 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29651 * config/mips/frame-header-opt.c: Include profile-count.h.
29652 * config/riscv/riscv.c: Include profile-count.h
29654 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29656 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
29658 (sm_set_flag_if_changed): Add bbs field.
29659 (execute_sm_if_changed_flag_set): Pass BBS.
29660 (execute_sm): Update.
29662 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29664 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
29667 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29669 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
29670 (peephole2): New peephole2 to emit the above.
29671 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
29673 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29675 * config/aarch64/aarch64.c (define_peephole2 above
29676 *sub_<shift>_<mode>): New peephole.
29678 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29680 * config/i386/i386.c (make_resolver_func): Update.
29681 * Makefile.in: Add profile-count.h and profile-count.o
29682 * auto-profile.c (afdo_indirect_call): Update to new API.
29683 (afdo_set_bb_count): Update.
29684 (afdo_propagate_edge): Update.
29685 (afdo_propagate_circuit): Update.
29686 (afdo_calculate_branch_prob): Update.
29687 (afdo_annotate_cfg): Update.
29688 * basic-block.h: Include profile-count.h
29689 (struct edge_def): Turn count to profile_count.
29690 (struct basic_block_def): Likewie.
29691 (REG_BR_PROB_BASE): Move to profile-count.h
29692 (RDIV): Move to profile-count.h
29693 * bb-reorder.c (max_entry_count): Turn to profile_count.
29694 (find_traces): Update.
29695 (rotate_loop):Update.
29696 (connect_traces):Update.
29697 (sanitize_hot_paths):Update.
29698 * bt-load.c (migrate_btr_defs): Update.
29699 * cfg.c (RDIV): Remove.
29700 (init_flow): Use alloc_block.
29701 (alloc_block): Uninitialize count.
29702 (unchecked_make_edge): Uninitialize count.
29703 (check_bb_profile): Update.
29704 (dump_edge_info): Update.
29705 (dump_bb_info): Update.
29706 (update_bb_profile_for_threading): Update.
29707 (scale_bbs_frequencies_int): Update.
29708 (scale_bbs_frequencies_gcov_type): Update.
29709 (scale_bbs_frequencies_profile_count): New.
29710 * cfg.h (update_bb_profile_for_threading): Update.
29711 (scale_bbs_frequencies_profile_count): Declare.
29712 * cfgbuild.c (compute_outgoing_frequencies): Update.
29713 (find_many_sub_basic_blocks): Update.
29714 * cfgcleanup.c (try_forward_edges): Update.
29715 (try_crossjump_to_edge): Update.
29716 * cfgexpand.c (expand_gimple_tailcall): Update.
29717 (construct_exit_block): Update.
29718 * cfghooks.c (verify_flow_info): Update.
29719 (dump_bb_for_graph): Update.
29720 (split_edge): Update.
29721 (make_forwarder_block): Update.
29722 (duplicate_block): Update.
29723 (account_profile_record): Update.
29724 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
29725 (get_estimated_loop_iterations): Update.
29726 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
29727 (single_likely_exit): Update.
29728 * cfgloopmanip.c (scale_loop_profile): Update.
29730 (set_zero_probability): Update.
29731 (lv_adjust_loop_entry_edge): Update.
29732 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
29733 (purge_dead_edges): Update.
29734 (rtl_account_profile_record): Update.
29735 * cgraph.c (cgraph_node::create): Uninitialize count.
29736 (symbol_table::create_edge): Uninitialize count.
29737 (cgraph_update_edges_for_call_stmt_node): Update.
29738 (cgraph_edge::dump_edge_flags): Update.
29739 (cgraph_node::dump): Update.
29740 (cgraph_edge::maybe_hot_p): Update.
29741 * cgraph.h: Include profile-count.h
29742 (create_clone), create_edge, create_indirect_edge): Update.
29743 (cgraph_node): Turn count to profile_count.
29744 (cgraph_edge0: Likewise.
29745 (make_speculative, clone): Update.
29746 (create_edge): Update.
29747 (init_lowered_empty_function): Update.
29748 * cgraphclones.c (cgraph_edge::clone): Update.
29749 (duplicate_thunk_for_node): Update.
29750 (cgraph_node::create_clone): Update.
29751 * cgraphunit.c (cgraph_node::analyze): Update.
29752 (cgraph_node::expand_thunk): Update.
29753 * final.c (dump_basic_block_info): Update.
29754 * gimple-streamer-in.c (input_bb): Update.
29755 * gimple-streamer-out.c (output_bb): Update.
29756 * graphite.c (print_global_statistics): Update.
29757 (print_graphite_scop_statistics): Update.
29758 * hsa-brig.c: Include basic-block.h.
29759 * hsa-dump.c: Include basic-block.h.
29760 * hsa-gen.c (T sum_slice): Update.
29761 (convert_switch_statements):Update.
29762 * hsa-regalloc.c: Include basic-block.h.
29763 * ipa-chkp.c (chkp_produce_thunks): Update.
29764 * ipa-cp.c (struct caller_statistics): Update.
29765 (init_caller_stats): Update.
29766 (gather_caller_stats): Update.
29767 (ipcp_cloning_candidate_p): Update.
29768 (good_cloning_opportunity_p): Update.
29769 (get_info_about_necessary_edges): Update.
29770 (dump_profile_updates): Update.
29771 (update_profiling_info): Update.
29772 (update_specialized_profile): Update.
29773 (perhaps_add_new_callers): Update.
29774 (decide_about_value): Update.
29775 (ipa_cp_c_finalize): Update.
29776 * ipa-devirt.c (struct odr_type_warn_count): Update.
29777 (struct decl_warn_count): Update.
29778 (struct final_warning_record): Update.
29779 (possible_polymorphic_call_targets): Update.
29780 (ipa_devirt): Update.
29781 * ipa-fnsummary.c (redirect_to_unreachable): Update.
29782 * ipa-icf.c (sem_function::merge): Update.
29783 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
29784 * ipa-inline.c (compute_uninlined_call_time): Update.
29785 (compute_inlined_call_time): Update.
29786 (want_inline_small_function_p): Update.
29787 (want_inline_self_recursive_call_p): Update.
29788 (edge_badness): Update.
29789 (lookup_recursive_calls): Update.
29790 (recursive_inlining): Update.
29791 (inline_small_functions): Update.
29792 (dump_overall_stats): Update.
29793 (dump_inline_stats): Update.
29794 * ipa-profile.c (ipa_profile_generate_summary): Update.
29795 (ipa_propagate_frequency): Update.
29796 (ipa_profile): Update.
29797 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29798 * ipa-utils.c (ipa_merge_profiles): Update.
29799 * loop-doloop.c (doloop_modify): Update.
29800 * loop-unroll.c (report_unroll): Update.
29801 (unroll_loop_runtime_iterations): Update.
29802 * lto-cgraph.c (lto_output_edge): Update.
29803 (lto_output_node): Update.
29804 (input_node): Update.
29805 (input_edge): Update.
29806 (merge_profile_summaries): Update.
29807 * lto-streamer-in.c (input_cfg): Update.
29808 * lto-streamer-out.c (output_cfg): Update.
29809 * mcf.c (create_fixup_graph): Update.
29810 (adjust_cfg_counts): Update.
29811 (sum_edge_counts): Update.
29812 * modulo-sched.c (sms_schedule): Update.
29813 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
29814 * predict.c (maybe_hot_count_p): Update.
29815 (probably_never_executed): Update.
29816 (dump_prediction): Update.
29817 (combine_predictions_for_bb): Update.
29818 (propagate_freq): Update.
29819 (handle_missing_profiles): Update.
29820 (counts_to_freqs): Update.
29821 (rebuild_frequencies): Update.
29822 (force_edge_cold): Update.
29823 * predict.h: Include profile-count.h
29824 (maybe_hot_count_p, counts_to_freqs): UPdate.
29825 * print-rtl-function.c: Do not include cfg.h
29826 * print-rtl.c: Include basic-block.h
29827 * profile-count.c: New file.
29828 * profile-count.h: New file.
29829 * profile.c (is_edge_inconsistent): Update.
29830 (correct_negative_edge_counts): Update.
29831 (is_inconsistent): Update.
29832 (set_bb_counts): Update.
29833 (read_profile_edge_counts): Update.
29834 (compute_frequency_overlap): Update.
29835 (compute_branch_probabilities): Update; Initialize and deinitialize
29837 (branch_prob): Update.
29838 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
29839 (edge_gcov_count): New.
29840 (bb_gcov_count): New.
29841 * shrink-wrap.c (try_shrink_wrapping): Update.
29842 * tracer.c (better_p): Update.
29843 * trans-mem.c (expand_transaction): Update.
29844 (ipa_tm_insert_irr_call): Update.
29845 (ipa_tm_insert_gettmclone_call): Update.
29846 * tree-call-cdce.c: Update.
29847 * tree-cfg.c (gimple_duplicate_sese_region): Update.
29848 (gimple_duplicate_sese_tail): Update.
29849 (gimple_account_profile_record): Update.
29850 (execute_fixup_cfg): Update.
29851 * tree-inline.c (copy_bb): Update.
29852 (copy_edges_for_bb): Update.
29853 (initialize_cfun): Update.
29854 (freqs_to_counts): Update.
29855 (copy_cfg_body): Update.
29856 (expand_call_inline): Update.
29857 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29858 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29859 (try_unroll_loop_completely): Update.
29860 (try_peel_loop): Update.
29861 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29862 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29863 * tree-ssa-loop-split.c (connect_loops): Update.
29864 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29865 * tree-ssa-reassoc.c (branch_fixup): Update.
29866 * tree-ssa-tail-merge.c (replace_block_by): Update.
29867 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29868 (compute_path_counts): Update.
29869 (update_profile): Update.
29870 (recompute_probabilities): Update.
29871 (update_joiner_offpath_counts): Update.
29872 (estimated_freqs_path): Update.
29873 (freqs_to_counts_path): Update.
29874 (clear_counts_path): Update.
29875 (ssa_fix_duplicate_block_edges): Update.
29876 (duplicate_thread_path): Update.
29877 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29878 (struct switch_conv_info): Update.
29879 * tree-tailcall.c (decrease_profile): Update.
29880 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29881 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29882 * value-prof.c (check_counter): Update.
29883 (gimple_divmod_fixed_value): Update.
29884 (gimple_mod_pow2): Update.
29885 (gimple_mod_subtract): Update.
29886 (gimple_ic_transform): Update.
29887 (gimple_stringop_fixed_value): Update.
29888 * value-prof.h (gimple_ic): Update.
29890 2017-06-02 Carl Love <cel@us.ibm.com>
29892 * config/rs6000/rs6000-c: Add support for built-in functions
29893 vector double vec_doublee (vector signed int);
29894 vector double vec_doublee (vector unsigned int);
29895 vector double vec_doublee (vector float);
29896 vector double vec_doubleh (vector signed int);
29897 vector double vec_doubleh (vector unsigned int);
29898 vector double vec_doubleh (vector float);
29899 vector double vec_doublel (vector signed int);
29900 vector double vec_doublel (vector unsigned int);
29901 vector double vec_doublel (vector float);
29902 vector double vec_doubleo (vector signed int);
29903 vector double vec_doubleo (vector unsigned int);
29904 vector double vec_doubleo (vector float);.
29905 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29906 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29908 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29909 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29910 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29912 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29913 vec_doublel, vec_doubleh.
29914 * doc/extend.texi: Update the built-in documentation file for the
29915 new built-in functions.
29917 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29920 * ipa-inline-analysis.c (free_growth_caches): Set
29921 edge_removal_hook_holder to NULL after removing it.
29923 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29925 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29926 comparision with zero.
29928 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29929 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29930 for early expansion of vec_min and vec_max builtins.
29931 (builtin_function_type): Add min/max unsigned variants to those
29932 identified as having unsigned arguments.
29934 2017-06-02 Olivier Hainque <hainque@adacore.com>
29936 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29938 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29940 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29941 Use VALL_F16 iterator rather than VALL.
29943 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29945 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29946 Emit CBNZ inside loop when doing a strong exchange and comparing
29947 against zero. Generate the CC flags after the loop.
29949 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29951 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29952 (dl_section_ref): New.
29953 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29954 On AIX, append an expression to subtract the size of the
29955 section length to dl_section_ref.
29957 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29959 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29960 for early expansion of vector absolute builtins.
29962 2017-06-02 Richard Biener <rguenther@suse.de>
29964 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29965 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29967 2017-06-02 Richard Biener <rguenther@suse.de>
29969 PR tree-optimization/80948
29970 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29971 stmt order as well.
29973 2017-06-02 Richard Biener <rguenther@suse.de>
29975 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29977 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29978 for inductions as relevant.
29980 2017-06-02 Richard Biener <rguenther@suse.de>
29982 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29983 (vectorizable_induction): ... this. Remove dead code.
29985 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29987 * builtins. (expand_builtin_alloca): Remove second parameter and
29988 infer its value from the first parameter instead.
29989 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29991 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29993 PR rtl-optimization/80903
29994 * loop-doloop.c (add_test): Unshare sequence.
29996 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29998 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
30000 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
30002 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
30004 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
30005 xlogue_layout::get_instance, logue_layout::xlogue_layout,
30006 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
30007 (xlogue_layout::get_stub_rtx): Make static.
30008 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
30009 (xlogue_layout::compute_stub_managed_regs): Rename to...
30010 (xlogue_layout::count_stub_managed_regs): ...this.
30011 (xlogue_layout::is_stub_managed_reg): New function.
30012 (xlogue_layout::m_stub_names): Rename to...
30013 (xlogue_layout::s_stub_names): ...this, make static.
30014 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
30015 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
30016 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
30017 xlogue_layout::s_stub_names): Instantiate statics.
30018 (stub_managed_regs): Remove.
30019 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
30020 (disable_call_ms2sysv_xlogues): Rename to...
30021 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
30022 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
30024 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
30025 change after reload_completed.
30026 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
30028 (ix86_expand_prologue): Likewise.
30029 (ix86_expand_epilogue): Likewise.
30030 (ix86_expand_split_stack_prologue): Likewise.
30031 (ix86_compute_frame_layout): Remove frame parameter ...
30032 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
30033 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
30035 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
30036 (ix86_frame): Move from here ...
30037 * config/i386/i386.h (ix86_frame): ... to here.
30038 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
30039 complete ix86_frame data structure instead. Remove some_ld_name.
30041 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
30043 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
30044 symbols that hold a DECL_VALUE_EXPR.
30046 2017-06-01 Martin Jambor <mjambor@suse.cz>
30048 PR tree-optimization/80898
30049 * tree-sra.c (process_subtree_disqualification): Removed.
30050 (disqualify_candidate): Do not acll
30051 process_subtree_disqualification.
30052 (subtree_mark_written_and_enqueue): New function.
30053 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
30054 RHS has been disqualified and re-queue LHS if necessary. Apart
30055 from that, ignore disqualified RHS.
30057 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30059 * config/s390/s390.c (s390_emit_epilogue): Disable early return
30060 address fetch for z10 or later.
30062 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30064 * config/arc/arc.md (tst_movb): Add guard when splitting.
30066 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30068 * config/arc/arc.c (arc_can_eliminate): Test against
30069 arc_frame_pointer_needed.
30071 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30073 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
30074 to prevent store reordering.
30075 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
30076 (type): Add block type.
30077 (stack_tie): Define special instruction to be used in
30080 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30082 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
30083 constraint. It is not valid for the pattern.
30084 (noncommutative_binary_comparison): Likewise.
30086 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30088 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
30091 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30093 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
30094 be used by the reg-alloc.
30096 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30098 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
30099 reg-alloc when having mul64 or mul32x16 instructions.
30100 (mulsidi3): Likewise.
30101 (umulsidi3): Likewise.
30102 (mulsi32x16): New pattern.
30103 (mulsi64): Likewise.
30104 (mulsidi64): Likewise.
30105 (umulsidi64): Likewise.
30106 (MUL32x16_REG): Define.
30107 (mul64_600): Use MUL32x16_REG.
30108 (mac64_600): Likewise.
30109 (umul64_600): Likewise.
30110 (umac64_600): Likewise.
30112 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30114 * config/arc/arc.md (mulsi3_700): Make it commutative.
30116 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
30118 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
30120 (*sign_extendsidi2_insn): Likewise for movstosw.
30122 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
30124 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
30125 the type of the input discriminant value. Convert the
30126 discriminant value of signedness vary.
30128 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
30130 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
30131 Add to -Wall section.
30133 2017-06-01 Richard Biener <rguenther@suse.de>
30135 PR middle-end/66313
30136 * fold-const.c (fold_plusminus_mult_expr): If the factored
30137 factor may be zero use a wrapping type for the inner operation.
30138 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
30139 and handle moved defs.
30140 (process_assignment): Properly guard the unary op case. Return a
30141 tri-state indicating that moving the stmt before the call may allow
30142 to continue. Pass through to_move.
30143 (find_tail_calls): Handle moving unrelated defs before
30146 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
30149 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
30150 splitter result in the canonical way.
30152 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
30154 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
30155 also for 32bit target. Update insn attributes.
30156 (zero-extendsidi2 splitter): Allow all registers for operand 1.
30158 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
30160 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
30161 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
30162 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
30163 (_mm_maskz_min_ss): New intrinsics.
30165 2017-05-31 Martin Liska <mliska@suse.cz>
30167 * tree-vect-loop.c (vect_create_epilog_for_reduction):
30168 Change comment style to one we normally use.
30169 (vectorizable_reduction): Likewise.
30170 (vectorizable_induction): Likewise.
30171 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
30172 (vectorizable_call): Likewise.
30173 (vectorizable_simd_clone_call): Likewise.
30174 (vectorizable_conversion): Likewise.
30175 (vectorizable_assignment): Likewise.
30176 (vectorizable_shift): Likewise.
30177 (vectorizable_operation): Likewise.
30178 (vectorizable_store): Likewise.
30179 (vectorizable_load): Likewise.
30180 * tree-vectorizer.h: Likewise.
30182 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
30184 * passes.c (emergency_dump_function): New.
30185 * tree-pass.h (emergency_dump_function): Declare.
30186 * plugin.c (plugins_internal_error_function): Remove.
30187 * plugin.h (plugins_internal_error_function): Remove declaration.
30188 * toplev.c (internal_error_function): New static function. Use it...
30189 (general_init): ...here.
30191 2017-05-31 Graham Markall <graham.markall@embecosm.com>
30193 * config/arc/arc.c (arc_print_operand): Handle constant operands.
30194 (arc_rtx_costs): Add costs for new patterns.
30195 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
30196 * config/arc/predicates.md: Add _1_2_3_operand predicate.
30198 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
30200 * tree-ssa-strlen.c (get_next_strinfo): New function.
30201 (get_stridx_plus_constant): Use it.
30202 (zero_length_string): Likewise.
30203 (adjust_related_strinfos): Likewise.
30204 (adjust_last_stmt): Likewise.
30206 2017-05-31 Richard Biener <rguenther@suse.de>
30209 * config/i386/i386.c (ix86_expand_builtin): Remove assert
30210 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
30212 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
30214 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
30215 loop_vinfo argument and use of dependence distance vectors.
30216 Check instead whether the two references differ only in their
30217 initial value and assume that they have the same alignment if the
30218 difference is a multiple of the vector alignment.
30219 (vect_analyze_data_refs_alignment): Update call accordingly.
30221 2017-05-31 Martin Liska <mliska@suse.cz>
30224 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
30226 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30228 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
30229 (create_intersect_range_checks): Move from ...
30230 * tree-data-ref.c (create_intersect_range_checks_index)
30231 (create_intersect_range_checks): ... to here.
30232 (create_runtime_alias_checks): New function factored from ...
30233 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
30234 here. Call above function.
30235 * tree-data-ref.h (create_runtime_alias_checks): New function.
30237 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30239 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
30240 segment length for dr_b and compute it in wide_int.
30242 2017-05-31 Richard Biener <rguenther@suse.de>
30244 PR tree-optimization/80906
30245 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
30246 and pass through iv_map.
30247 (copy_bb_and_scalar_dependences): Adjust.
30248 (translate_pending_phi_nodes): Likewise.
30249 (copy_loop_close_phi_args): Handle code-generating IVs instead
30252 2017-05-30 David Malcolm <dmalcolm@redhat.com>
30254 * diagnostic-color.c (color_dict): Add "type-diff".
30255 (parse_gcc_colors): Update comment.
30256 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
30257 -fdiagnostics-show-template-tree and -fno-elide-type.
30258 (GCC_COLORS): Add type-diff to example.
30260 (-fdiagnostics-show-template-tree): New.
30261 (-fno-elide-type): New.
30262 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
30263 the pp_format_decoder callback. Call any m_format_postprocessor's
30265 (pretty_printer::pretty_printer): Initialize
30266 m_format_postprocessor.
30267 (pretty_printer::~pretty_printer): Delete any
30268 m_format_postprocessor.
30269 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
30270 (class format_postprocessor): New class.
30271 (struct pretty_printer::format_decoder): Document the new parameters.
30272 (struct pretty_printer::m_format_postprocessor): New field.
30273 * tree-diagnostic.c (default_tree_printer): Update for new
30274 bool and const char ** params.
30275 * tree-diagnostic.h (default_tree_printer): Likewise.
30277 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
30279 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
30280 (lwa_operand): Delete rs6000_gen_cell_microcode test.
30281 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
30282 rs6000_gen_cell_microcode code.
30283 (rs6000_final_prescan_insn): Delete.
30284 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
30285 "warn-cell-microcode" entries.
30286 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
30287 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
30288 throughout. Change cc_reg_not_micro_cr0_operand to
30289 cc_reg_not_cr0_operand throughout.
30290 (*extendhi<mode>2_noload): Delete.
30291 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
30292 (mwarn-cell-microcode): Delete.
30293 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
30294 -mgen-cell-microcode and -mwarn-cell-microcode.
30296 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
30299 * config/i386/constraints.md (Yd): New constraint.
30301 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
30302 and (?Yd, r) alternatives. Update insn attributes.
30303 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
30304 and (?*Yd, r) alternatives. Update insn attributes.
30305 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
30307 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
30309 * gimplify.c (gimplify_modify_expr): Don't create a
30310 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
30313 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
30315 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
30317 2017-05-30 Richard Biener <rguenther@suse.de>
30319 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
30320 and reduc_def fields.
30321 (STMT_VINFO_REDUC_TYPE): New define.
30322 (STMT_VINFO_REDUC_DEF): Likewise.
30323 (vect_force_simple_reduction): Adjust prototype.
30324 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
30325 (vect_is_simple_reduction): Remove check_reduction argument.
30326 (vect_force_simple_reduction): Adjust and set
30327 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30328 (vectorizable_reduction): Do not re-do reduction analysis
30329 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30330 * tree-parloops.c (gather_scalar_reductions): Adjust.
30332 2017-05-30 Richard Biener <rguenther@suse.de>
30334 PR middle-end/80901
30335 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
30338 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
30340 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
30341 Introduce unknown_misalignment parameter and remove vf.
30342 (vect_peeling_hash_get_lowest_cost):
30343 Pass unknown_misalignment parameter.
30344 (vect_enhance_data_refs_alignment):
30345 Fix unsupportable data ref treatment.
30347 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30349 * tree-vect-data-refs.c (vect_get_data_access_cost):
30350 Workaround for SLP handling.
30351 (vect_enhance_data_refs_alignment):
30352 Compute costs for doing no peeling at all, compare to the best
30353 peeling costs so far and avoid peeling if cheaper.
30355 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30357 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
30358 Return peeling info and set costs to zero for unlimited cost
30360 (vect_enhance_data_refs_alignment): Also inspect all datarefs
30361 with unknown misalignment. Compute and costs for unknown
30362 misalignment, compare them to the costs for known misalignment
30363 and choose the cheapest for peeling.
30365 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30367 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
30368 (vect_get_peeling_costs_all_drs): Create function.
30369 (vect_peeling_hash_get_lowest_cost):
30370 Use vect_get_peeling_costs_all_drs.
30371 (vect_peeling_supportable): Create function.
30372 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
30374 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30376 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
30377 DR_HAS_NEGATIVE_STEP.
30378 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
30379 (vect_enhance_data_refs_alignment): Use.
30380 (vect_duplicate_ssa_name_ptr_info): Use.
30381 * tree-vectorizer.h (dr_misalignment): Use.
30382 (known_alignment_for_access_p): Use.
30384 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
30387 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
30389 (has_section_name): New function.
30391 2017-05-30 Martin Liska <mliska@suse.cz>
30394 * auto-profile.c (get_function_decl_from_block): Fix
30397 2017-05-30 Richard Biener <rguenther@suse.de>
30399 PR middle-end/80876
30400 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
30402 2017-05-30 Martin Liska <mliska@suse.cz>
30404 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
30405 * dumpfile.h (struct dump_file_info): Remove ctors.
30407 2017-05-30 Martin Liska <mliska@suse.cz>
30409 * predict.def: Fix GNU coding style.
30411 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30413 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30414 Mark 'to' argument with ATTRIBUTE_UNUSED.
30416 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30418 * config/xtensa/xtensa.c (xtensa_emit_call): Use
30419 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
30420 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
30423 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
30425 * doc/install.texi (Options specification): Restore entry of
30426 --enable-sjlj-exceptions.
30428 2017-05-27 Michael Eager <eager@eagercon.com>
30431 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
30433 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
30435 * config/microblaze/microblaze.h
30436 (FIXED_REGISTERS): Update in macro.
30437 (CALL_USED_REGISTERS): Update in macro.
30439 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
30441 * doc/install.texi: Add links to macOS binary distributions.
30443 2017-05-27 Jakub Jelinek <jakub@redhat.com>
30447 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30449 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30451 2017-05-26 Martin Liska <mliska@suse.cz>
30453 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
30455 2017-05-26 Martin Liska <mliska@suse.cz>
30457 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
30458 always leading ';; '.
30459 (dump_bb_info): Likewise.
30460 (brief_dump_cfg): Likewise.
30461 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
30462 * dumpfile.c: Remove usage of TDF_VERBOSE.
30463 * dumpfile.h (enum dump_kind): Likewise.
30464 (dump_gimple_bb_header): Do not use TDF_COMMENT.
30465 * print-tree.c (debug_verbose): Remove.
30466 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
30467 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
30468 * tree-diagnostic.c (default_tree_printer): Replace
30469 TDF_DIAGNOSTIC with TDF_SLIM.
30471 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30473 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
30474 in parameter loop, rather than loop_vinfo.
30475 (create_intersect_range_checks): Ditto.
30476 (vect_create_cond_for_alias_checks): Update call to above functions.
30478 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30480 PR tree-optimization/80815
30481 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
30482 for merging runtime alias checks. Handle negative DR_STEPs.
30484 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30486 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
30488 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
30489 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
30490 out code pruning runtime alias checks.
30491 * tree-data-ref.c (prune_runtime_alias_test_list): New function
30492 factored out from above.
30493 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30495 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30497 (prune_runtime_alias_test_list): New decalaration.
30499 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30501 * tree-vect-data-refs.c (compare_tree): Rename and move ...
30502 * tree-data-ref.c (data_ref_compare_tree): ... to here.
30503 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
30504 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
30505 (operator==, comp_dr_with_seg_len_pair): Ditto.
30506 (vect_prune_runtime_alias_test_list): Ditto.
30508 2017-05-26 Martin Liska <mliska@suse.cz>
30511 * params.def: Bound partial-inlining-entry-probability param.
30513 2017-05-26 Marek Polacek <polacek@redhat.com>
30516 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
30519 2017-05-26 Richard Biener <rguenther@suse.de>
30521 PR tree-optimization/80842
30522 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
30525 2017-05-26 Richard Biener <rguenther@suse.de>
30527 PR tree-optimization/80844
30528 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
30530 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
30532 * doc/md.texi (Machine Constraints): Update x86 family
30533 machine constraints section to match 'config/i386/constraints.md'.
30535 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
30537 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
30539 2017-05-25 Nathan Sidwell <nathan@acm.org>
30541 * doc/invoke.texi (--enable-languages): Update documentation.
30543 2017-05-25 Martin Liska <mliska@suse.cz>
30545 * dumpfile.c: Add TDF_FOLDING.
30546 * dumpfile.h (enum dump_kind): Likewise.
30547 * genmatch.c (dt_simplify::gen_1): Use it.
30549 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30551 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
30553 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30555 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30556 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
30558 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30560 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
30561 * match.pd (X == C): Rewrite it here.
30562 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
30563 with_certain_nonzero_bits2): New predicates.
30564 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
30566 2017-05-24 Nathan Sidwell <nathan@acm.org>
30568 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
30571 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
30574 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30576 * config/powerpcspe: New port. Files are copied from the rs6000
30577 port, with "rs6000" in filenames replaced by "powerpcspe".
30579 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
30581 PR rtl-optimization/80754
30582 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
30584 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
30586 * config/sparc/sparc.md (length): Return the correct value for -mflat
30587 sibcalls to match output_sibcall.
30589 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30593 * config/rs6000/rs6000.c (struct machine_function): Add new field
30595 (rs6000_get_separate_components): Init that field, use it.
30596 (rs6000_components_for_bb): Use the field.
30598 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
30600 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
30602 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
30604 PR middle-end/80823
30605 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
30607 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30610 * config/s390/s390.c (s390_check_qrst_address): Check incoming
30611 address against address_operand predicate.
30612 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
30614 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
30616 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
30617 parameters passed indirectly.
30619 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
30621 * config/i386/i386.md (*movdi_internal): Remove SSE4
30622 alternative 18 (?r, *v). Update insn attributes.
30623 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
30624 Update insn attributes.
30625 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
30626 Update insn attributes.
30627 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
30628 alternative 1 (r, v). Remove isa attribute.
30629 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
30630 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
30631 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
30633 2017-05-23 Tom de Vries <tom@codesourcery.com>
30635 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
30638 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30640 * cgraphunit.c (symbol_table::process_new_functions): Update.
30641 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
30642 (inline_generate_summary): Rename to ...
30643 (ipa_fn_summary_generate): ... this one.
30644 (inline_read_summary): Rename to ...
30645 (ipa_fn_summary_read): ... this one.
30646 (inline_write_summary): Rename to ...
30647 (ipa_fn_summary_write): ... this one.
30648 (inline_free_summary): Rename to ...
30649 (ipa_free_fn_summary): ... this one.
30650 (pass_data_local_fn_summary, pass_local_fn_summary,
30651 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
30652 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
30653 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
30654 make_pass_ipa_fn_summary): New.
30655 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
30656 inline_write_summary, inline_free_summary): Remove.
30657 (ipa_free_fn_summary) : New.
30658 * ipa-inline.c (ipa_inline): Update.
30659 (pass_ipa_inline): Do not generate summaries.
30660 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
30662 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
30663 and add pass_ipa_fn_summary.
30664 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
30666 (make_pass_inline_parameters): Remove.
30668 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
30670 * omp-low.c (struct omp_context): Remove "default_kind" member.
30673 * omp-offload.c (execute_oacc_device_lower): Remove the
30674 parallelism dimensions function attributes for unparallelized
30675 OpenACC kernels constructs.
30677 2017-05-23 Martin Liska <mliska@suse.cz>
30679 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
30681 (cgraph_edge::make_speculative): Likewise.
30682 (cgraph_edge::resolve_speculation): Likewise.
30683 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30684 (cgraph_node::dump): Likewise.
30685 * cgraph.h: Likewise.
30686 * cgraphunit.c (analyze_functions): Likewise.
30687 (symbol_table::compile): Likewise.
30688 * ipa-cp.c (print_all_lattices): Likewise.
30689 (determine_versionability): Likewise.
30690 (initialize_node_lattices): Likewise.
30691 (ipcp_verify_propagated_values): Likewise.
30692 (estimate_local_effects): Likewise.
30693 (update_profiling_info): Likewise.
30694 (create_specialized_node): Likewise.
30695 (perhaps_add_new_callers): Likewise.
30696 (decide_about_value): Likewise.
30697 (decide_whether_version_node): Likewise.
30698 (identify_dead_nodes): Likewise.
30699 (ipcp_store_bits_results): Likewise.
30700 * ipa-devirt.c (dump_targets): Likewise.
30701 (ipa_devirt): Likewise.
30702 * ipa-icf.c (sem_item::dump): Likewise.
30703 (sem_function::equals): Likewise.
30704 (sem_variable::equals): Likewise.
30705 (sem_item_optimizer::read_section): Likewise.
30706 (sem_item_optimizer::execute): Likewise.
30707 (congruence_class::dump): Likewise.
30708 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
30709 (dump_inline_summary): Likewise.
30710 (estimate_node_size_and_time): Likewise.
30711 (inline_analyze_function): Likewise.
30712 * ipa-inline-transform.c (inline_call): Likewise.
30713 * ipa-inline.c (report_inline_failed_reason): Likewise.
30714 (want_early_inline_function_p): Likewise.
30715 (edge_badness): Likewise.
30716 (update_edge_key): Likewise.
30717 (inline_small_functions): Likewise.
30718 * ipa-profile.c (ipa_profile): Likewise.
30719 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
30720 (ipa_make_edge_direct_to_target): Likewise.
30721 (remove_described_reference): Likewise.
30722 (ipa_impossible_devirt_target): Likewise.
30723 (propagate_controlled_uses): Likewise.
30724 (ipa_print_node_params): Likewise.
30725 (ipcp_transform_function): Likewise.
30726 * ipa-pure-const.c (pure_const_read_summary): Likewise.
30727 (propagate_pure_const): Likewise.
30728 * ipa-reference.c (generate_summary): Likewise.
30729 (read_write_all_from_decl): Likewise.
30730 (propagate): Likewise.
30731 (ipa_reference_read_optimization_summary): Likewise.
30732 * ipa-utils.c (ipa_merge_profiles): Likewise.
30733 * ipa.c (walk_polymorphic_call_targets): Likewise.
30734 (symbol_table::remove_unreachable_nodes): Likewise.
30735 (ipa_single_use): Likewise.
30736 * passes.c (execute_todo): Likewise.
30737 * predict.c (drop_profile): Likewise.
30738 * symtab.c (symtab_node::get_dump_name): New function.
30739 (symtab_node::dump_name): Likewise.
30740 (symtab_node::dump_asm_name): Likewise.
30741 (symtab_node::dump_references): Likewise.
30742 (symtab_node::dump_referring): Likewise.
30743 (symtab_node::dump_base): Likewise.
30744 (symtab_node::debug_symtab): Likewise.
30745 * tree-sra.c (convert_callers_for_node): Likewise.
30746 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
30747 * value-prof.c (init_node_map): Likewise.
30749 2017-05-23 Martin Liska <mliska@suse.cz>
30751 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
30752 and symtab_node::debug_symtab to symbol_table::debug.
30753 * cgraphunit.c (analyze_functions): Use the renamed function.
30754 (symbol_table::compile): Likewise.
30755 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
30756 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
30757 * passes.c (execute_todo): Likewise.
30758 * symtab.c (symbol_table::dump): New function.
30759 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
30761 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30763 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
30764 that nonconst implies exec.
30766 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30768 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30769 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30770 (inline_edge_summary_vec): Turn into ...
30771 (ipa_call_summaries): ... this one.
30772 (redirect_to_unreachable, edge_set_predicate,
30773 evaluate_properties_for_edge, inline_summary_alloc,
30774 reset_ipa_call_summary, reset_inline_summary,
30775 inline_summary_t::duplicate): Update.
30776 (inline_edge_duplication_hook): Turn to ...
30777 (ipa_call_summary_t::duplicate): ... this one.
30778 (inline_edge_removal_hook): Turn to ...
30779 (ipa_call_summary_t::remove): ... this one.
30780 (dump_inline_edge_summary): Turn to ...
30781 (dump_ipa_call_summary): ... this one.
30782 (estimate_function_body_sizes): Update.
30783 (inline_update_callee_summaries): Update.
30784 (remap_edge_change_prob): Update.
30785 (remap_edge_summaries): Update.
30786 (inline_merge_summary): Update.
30787 (do_estimate_edge_time): Update.
30788 (inline_generate_summary): Update.
30789 (inline_read_section): Update.
30790 (inline_read_summary): Update.
30791 (inline_free_summary): Update.
30792 * ipa-inline.c (can_inline_edge_p): Update.
30793 (compute_inlined_call_time): Update.
30794 (want_inline_small_function_p): Update.
30795 (edge_badness): Update.
30796 (early_inliner): Update.
30797 * ipa-inline.h (inline_edge_summary): Turn to ...
30798 (ipa_call_summary): ... this one.
30799 (ipa_call_summary_t): New class.
30800 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30801 (ipa_call_summaries): New.
30802 (inline_edge_summary): Remove.
30803 (estimate_edge_growth): Update.
30804 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30805 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30806 * ipa-split.c (execute_split_functions): Update.
30807 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30809 2017-05-23 Tom de Vries <tom@codesourcery.com>
30811 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30812 attributes): Document rdrand effective target.
30814 2017-05-23 Tom de Vries <tom@codesourcery.com>
30816 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30817 attributes): Sort alphabetically.
30819 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
30821 * config/avr/genmultilib.awk: Use gsub instead of gensub.
30823 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
30826 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
30827 V2DF/V2DI splat into two separate patterns, one that handles
30828 registers, and the other that only handles memory. Drop support
30829 for splatting from a GPR on ISA 2.07 and then splitting the
30830 splat into direct move and splat.
30831 (vsx_splat_<mode>_reg): Likewise.
30832 (vsx_splat_<mode>_mem): Likewise.
30834 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
30836 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
30838 2017-05-22 Jakub Jelinek <jakub@redhat.com>
30840 PR middle-end/80809
30841 * omp-low.c (finish_taskreg_remap): New function.
30842 (finish_taskreg_scan): If unit size of ctx->record_type
30843 is non-constant, unshare the size expression and replace
30844 decls in it with possible outer var refs.
30846 PR middle-end/80809
30847 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
30848 GOVD_SHARED rather than GOVD_PRIVATE with it.
30849 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30850 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30852 PR middle-end/80853
30853 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30854 as last argument to build_outer_var_ref for pointer bases of array
30855 section reductions.
30857 2017-05-19 Martin Sebor <msebor@redhat.com>
30859 * print-tree.c (print_node): Print DECL_READ_P flag.
30861 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30863 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30864 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30865 * cgraph.c: Likewise.
30866 * cgraphunit.c: Likewise.
30867 * gengtype.c: Likewise.
30868 * ipa-cp.c: Likewise.
30869 * ipa-devirt.c: Likewise.
30870 * ipa-icf.c: Likewise.
30871 * ipa-predicate.c: Likewise.
30872 * ipa-profile.c: Likewise.
30873 * ipa-prop.c: Likewise.
30874 * ipa-split.c: Likewise.
30876 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30877 edge_predicate_pool, dump_inline_hints,
30878 inline_summary::account_size_time, redirect_to_unreachable,
30879 edge_set_predicate, set_hint_predicate,
30880 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30881 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30882 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30883 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30884 ipa_call_summary_t::remove, initialize_growth_caches,
30885 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30886 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30887 mark_modified, unmodified_parm_1, unmodified_parm,
30888 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30889 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30890 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30891 will_be_nonconstant_predicate, record_modified_bb_info,
30892 get_minimal_bb, record_modified, param_change_prob,
30893 phi_result_unknown_predicate, predicate_for_phi_result,
30894 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30895 estimate_function_body_sizes, compute_inline_parameters,
30896 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30897 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30898 inline_update_callee_summaries, remap_edge_change_prob,
30899 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30900 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30901 inline_analyze_function, inline_summary_t::insert,
30902 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30903 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30904 inline_free_summary): Move to ipa-fnsummary.h
30905 (predicate_t): Remove.
30906 * ipa-fnsummary.c: New file.
30907 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30908 (enum inline_hints_vals, inline_hints, agg_position_info,
30909 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30910 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30911 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30912 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30913 inline_read_summary, inline_write_summary, inline_free_summary,
30914 inline_analyze_function, initialize_inline_failed,
30915 inline_merge_summary, inline_update_overall_summary,
30916 compute_inline_parameters): Move to ipa-fnsummary.h
30917 * ipa-fnsummary.h: New file.
30918 * ipa-inline-transform.h: Include ipa-inline.h.
30919 * ipa-inline.c: LIkewise.
30921 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30923 * ipa-inline.c (edge_badness): Use inlined_time instead of
30924 inline_summaries->get.
30926 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30928 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30930 2017-05-22 Nathan Sidwell <nathan@acm.org>
30932 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30933 (fdump-lang): Document 'raw' option.
30934 * dumpfile.h (TDI_tu): Delete.
30935 * dumpfile.c (dump_files): Remove translation-unit.
30936 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30938 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30940 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30941 command option from $(AWK) call.
30942 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30943 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30944 [FORMAT]: Remove handling of variable.
30945 * config/avr/t-multilib: Regenerate.
30947 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30949 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30951 (dump_inline_summary): Do not print self_time.
30952 (estimate_function_body_sizes): Do not set self_time.
30953 (compute_inline_parameters): Likewise.
30954 (inline_read_section, inline_write_summary): Do not stream self_time.
30955 * ipa-inline.h (inline_summary): Drop self_time.
30957 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30959 * ipa-inline-analysis.c (account_size_time): Rename to ...
30960 (inline_summary::account_size_time): ... this one.
30961 (reset_ipa_call_summary): Turn to ...
30962 (ipa_call_summary::reset): ... this one.
30963 (reset_inline_summary): Turn to ...
30964 (inline_summary::reset): ... this one.
30965 (inline_summary_t::remove): Update.
30966 (inline_summary_t::duplicate): Update.
30967 (ipa_call_summary_t::remove): Update.
30968 (dump_inline_summary): Update.
30969 (estimate_function_body_sizes): Update.
30970 (compute_inline_parameters): Update.
30971 (estimate_node_size_and_time): Update.
30972 (inline_merge_summary): Update.
30973 (inline_update_overall_summary): Update.
30974 (inline_read_section): Update.
30975 (inline_write_summary): Update.
30976 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30977 add account_size_time and reset member functions.
30978 (ipa_call_summary): Add reset function.
30979 * ipa-predicate.h (predicate::operator &): Constify.
30981 2017-05-22 Richard Biener <rguenther@suse.de>
30983 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30985 2017-05-19 Jason Merrill <jason@redhat.com>
30987 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30989 2017-05-19 Marek Polacek <polacek@redhat.com>
30992 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30993 TYPE_OVERFLOW_WRAPS checks.
30995 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30997 * tree-core.h (enum omp_clause_default_kind): Add
30998 "OMP_CLAUSE_DEFAULT_PRESENT".
30999 * tree-pretty-print.c (dump_omp_clause): Handle it.
31000 * gimplify.c (enum gimplify_omp_var_data): Add
31001 "GOVD_MAP_FORCE_PRESENT".
31002 (gimplify_adjust_omp_clauses_1): Map it to
31003 "GOMP_MAP_FORCE_PRESENT".
31004 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
31006 * gimplify.c (oacc_default_clause): Clarify.
31008 2017-05-19 Nathan Sidwell <nathan@acm.org>
31010 LANG_HOOK_REGISTER_DUMPS
31011 * toplev.c (general_init): Call register dump lang hook.
31012 * doc/invoke.texi: Document -fdump-lang option family.
31013 * dumpfile.c (dump_files): Remove class dump here.
31014 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
31015 * dumpfile.h (tree_dump_index): Remove TDI_class.
31016 * langhooks-def.h (lhd_register_dumps): Declare.
31017 (LANG_HOOKS_REGISTER_DUMPS): Define.
31018 (LANG_HOOKS_INITIALIZER): Add it.
31019 * langhooks.c (lhd_register_dumps): Define.
31020 * langhooks.h (struct lang_hooks): Add register_dumps.
31022 2017-05-19 Nathan Sidwell <nathan@acm.org>
31024 * context.h (context::set_passes): New.
31025 * context.c (context::context): Do not create pass manager.
31026 * toplev.c (general_init): Create pass manager here.
31028 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
31030 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
31031 use this splitter if two add or or instructions would also work for
31032 the constant we want to generate.
31034 2017-05-19 Richard Biener <rguenther@suse.de>
31037 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
31038 predicate evaluation.
31040 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
31042 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
31044 * ipa-inline.c (want_inline_small_function_p): Do not cast to
31047 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
31049 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
31050 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
31051 (inline_edge_summary_vec): Turn into ...
31052 (ipa_call_summaries): ... this one.
31053 (redirect_to_unreachable, edge_set_predicate,
31054 evaluate_properties_for_edge, inline_summary_alloc,
31055 reset_ipa_call_summary, reset_inline_summary,
31056 inline_summary_t::duplicate): Update.
31057 (inline_edge_duplication_hook): Turn to ...
31058 (ipa_call_summary_t::duplicate): ... this one.
31059 (inline_edge_removal_hook): Turn to ...
31060 (ipa_call_summary_t::remove): ... this one.
31061 (dump_inline_edge_summary): Turn to ...
31062 (dump_ipa_call_summary): ... this one.
31063 (estimate_function_body_sizes): Update.
31064 (inline_update_callee_summaries): Update.
31065 (remap_edge_change_prob): Update.
31066 (remap_edge_summaries): Update.
31067 (inline_merge_summary): Update.
31068 (do_estimate_edge_time): Update.
31069 (inline_generate_summary): Update.
31070 (inline_read_section): Update.
31071 (inline_read_summary): Update.
31072 (inline_free_summary): Update.
31073 * ipa-inline.c (can_inline_edge_p): Update.
31074 (compute_inlined_call_time): Update.
31075 (want_inline_small_function_p): Update.
31076 (edge_badness): Update.
31077 (early_inliner): Update.
31078 * ipa-inline.h (inline_edge_summary): Turn to ...
31079 (ipa_call_summary): ... this one.
31080 (ipa_call_summary_t): New class.
31081 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
31082 (ipa_call_summaries): New.
31083 (inline_edge_summary): Remove.
31084 (estimate_edge_growth): Update.
31085 * ipa-profile.c (ipa_propagate_frequency_1): Update.
31086 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
31087 * ipa-split.c (execute_split_functions): Update.
31088 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
31090 2017-05-19 Richard Biener <rguenther@suse.de>
31092 PR middle-end/80764
31093 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
31095 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
31097 * config/rs6000/rs6000.c (struct machine_function): Add field
31098 fpr_is_wrapped_separately.
31099 (rs6000_get_separate_components): Use 64 components. Handle the
31100 new FPR components.
31101 (rs6000_components_for_bb): Handle the FPR components.
31102 (rs6000_emit_prologue_components): Handle the FPR components.
31103 (rs6000_emit_epilogue_components): Handle the FPR components.
31104 (rs6000_set_handled_components): Handle the FPR components.
31105 (rs6000_emit_prologue): Don't output prologue code for those FPRs
31106 that are already separately shrink-wrapped.
31107 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
31108 that are already separately shrink-wrapped.
31110 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31113 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
31116 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
31117 (define_peephole2 for Altivec d-form load): Add peepholes to catch
31118 cases where the register allocator uses a move and an offsettable
31119 memory operation to/from a FPR register on ISA 2.06/2.07.
31120 (define_peephole2 for Altivec d-form store): Likewise.
31122 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
31125 * config/i386/mmx.md (*mov<mode>_internal): Enable
31126 alternatives 11, 12, 13 and 14 also for 32bit targets.
31127 Remove alternatives 15, 16, 17 and 18.
31128 * config/i386/sse.md (vec_concatv2di): Change
31129 alternative (!x, *y) to (x, ?!*Yn).
31131 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
31133 * dumpfile.h (enum dump_kind): Remove stray comma.
31135 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
31137 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
31138 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
31139 predicate::num_conditions
31140 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
31141 (CHANGED): turn into predicate::changed.
31142 (agg_position_info): Move to ipa-predicate.h
31143 (add_condition, predicate::add_clause, predicate::operator &=,
31144 predicate::or_with, predicate::evaluate, predicate::probability,
31145 dump_condition, dump_clause, predicate::dump,
31146 predicate::remap_after_duplication, predicate::remap_after_inlining,
31147 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
31148 (evaluate_conditions_for_known_args): Update.
31149 (set_cond_stmt_execution_predicate): Update.
31150 * ipa-inline.h: Include ipa-predicate.h
31151 (condition, inline_param_summary, conditions, agg_position_info,
31152 predicate): Move to ipa-predicate.h
31153 * ipa-predicate.c: New file.
31154 * ipa-predicate.h: New file.
31156 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
31158 * final.c (leaf_function_p): Check we are not in a sequence.
31160 2017-05-18 Martin Liska <mliska@suse.cz>
31162 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
31163 * dumpfile.c (dump_register): Use new enum dump_kind.
31164 (get_dump_file_name): Likewise.
31165 (dump_enable_all): Likewise.
31166 (dump_switch_p_1): Likewise.
31167 (enable_rtl_dump_file): Remove usage of TDF_RTL.
31168 * dumpfile.h (enum dump_kind): New enum type.
31169 (struct dump_file_info): Create constructor and
31170 format fields and comments.
31171 * passes.c (pass_manager::register_one_dump_file):
31173 * statistics.c (statistics_early_init): Likewise.
31174 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
31175 TDF_TREE with TDF_SLIM.
31176 (gather_memory_references_ref): Likewise.
31178 2017-05-18 Martin Liska <mliska@suse.cz>
31180 * vec.h (struct vnull): Use it.
31182 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
31184 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
31185 (true_predicate, false_predicate, true_predicate_p,
31186 false_predicate_p): Remove.
31187 (single_cond_predicate, not_inlined_predicate): Turn to member function
31189 (add_condition): Update.
31190 (add_clause): Turn to...
31191 (predicate::add_clause): ... this one; update; allow passing NULL
31193 (and_predicates): Turn to ...
31194 (predicate::operator &=): ... this one.
31195 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
31196 (or_predicates): Turn to ...
31197 (predicate::or_with): ... this one.
31198 (evaluate_predicate): Turn to ...
31199 (predicate::evaluate): ... this one.
31200 (predicate_probability): Turn to ...
31201 (predicate::probability): ... this one.
31202 (dump_condition): Update.
31203 (dump_predicate): Turn to ...
31204 (predicate::dump): ... this one.
31205 (account_size_time): Update.
31206 (edge_set_predicate): Update.
31207 (set_hint_predicate): UPdate.
31208 (evaluate_conditions_for_known_args): Update.
31209 (evaluate_properties_for_edge): Update.
31210 (remap_predicate_after_duplication): Turn to...
31211 (predicate::remap_after_duplication): ... this one.
31212 (remap_hint_predicate_after_duplication): Update.
31213 (inline_summary_t::duplicate): UPdate.
31214 (dump_inline_edge_summary): Update.
31215 (dump_inline_summary): Update.
31216 (set_cond_stmt_execution_predicate): Update.
31217 (set_switch_stmt_execution_predicate): Update.
31218 (compute_bb_predicates): Update.
31219 (will_be_nonconstant_expr_predicate): Update.
31220 (will_be_nonconstant_predicate): Update.
31221 (phi_result_unknown_predicate): Update.
31222 (predicate_for_phi_result): Update.
31223 (array_index_predicate): Update.
31224 (estimate_function_body_sizes): Update.
31225 (estimate_node_size_and_time): Update.
31226 (estimate_ipcp_clone_size_and_time): Update.
31227 (remap_predicate): Rename to ...
31228 (predicate::remap_after_inlining): ... this one.
31229 (remap_hint_predicate): Update.
31230 (inline_merge_summary): Update.
31231 (inline_update_overall_summary): Update.
31232 (estimate_size_after_inlining): Update.
31233 (read_predicate): Rename to ...
31234 (predicate::stream_in): ... this one.
31235 (read_inline_edge_summary): Update.
31236 (write_predicate): Rename to ...
31237 (predicate::stream_out): ... this one.
31238 (write_inline_edge_summary): Update.
31239 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
31240 (clause_t): Turn to uint32_t
31241 (predicate): Turn to class; implement constructor and operators
31243 (size_time_entry): Update.
31244 (inline_summary): Update.
31245 (inline_edge_summary): Update.
31247 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
31249 * fold-const.c (fold_binary_loc): Move transformation...
31250 * match.pd (C - X CMP X): ... here.
31252 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
31254 * config/sparc/sparc.c (sparc_option_override): Set function
31255 alignment for -mcpu=niagara7 to 64 to match the I$ line.
31256 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
31258 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
31260 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
31262 2017-05-18 Marek Polacek <polacek@redhat.com>
31265 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
31266 (pass_ubsan::execute): Call gimple_assign_single_p instead of
31267 gimple_assign_load_p.
31269 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
31271 PR middle-end/80692
31272 * real.c (do_compare): Give decimal_do_compare preference over
31273 comparing just the signs.
31275 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
31277 * doc/md.texi (Canonicalization of Instructions): Describe the
31278 canonical form of instructions that inherently set a condition
31281 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
31283 PR middle-end/80775
31284 * tree-cfg.c: Move deletion of unreachable case statements to after
31285 the merging of consecutive case labels.
31287 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
31289 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
31290 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
31291 restoring of callee-saved registers.
31293 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
31295 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
31296 * config/visium/visium.c (single_set_and_flags): Likewise.
31297 * config/visium/visium.md (Substitutions): Likewise.
31299 2017-05-17 Martin Liska <mliska@suse.cz>
31301 * cfg.c: Introduce dump_flags_t type and
31302 use it instead of int type.
31304 * cfghooks.c: Likewise.
31305 * cfghooks.h (struct cfg_hooks): Likewise.
31306 * cfgrtl.c: Likewise.
31307 * cfgrtl.h: Likewise.
31308 * cgraph.c (cgraph_node::get_body): Likewise.
31309 * coretypes.h: Likewise.
31310 * domwalk.c: Likewise.
31311 * domwalk.h: Likewise.
31312 * dumpfile.c (struct dump_option_value_info): Likewise.
31313 (dump_enable_all): Likewise.
31314 (dump_switch_p_1): Likewise.
31315 (opt_info_switch_p): Likewise.
31316 * dumpfile.h (enum tree_dump_index): Likewise.
31317 (struct dump_file_info): Likewise.
31318 * genemit.c: Likewise.
31319 * generic-match-head.c: Likewise.
31320 * gengtype.c (open_base_files): Likewise.
31321 * gimple-pretty-print.c: Likewise.
31322 * gimple-pretty-print.h: Likewise.
31323 * graph.c (print_graph_cfg): Likewise.
31324 * graphite-scop-detection.c (dot_all_sese): Likewise.
31325 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
31326 * loop-unroll.c (report_unroll): Likewise.
31327 * passes.c (pass_manager::register_one_dump_file): Likewise.
31328 * print-tree.c: Likewise.
31329 * statistics.c: Likewise.
31330 * tree-cfg.c: Likewise.
31331 * tree-cfg.h: Likewise.
31332 * tree-dfa.c: Likewise.
31333 * tree-dfa.h: Likewise.
31334 * tree-dump.c (dump_function): Likewise.
31335 * tree-dump.h (struct dump_info): Likewise.
31336 * tree-pretty-print.c: Likewise.
31337 * tree-pretty-print.h: Likewise.
31338 * tree-ssa-live.c: Likewise.
31339 * tree-ssa-live.h: Likewise.
31340 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
31341 * tree-vect-loop.c: Likewise.
31342 * tree-vect-slp.c: Likewise.
31344 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
31345 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31347 PR tree-optimization/80457
31348 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
31349 of all arguments to a statement as scalar_to_vec operations.
31350 (vectorizable_call): Adjust call to vect_model_simple_cost for
31352 (vectorizable_conversion): Likewise.
31353 (vectorizable_assignment): Likewise.
31354 (vectorizable_shift): Likewise.
31355 (vectorizable_operation): Likewise.
31356 (vectorizable_comparison): Likewise.
31357 (vect_is_simple_cond): Record the def types for operands.
31358 (vectorizable_condition): Likewise, call vect_model_simple_cost.
31359 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
31360 for statement argument count.
31362 2017-05-16 Carl Love <cel@us.ibm.com>
31364 * config/rs6000/rs6000-c: Add support for built-in functions
31365 vector unsigned long long vec_bperm (vector unsigned long long,
31366 vector unsigned char)
31367 vector signed long long vec_mule (vector signed int,
31369 vector unsigned long long vec_mule (vector unsigned int,
31370 vector unsigned int)
31371 vector signed long long vec_mulo (vector signed int,
31373 vector unsigned long long vec_mulo (vector unsigned int,
31374 vector unsigned int)
31375 vector signed char vec_sldw (vector signed char,
31376 vector signed char,
31378 vector unsigned char vec_sldw (vector unsigned char,
31379 vector unsigned char,
31381 vector signed short vec_sldw (vector signed short,
31382 vector signed short,
31384 vector unsigned short vec_sldw (vector unsigned short,
31385 vector unsigned short,
31387 vector signed int vec_sldw (vector signed int,
31390 vector unsigned int vec_sldw (vector unsigned int,
31391 vector unsigned int,
31393 vector signed long long vec_sldw (vector signed long long,
31394 vector signed long long,
31396 vector unsigned long long vec_sldw (vector unsigned long long,
31397 vector unsigned long long,
31399 * config/rs6000/rs6000-c: Add support for built-in functions
31400 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
31401 * config/rs6000/altivec.h: Add defintion for vec_sldw.
31402 * doc/extend.texi: Update the built-in documentation for the
31403 new built-in functions.
31405 2017-05-16 Marek Polacek <polacek@redhat.com>
31409 * tree.c (save_expr): Don't fold the expression.
31411 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
31413 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
31414 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
31415 and (?*y,m). Update insn attributes.
31417 2017-05-16 Martin Liska <mliska@suse.cz>
31419 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
31420 flags argument of print_gimple_stmt, print_gimple_expr,
31421 print_generic_stmt and print_generic_expr.
31422 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
31423 * coretypes.h: Likewise.
31424 * except.c (dump_eh_tree): Likewise.
31425 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
31426 * gimple-pretty-print.h: Likewise.
31427 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
31428 (backprop::push_to_worklist): Likewise.
31429 (backprop::pop_from_worklist): Likewise.
31430 (backprop::process_use): Likewise.
31431 (backprop::intersect_uses): Likewise.
31432 (note_replacement): Likewise.
31433 * gimple-ssa-store-merging.c
31434 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
31435 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
31436 (pass_store_merging::execute): Likewise.
31437 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
31438 (ssa_base_cand_dump_callback): Likewise.
31439 (dump_incr_vec): Likewise.
31440 (replace_refs): Likewise.
31441 (replace_mult_candidate): Likewise.
31442 (create_add_on_incoming_edge): Likewise.
31443 (create_phi_basis): Likewise.
31444 (insert_initializers): Likewise.
31445 (all_phi_incrs_profitable): Likewise.
31446 (introduce_cast_before_cand): Likewise.
31447 (replace_one_candidate): Likewise.
31448 * gimplify.c (gimplify_expr): Likewise.
31449 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
31450 (set_rename): Likewise.
31451 (rename_uses): Likewise.
31452 (copy_loop_phi_nodes): Likewise.
31453 (add_close_phis_to_merge_points): Likewise.
31454 (copy_loop_close_phi_args): Likewise.
31455 (copy_cond_phi_args): Likewise.
31456 (graphite_copy_stmts_from_block): Likewise.
31457 (translate_pending_phi_nodes): Likewise.
31458 * graphite-poly.c (print_pdr): Likewise.
31459 (dump_gbb_cases): Likewise.
31460 (dump_gbb_conditions): Likewise.
31461 (print_scop_params): Likewise.
31462 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
31463 (build_cross_bb_scalars_use): Likewise.
31464 (gather_bbs::before_dom_children): Likewise.
31465 * hsa-dump.c (dump_hsa_immed): Likewise.
31466 * ipa-cp.c (print_ipcp_constant_value): Likewise.
31467 (get_replacement_map): Likewise.
31468 * ipa-inline-analysis.c (dump_condition): Likewise.
31469 (estimate_function_body_sizes): Likewise.
31470 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
31471 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
31472 * ipa-prop.c (ipa_dump_param): Likewise.
31473 (ipa_print_node_jump_functions_for_edge): Likewise.
31474 (ipa_modify_call_arguments): Likewise.
31475 (ipa_modify_expr): Likewise.
31476 (ipa_dump_param_adjustments): Likewise.
31477 (ipa_dump_agg_replacement_values): Likewise.
31478 (ipcp_modif_dom_walker::before_dom_children): Likewise.
31479 * ipa-pure-const.c (check_stmt): Likewise.
31480 (pass_nothrow::execute): Likewise.
31481 * ipa-split.c (execute_split_functions): Likewise.
31482 * omp-offload.c (dump_oacc_loop_part): Likewise.
31483 (dump_oacc_loop): Likewise.
31484 * trans-mem.c (tm_log_emit): Likewise.
31485 (tm_memopt_accumulate_memops): Likewise.
31486 (dump_tm_memopt_set): Likewise.
31487 (dump_tm_memopt_transform): Likewise.
31488 * tree-cfg.c (gimple_verify_flow_info): Likewise.
31489 (print_loop): Likewise.
31490 * tree-chkp-opt.c (chkp_print_addr): Likewise.
31491 (chkp_gather_checks_info): Likewise.
31492 (chkp_get_check_result): Likewise.
31493 (chkp_remove_check_if_pass): Likewise.
31494 (chkp_use_outer_bounds_if_possible): Likewise.
31495 (chkp_reduce_bounds_lifetime): Likewise.
31496 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
31497 (chkp_mark_completed_bounds): Likewise.
31498 (chkp_register_incomplete_bounds): Likewise.
31499 (chkp_mark_invalid_bounds): Likewise.
31500 (chkp_maybe_copy_and_register_bounds): Likewise.
31501 (chkp_build_returned_bound): Likewise.
31502 (chkp_get_bound_for_parm): Likewise.
31503 (chkp_build_bndldx): Likewise.
31504 (chkp_get_bounds_by_definition): Likewise.
31505 (chkp_generate_extern_var_bounds): Likewise.
31506 (chkp_get_bounds_for_decl_addr): Likewise.
31507 * tree-chrec.c (chrec_apply): Likewise.
31508 * tree-data-ref.c (dump_data_reference): Likewise.
31509 (dump_subscript): Likewise.
31510 (dump_data_dependence_relation): Likewise.
31511 (analyze_overlapping_iterations): Likewise.
31512 * tree-inline.c (expand_call_inline): Likewise.
31513 (tree_function_versioning): Likewise.
31514 * tree-into-ssa.c (dump_defs_stack): Likewise.
31515 (dump_currdefs): Likewise.
31516 (dump_names_replaced_by): Likewise.
31517 (dump_update_ssa): Likewise.
31518 (update_ssa): Likewise.
31519 * tree-object-size.c (pass_object_sizes::execute): Likewise.
31520 * tree-parloops.c (build_new_reduction): Likewise.
31521 (try_create_reduction_list): Likewise.
31522 (ref_conflicts_with_region): Likewise.
31523 (oacc_entry_exit_ok_1): Likewise.
31524 (oacc_entry_exit_single_gang): Likewise.
31525 * tree-pretty-print.h: Likewise.
31526 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
31527 (get_scalar_evolution): Likewise.
31528 (add_to_evolution): Likewise.
31529 (get_loop_exit_condition): Likewise.
31530 (analyze_evolution_in_loop): Likewise.
31531 (analyze_initial_condition): Likewise.
31532 (analyze_scalar_evolution): Likewise.
31533 (instantiate_scev): Likewise.
31534 (number_of_latch_executions): Likewise.
31535 (gather_chrec_stats): Likewise.
31536 (final_value_replacement_loop): Likewise.
31537 (scev_const_prop): Likewise.
31538 * tree-sra.c (dump_access): Likewise.
31539 (disqualify_candidate): Likewise.
31540 (create_access): Likewise.
31541 (reject): Likewise.
31542 (maybe_add_sra_candidate): Likewise.
31543 (create_access_replacement): Likewise.
31544 (analyze_access_subtree): Likewise.
31545 (analyze_all_variable_accesses): Likewise.
31546 (sra_modify_assign): Likewise.
31547 (initialize_constant_pool_replacements): Likewise.
31548 (find_param_candidates): Likewise.
31549 (decide_one_param_reduction): Likewise.
31550 (replace_removed_params_ssa_names): Likewise.
31551 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
31552 * tree-ssa-copy.c (dump_copy_of): Likewise.
31553 (copy_prop_visit_cond_stmt): Likewise.
31554 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
31555 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
31556 (record_equivalences_from_stmt): Likewise.
31557 * tree-ssa-dse.c (compute_trims): Likewise.
31558 (delete_dead_call): Likewise.
31559 (delete_dead_assignment): Likewise.
31560 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
31561 (forward_propagate_into_cond): Likewise.
31562 (pass_forwprop::execute): Likewise.
31563 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
31564 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
31566 (move_computations_worker): Likewise.
31567 (execute_sm): Likewise.
31568 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
31569 (remove_exits_and_undefined_stmts): Likewise.
31570 (remove_redundant_iv_tests): Likewise.
31571 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
31572 (adjust_iv_update_pos): Likewise.
31573 * tree-ssa-math-opts.c (bswap_replace): Likewise.
31574 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
31575 (value_replacement): Likewise.
31576 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
31577 * tree-ssa-pre.c (print_pre_expr): Likewise.
31578 (get_representative_for): Likewise.
31579 (create_expression_by_pieces): Likewise.
31580 (insert_into_preds_of_block): Likewise.
31581 (eliminate_insert): Likewise.
31582 (eliminate_dom_walker::before_dom_children): Likewise.
31583 (eliminate): Likewise.
31584 (remove_dead_inserted_code): Likewise.
31585 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
31586 * tree-ssa-reassoc.c (get_rank): Likewise.
31587 (eliminate_duplicate_pair): Likewise.
31588 (eliminate_plus_minus_pair): Likewise.
31589 (eliminate_not_pairs): Likewise.
31590 (undistribute_ops_list): Likewise.
31591 (eliminate_redundant_comparison): Likewise.
31592 (update_range_test): Likewise.
31593 (optimize_range_tests_var_bound): Likewise.
31594 (optimize_vec_cond_expr): Likewise.
31595 (rewrite_expr_tree): Likewise.
31596 (rewrite_expr_tree_parallel): Likewise.
31597 (linearize_expr): Likewise.
31598 (break_up_subtract): Likewise.
31599 (linearize_expr_tree): Likewise.
31600 (attempt_builtin_powi): Likewise.
31601 (attempt_builtin_copysign): Likewise.
31602 (transform_stmt_to_copy): Likewise.
31603 (transform_stmt_to_multiply): Likewise.
31604 (dump_ops_vector): Likewise.
31605 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
31606 (print_scc): Likewise.
31607 (set_ssa_val_to): Likewise.
31608 (visit_reference_op_store): Likewise.
31609 (visit_use): Likewise.
31610 (sccvn_dom_walker::before_dom_children): Likewise.
31611 (run_scc_vn): Likewise.
31612 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
31614 (expr_hash_elt::print): Likewise.
31615 (const_and_copies::pop_to_marker): Likewise.
31616 (const_and_copies::record_const_or_copy_raw): Likewise.
31617 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
31618 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
31619 (dump_predicates): Likewise.
31620 (find_uninit_use): Likewise.
31621 (warn_uninitialized_phi): Likewise.
31622 (pass_late_warn_uninitialized::execute): Likewise.
31623 * tree-ssa.c (verify_vssa): Likewise.
31624 (verify_ssa): Likewise.
31625 (maybe_optimize_var): Likewise.
31626 * tree-vrp.c (dump_value_range): Likewise.
31627 (dump_all_value_ranges): Likewise.
31628 (dump_asserts_for): Likewise.
31629 (register_edge_assert_for_2): Likewise.
31630 (vrp_visit_cond_stmt): Likewise.
31631 (vrp_visit_switch_stmt): Likewise.
31632 (vrp_visit_stmt): Likewise.
31633 (vrp_visit_phi_node): Likewise.
31634 (simplify_cond_using_ranges_1): Likewise.
31635 (fold_predicate_in): Likewise.
31636 (evrp_dom_walker::before_dom_children): Likewise.
31637 (evrp_dom_walker::push_value_range): Likewise.
31638 (evrp_dom_walker::pop_value_range): Likewise.
31639 (execute_early_vrp): Likewise.
31641 2017-05-16 Richard Biener <rguenther@suse.de>
31643 * dwarf2out.c (loc_list_from_tree_1): Do not create
31644 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
31646 2017-05-16 Richard Biener <rguenther@suse.de>
31648 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
31650 (note_variable_value_in_expr): If we resolved the decl ref
31651 do not push to the stack.
31653 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
31655 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
31656 operations in fast-math mode.
31657 (vaddq_f16): Likewise.
31658 (vmul_f16): Likewise.
31659 (vmulq_f16): Likewise.
31660 (vsub_f16): Likewise.
31661 (vsubq_f16): Likewise.
31662 * config/arm/neon.md (add<mode>3): New.
31664 (fma:<VH:mode>3): New. Also remove outdated comment.
31667 2017-05-16 Martin Liska <mliska@suse.cz>
31671 * ipa-devirt.c (warn_types_mismatch): Fix typo.
31672 (odr_types_equivalent_p): Likewise.
31674 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
31676 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
31678 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
31681 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
31682 non-interunit SSE move alternatives with '?'.
31683 (zero-extendsidi peephole2): New peephole to skip intermediate
31684 general register in SSE zero-extend sequence.
31686 2017-05-15 Jeff Law <law@redhat.com>
31688 * reorg.c (relax_delay_slots): Create a new variable to hold
31689 the temporary target rather than clobbering TARGET_LABEL.
31691 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
31692 missing argument to extract_bit_field call.
31693 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
31695 2017-05-15 Martin Liska <mliska@suse.cz>
31698 * gcc.c (process_command): Do not allow empty argument of -o option.
31700 2017-05-15 Renlin Li <renlin.li@arm.com>
31702 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
31703 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
31704 * config/aarch64/constraints.md (Usf): Add long call check.
31705 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
31706 (call_value): Likewise.
31707 (sibcall): Likewise.
31708 (sibcall_value): Likewise.
31710 (call_value_insn): New.
31711 (sibcall_insn): Update rtx pattern.
31712 (sibcall_value_insn): Likewise.
31713 (call_internal): Remove.
31714 (call_value_internal): Likewise.
31715 (sibcall_internal): Likewise.
31716 (sibcall_value_internal): Likewise.
31717 (call_reg): Likewise.
31718 (call_symbol): Likewise.
31719 (call_value_reg): Likewise.
31720 (call_value_symbol): Likewise.
31722 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
31725 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
31727 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
31729 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
31730 compatible with CCGOCmode and with CCZmode.
31732 2017-05-14 Martin Sebor <msebor@redhat.com>
31734 PR middle-end/77671
31735 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
31736 (gimple_fold_builtin_snprintf): Same.
31737 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
31738 (gimple_fold_builtin_snprintf): Same.
31739 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
31740 of character types.
31741 (is_call_safe): New function.
31742 (try_substitute_return_value): Call it.
31743 (try_simplify_call): New function.
31744 (pass_sprintf_length::handle_gimple_call): Call it.
31746 2017-05-14 Martin Sebor <msebor@redhat.com>
31748 PR middle-end/80669
31749 * builtins.c (expand_builtin_stpncpy): Simplify.
31751 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
31753 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
31754 * config/i386/i386.h
31755 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
31756 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
31757 (struct machine_function): Add new members call_ms2sysv,
31758 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
31759 (struct machine_frame_state): New fields sp_realigned and
31760 sp_realigned_offset.
31761 * config/i386/i386.c
31762 (enum xlogue_stub): New enum.
31763 (enum xlogue_stub_sets): New enum.
31764 (class xlogue_layout): New class.
31765 (struct ix86_frame): New fields stack_realign_allocate_offset,
31766 stack_realign_offset and outlined_save_offset. Modify comments to
31767 detail stack layout when using out-of-line stubs.
31768 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
31769 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
31770 -mcall-ms2sysv-xlogues.
31771 (stub_managed_regs): New static variable.
31772 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
31773 registers managed by out-of-line stub.
31774 (disable_call_ms2sysv_xlogues): New function.
31775 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
31776 m->call_ms2sysv when appropriate and compute frame layout for
31778 (sp_valid_at, fp_valid_at): New inline functions.
31779 (choose_basereg): New function.
31780 (choose_baseaddr): Add align parameter, use choose_basereg and modify
31782 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
31783 Use align parameter of choose_baseaddr to generated aligned SSE movs
31785 (pro_epilogue_adjust_stack): Modify to track
31786 machine_frame_state::sp_realigned.
31787 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
31788 (ix86_nsaved_sseregs): Likewise.
31789 (ix86_emit_save_regs): Likewise.
31790 (ix86_emit_save_regs_using_mov): Likewise.
31791 (ix86_emit_save_sse_regs_using_mov): Likewise.
31792 (get_scratch_register_on_entry): Likewise.
31793 (gen_frame_set): New function.
31794 (gen_frame_load): Likewise.
31795 (gen_frame_store): Likewise.
31796 (emit_outlined_ms2sysv_save): Likewise.
31797 (emit_outlined_ms2sysv_restore): Likewise.
31798 (ix86_expand_prologue): Modify stack re-alignment code and call
31799 emit_outlined_ms2sysv_save when appropriate.
31800 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
31801 parameter rtx_insn *insn, which allows the function to be used to only
31802 generate the notes.
31803 (ix86_expand_epilogue): Modify validity checks of frame and stack
31804 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
31805 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
31806 * config/i386/predicates.md
31807 (save_multiple): New predicate.
31808 (restore_multiple): Likewise.
31809 * config/i386/sse.md
31810 (save_multiple<mode>): New pattern.
31811 (save_multiple_realign<mode>): Likewise.
31812 (restore_multiple<mode>): Likewise.
31813 (restore_multiple_and_return<mode>): Likewise.
31814 (restore_multiple_leave_return<mode>): Likewise.
31815 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
31817 2017-05-14 Julia Koval <julia.koval@intel.com>
31819 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
31820 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
31821 (__builtin_ia32_xsetbv): New builtins.
31822 * config/i386/i386.c (ix86_expand_special_args_builtin):
31824 (ix86_expand_builtin): Special expand for new intrinsics.
31825 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
31826 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
31827 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
31829 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31831 * cfganal.c (inverted_post_order_compute): Change argument type
31833 * cfganal.h (inverted_post_order_compute): Adjust prototype.
31834 * df-core.c (rest_of_handle_df_initialize): Adjust.
31835 (rest_of_handle_df_finish): Likewise.
31836 (df_analyze_1): Likewise.
31837 (df_analyze): Likewise.
31838 (loop_inverted_post_order_compute): Change argument to be a vec *.
31839 (df_analyze_loop): Adjust.
31840 (df_get_n_blocks): Likewise.
31841 (df_get_postorder): Likewise.
31842 * df.h (struct df_d): Change field to be a vec.
31843 * lcm.c (compute_laterin): Adjust.
31844 (compute_available): Likewise.
31845 * lra-lives.c (lra_create_live_ranges_1): Likewise.
31846 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
31847 * tree-ssa-pre.c (compute_antic): Likewise.
31849 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31851 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31852 (depth_first_search::depth_first_search): Change structure init
31853 function to this constructor.
31854 (depth_first_search::add_bb): Rename function to this member.
31855 (depth_first_search::execute): Likewise.
31856 (flow_dfs_compute_reverse_finish): Adjust.
31858 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31860 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31861 (longest_simple_path): Likewise.
31862 * shrink-wrap.c (spread_components): Likewise.
31863 (disqualify_problematic_components): Likewise.
31864 (emit_common_heads_for_components): Likewise.
31865 (emit_common_tails_for_components): Likewise.
31866 (insert_prologue_epilogue_for_components): Likewise.
31868 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31870 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31873 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31875 * df-core.c (df_set_blocks): Start using auto_bitmap.
31876 (df_compact_blocks): Likewise.
31877 * df-problems.c (df_rd_confluence_n): Likewise.
31878 * df-scan.c (df_insn_rescan_all): Likewise.
31879 (df_process_deferred_rescans): Likewise.
31880 (df_update_entry_block_defs): Likewise.
31881 (df_update_exit_block_uses): Likewise.
31882 (df_entry_block_bitmap_verify): Likewise.
31883 (df_exit_block_bitmap_verify): Likewise.
31884 (df_scan_verify): Likewise.
31885 * lra-constraints.c (lra_constraints): Likewise.
31886 (undo_optional_reloads): Likewise.
31887 (lra_undo_inheritance): Likewise.
31888 * lra-remat.c (calculate_gen_cands): Likewise.
31889 (do_remat): Likewise.
31890 * lra-spills.c (assign_spill_hard_regs): Likewise.
31891 (spill_pseudos): Likewise.
31892 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31893 (bitmap_set_subtract_values): Likewise.
31895 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31897 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31898 management with auto_bitmap.
31899 (fix_inter_tick): Likewise.
31900 (fix_recovery_deps): Likewise.
31901 * ira.c (add_store_equivs): Likewise.
31902 (find_moveable_pseudos): Likewise.
31903 (split_live_ranges_for_shrink_wrap): Likewise.
31904 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31905 (rtx_reuse_manager::seen_def_p): Likewise.
31906 (rtx_reuse_manager::set_seen_def): Likewise.
31907 * print-rtl.h (class rtx_reuse_manager): Likewise.
31909 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31911 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31913 (migrate_btr_def): Likewise.
31914 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31915 * df-core.c (loop_post_order_compute): Likewise.
31916 (loop_inverted_post_order_compute): Likewise.
31917 * hsa-common.h: Likewise.
31918 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31919 * init-regs.c (initialize_uninitialized_regs): Likewise.
31920 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31921 (inline_small_functions): Likewise.
31922 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31923 * ira.c (combine_and_move_insns): Likewise.
31924 (build_insn_chain): Likewise.
31925 * loop-invariant.c (find_invariants): Likewise.
31926 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31927 * predict.c (tree_predict_by_opcode): Likewise.
31928 (predict_paths_leading_to): Likewise.
31929 (predict_paths_leading_to_edge): Likewise.
31930 (estimate_loops_at_level): Likewise.
31931 (estimate_loops): Likewise.
31932 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31933 (spread_components): Likewise.
31934 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31935 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31936 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31937 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31938 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31939 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31940 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31941 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31942 (mark_threaded_blocks): Likewise.
31943 (thread_through_all_blocks): Likewise.
31944 * tree-ssa.c (verify_ssa): Likewise.
31945 (execute_update_addresses_taken): Likewise.
31946 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31948 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31950 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31952 (post_order_compute): Likewise.
31953 (inverted_post_order_compute): Likewise.
31954 (pre_and_rev_post_order_compute_fn): Likewise.
31956 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31958 * genrecog.c (int_set::int_set): Explicitly construct our
31959 auto_vec base class.
31960 * vec.h (auto_vec::auto_vec): New constructor.
31962 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31964 * bitmap.h (class auto_bitmap): New constructor taking
31965 bitmap_obstack * argument.
31967 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31969 * bitmap.h (class auto_bitmap): Change type of m_bits to
31970 bitmap_head, and adjust ctor / dtor and member operators.
31972 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31974 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31975 when returned register mode doesn't match original mode.
31977 2017-05-12 Jeff Law <law@redhat.com>
31978 Jakub Jelinek <jakub@redhat.com>
31980 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31981 we look for cc setter after the compare-elim changes.
31982 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31983 within the vector to match what compare-elim now expects.
31984 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31985 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31987 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31988 after the compare-elim changes.
31989 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31990 the vector to match what compare-elim now expects.
31991 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31992 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31993 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31994 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31995 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31997 * config/visium/visium.c (single_set_and_flags): Fix where
31998 we look for cc setter after the compare-elim changes.
31999 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
32000 with the vector to match what compare-elim now expects.
32001 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
32002 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
32003 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
32004 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
32005 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
32006 (neg<mode>2_insn_set_overflow): Likewise.
32008 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
32010 PR middle-end/79794
32011 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
32012 maybe_expand_insn call, set ops[0].target. If still set after call,
32013 set alt_rtl. Add extra arg to recursive calls.
32014 (extract_bit_field): Add alt_rtl argument. Pass to
32016 * expmed.h (extract_bit_field): Fix prototype.
32017 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
32018 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
32019 to extract_bit_field_calls.
32020 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
32021 Pass alt_rtl to extract_bit_field calls.
32022 * calls.c (store_unaligned_arguments_into_psuedos)
32023 load_register_parameters): Pass extra NULL to extract_bit_field calls.
32024 * optabs.c (maybe_legitimize_operand): Clear op->target when call
32026 * optabs.h (struct expand_operand): Add target bitfield.
32028 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
32030 * compare-elim.c (try_eliminate_compare): Canonicalize
32031 operation with embedded compare to
32032 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
32033 (set (reg) (operation)].
32035 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
32037 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
32040 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
32041 cost of adding a carry flag for ADC instruction.
32042 [case MINUS]: Ignore the cost of subtracting a carry flag
32043 for SBB instruction.
32045 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
32047 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
32049 * config/rs6000/bmiintrin.h: New file.
32050 * config/rs6000/bmi2intrin.h: New file.
32051 * config/rs6000/x86intrin.h: New file.
32053 2017-05-12 Jeff Law <law@redhat.com>
32055 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
32058 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
32060 PR middle-end/80707
32061 * tree-cfg.c: Remove cfg edges of unreachable case statements.
32063 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
32065 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
32066 early expansion of vector divide builtins.
32067 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
32068 builtins identified as having unsigned arguments.
32070 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
32072 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
32073 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
32074 expansion of vector logical operations (and, andc, or, xor,
32077 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
32079 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
32080 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
32082 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
32084 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
32085 early GIMPLE expansion of vector multiplies.
32087 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32089 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
32090 TARGET_HAVE_MOVT conditional.
32091 (movt splitter): Likewise.
32093 2017-05-12 Richard Biener <rguenther@suse.de>
32095 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
32096 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
32097 Fold all stmts not inplace.
32099 2017-05-12 Richard Biener <rguenther@suse.de>
32101 PR tree-optimization/80713
32102 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
32103 inserted_exprs bit for not removed stmts.
32105 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
32107 PR middle-end/69921
32108 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
32109 parallelized" attribute for parallelized OpenACC kernels.
32110 * omp-offload.c (execute_oacc_device_lower): Use it.
32112 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
32113 Set "oacc kernels" attribute.
32114 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
32115 parameter. Adjust all users.
32116 (oacc_fn_attrib_kernels_p): Remove function.
32117 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
32118 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
32119 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
32120 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
32121 assert "oacc kernels" attribute is set.
32123 2017-05-11 Carl Love <cel@us.ibm.com>
32125 * config/rs6000/rs6000-c: Add support for built-in functions
32126 vector unsigned char vec_popcnt (vector signed char)
32127 vector unsigned char vec_popcnt (vector unsigned char)
32128 vector unsigned short vec_popcnt (vector signed short)
32129 vector unsigned short vec_popcnt (vector unsigned short)
32130 vector unsigned int vec_popcnt (vector signed int)
32131 vector unsigned int vec_popcnt (vector unsigned int)
32132 vector unsigned long long vec_popcnt (vector signed long long)
32133 vector unsigned long long vec_popcnt (vector unsigned long long)
32134 vector signed long long vec_slo (vector signed long long,
32135 vector signed char)
32136 vector signed long long vec_slo (vector signed long long,
32137 vector unsigned char)
32138 vector unsigned long long vec_slo (vector unsigned long long,
32139 vector signed char)
32140 vector unsigned long long vec_slo (vector unsigned long long,
32141 vector unsigned char)
32142 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
32143 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
32144 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
32145 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
32146 * doc/extend.texi: Update the built-in documentation file for the
32147 new built-in functions.
32149 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
32151 * attribs.h (sorted_attr_string): Move machine independent
32152 functions for target clone support from the i386 port to common
32153 code. Rename ix86_function_versions to common_function_versions.
32154 Rename make_name to make_unique_name.
32155 (common_function_versions): Likewise.
32156 (make_unique_name): Likewise.
32157 (make_dispatcher_decl): Likewise.
32158 (is_function_default_version): Likewise.
32159 * attribs.c (attr_strcmp): Likewise.
32160 (sorted_attr_string): Likewise.
32161 (common_function_versions): Likewise.
32162 (make_unique_name): Likewise.
32163 (make_dispatcher_decl): Likewise.
32164 (is_function_default_version): Likewise.
32165 * config/i386/i386.c (attr_strcmp): Likewise.
32166 (sorted_attr_string): Likewise.
32167 (ix86_function_versions): Likewise.
32168 (make_name): Likewise.
32169 (make_dispatcher_decl): Likewise.
32170 (is_function_default_version): Likewise.
32171 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
32173 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32176 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
32177 Account for direct move costs for vec_construct of integer
32180 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
32183 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
32184 (UNSPEC_STX_ATOMIC): Ditto.
32185 (loaddi_via_sse): New insn.
32186 (storedi_via_sse): Ditto.
32187 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
32188 Update corresponding peephole2 patterns.
32189 (atomic_storedi_fpu): Ditto.
32191 2017-05-11 Julia Koval <julia.koval@intel.com>
32193 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
32194 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
32196 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
32197 (__builtin_ia32_rsqrt14ss_mask): New builtins.
32198 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
32200 2017-05-11 Nathan Sidwell <nathan@acm.org>
32202 * graphite-poly.c: Include dumpfile.h.
32204 * dumpfle.h (dump_function): Declare here ...
32205 * tree-dump.h (dump_function): ... not here.
32206 * dumpfile.c: #include tree-cfg.h.
32207 (dump_function): Move here from ...
32208 * tree-dump.c (dump_function): ... here.
32209 * gimplify.c: #include splay-tree.h, not tree-dump.h.
32210 * graphite-poly.c: Don't include tree-dump.h.
32211 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
32212 * print-tree.c: Likewise.
32213 * stor-layout.c: Likewise.
32214 * tree-nested.c: Likewise.
32216 * dumpfile.c (dump_start): Use TDF_FLAGS.
32217 (dump_enable_all): Fix TDF_KIND check thinko.
32219 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
32221 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32222 array entries to represent two legal parameterizations of the
32223 overloaded __builtin_cmpb function, as represented by the
32224 P6_OV_BUILTIN_CMPB constant.
32225 (altivec_resolve_overloaded_builtin): Add special case handling
32226 for the __builtin_cmpb function, as represented by the
32227 P6_OV_BUILTIN_CMPB constant.
32228 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
32229 (BU_P6_64BIT_2): New macro.
32230 (BU_P6_OVERLOAD_2): New macro
32231 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
32232 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
32233 (CMPB): Add overload support to represent both 32-bit and 64-bit
32234 compare-bytes function.
32235 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
32236 support for TARGET_CMPB.
32237 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
32238 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32239 documentation of the __builtin_cmpb overloaded built-in function.
32241 2017-05-11 Richard Biener <rguenther@suse.de>
32243 PR tree-optimization/80705
32244 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
32245 bases are not vectorizable.
32247 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32249 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
32250 when counting register pressure.
32252 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32254 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
32255 (iv_ca_more_deps): Renamed to ...
32256 (iv_ca_compare_deps): ... this.
32257 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
32259 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32261 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
32263 (determine_group_iv_costs): ... here.
32264 (find_inv_vars_cb): Record inv var if it's not recorded before.
32266 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32268 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
32269 (get_shiftadd_cost): Ditto.
32271 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32273 * tree-ssa-address.c: Include header file.
32274 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
32276 (add_to_parts): Refactor.
32277 (addr_to_parts): New parameter. Update use of move_hint_to_base.
32278 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
32281 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32283 PR tree-optimization/53090
32284 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
32286 (extract_cond_operands): Detect condition with IV on both sides
32287 and return COMP_IV_EXPR_2.
32288 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
32289 (rewrite_use_compare): Simplify by removing call to function
32290 extract_cond_operands.
32292 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32294 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
32295 (extract_cond_operands): Detect condition comparing against non-
32296 invariant bound and return appropriate enum value.
32297 (find_interesting_uses_cond): Update use of extract_cond_operands.
32298 Handle its return value accordingly.
32299 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
32301 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32303 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
32304 nonlinear iv_use computation in loop invariant sensitive way.
32306 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32308 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
32309 (find_iv_candidates): Call relate_compare_use_with_all_cands.
32311 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32313 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
32314 (dump_cand): Support iv_cand.inv_exprs.
32315 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
32317 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
32320 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32322 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
32324 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
32325 as local function. Include necessary header files.
32326 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
32328 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32330 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
32332 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32334 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
32335 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
32336 RSHIFT_EXPR and BIT_NOT_EXPR.
32338 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32340 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
32341 (adjust_setup_cost): New parameter supporting round up adjustment.
32342 (struct address_cost_data): Delete.
32343 (force_expr_to_var_cost): Don't bound cost with spill_cost.
32344 (split_address_cost, ptr_difference_cost): Delete.
32345 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
32346 (struct ainc_cost_data): New struct.
32347 (get_address_cost_ainc): New function.
32348 (get_address_cost, get_computation_cost): Reimplement.
32349 (determine_group_iv_cost_address): Record inv_expr for all uses of
32351 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
32352 (iv_ca_has_deps): Reimplemented to ...
32353 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
32355 (iv_ca_extend): Call iv_ca_more_deps.
32357 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32359 * tree-ssa-address.c (struct mem_address): Move to header file.
32360 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
32361 * tree-ssa-address.h (struct mem_address): Move from C file.
32362 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
32364 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32366 * tree-affine.h (aff_combination_type): New interface.
32367 (aff_combination_zero_p): Remove static.
32368 (aff_combination_const_p): New interface.
32369 (aff_combination_singleton_var_p): New interfaces.
32371 2017-05-11 Richard Biener <rguenther@suse.de>
32373 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
32374 Skip unreachable blocks and destinations.
32375 (eliminate): Move stmt removal and fixup ...
32376 (fini_eliminate): ... here. Skip inserted exprs.
32377 (pass_pre::execute): Move fini_pre after fini_eliminate.
32378 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
32379 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
32380 PRE to get rid of dead code that has invalid SSA form and
32381 split critical edges again.
32383 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32385 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32387 2017-05-11 Richard Biener <rguenther@suse.de>
32389 * passes.c (execute_function_todo): Verify loops if they are
32390 said to be up-to-date.
32391 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
32392 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
32394 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
32397 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
32398 handle calling assemble_external ourself.
32401 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
32402 modes with zero size. Enhance comment.
32404 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32406 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
32407 built-ins for vec_xl and vec_xst with short and char pointer
32410 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32412 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
32413 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
32414 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
32415 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
32416 (_mm_maskz_min_round_ss): New intrinsics.
32417 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
32418 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
32419 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
32420 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
32421 (__builtin_ia32_minss_mask_round): New builtins.
32422 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32423 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32424 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
32426 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
32427 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
32429 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
32432 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32434 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
32435 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
32436 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
32437 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
32438 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
32439 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
32440 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
32441 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32442 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32443 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
32444 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
32445 (__builtin_ia32_mulss_mask_round): New builtins.
32446 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32447 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32448 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
32450 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
32451 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32453 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32456 2017-05-10 Julia Koval <julia.koval@intel.com>
32458 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
32459 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
32460 (_mm256_setr_m128i): New intrinsics.
32462 2017-05-10 Julia Koval <julia.koval@intel.com>
32464 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
32465 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
32466 (_mm_maskz_rcp14_ss): New intrinsics.
32467 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
32468 (__builtin_ia32_rcp14ss_mask): New builtins.
32469 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
32471 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
32473 PR tree-optimization/51513
32474 * tree-cfg.c (gimple_seq_unreachable_p): New function.
32475 (assert_unreachable_fallthru_edge_p): Use it.
32476 (group_case_labels_stmt): Likewise.
32477 * tree-cfg.h: Prototype it.
32478 * stmt.c: Include cfghooks.h and tree-cfg.h.
32479 (emit_case_dispatch_table) <gap_label>: New local variable.
32480 Use it to fill dispatch table gaps.
32481 Test for default_label before updating probabilities.
32482 (expand_case) <default_label>: Remove unneeded initialization.
32483 Test for unreachable default case statement and remove its edge.
32484 Set default_label accordingly.
32485 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
32487 2017-05-10 Carl Love <cel@us.ibm.com>
32489 * config/rs6000/rs6000-c: Add support for built-in functions
32490 vector signed char vec_neg (vector signed char)
32491 vector signed short int vec_neg (vector short int)
32492 vector signed int vec_neg (vector signed int)
32493 vector signed long long vec_neg (vector signed long long)
32494 vector float vec_neg (vector float)
32495 vector double vec_neg (vector double)
32496 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
32498 * config/rs6000/altivec.h: Add define for vec_neg
32499 * doc/extend.texi: Update the built-in documentation for the
32500 new built-in functions.
32502 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32504 PR tree-optimization/77644
32505 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
32507 2017-05-10 Nathan Sidwell <nathan@acm.org>
32509 * dumpfile.h (TDI_lang_all): New.
32510 (TDF_KIND): New. Renumber others
32511 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
32513 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
32515 (get_dump_file_name): Adjust suffix generation.
32516 (dump_enable_all): Use TDF_KIND.
32517 * doc/invoke.texi (-fdump-lang-all): Document.
32519 * dumpfile.h: Tabify.
32521 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
32524 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
32525 Move member access before delete.
32527 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
32529 * tree-inline.c (expand_call_inline): Split block at stmt
32532 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
32535 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
32536 are now unused after splitting mov{sf,sd}_hardfloat.
32537 (f32_lr2): Likewise.
32538 (f32_lm): Likewise.
32539 (f32_lm2): Likewise.
32540 (f32_li): Likewise.
32541 (f32_li2): Likewise.
32542 (f32_lv): Likewise.
32543 (f32_sr): Likewise.
32544 (f32_sr2): Likewise.
32545 (f32_sm): Likewise.
32546 (f32_sm2): Likewise.
32547 (f32_si): Likewise.
32548 (f32_si2): Likewise.
32549 (f32_sv): Likewise.
32550 (f32_dm): Likewise.
32551 (f32_vsx): Likewise.
32552 (f32_av): Likewise.
32553 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
32554 For movsf, order stores so the VSX stores occur before the GPR
32555 store which encourages the register allocator to use a traditional
32556 FPR instead of a GPR. For movsd, order the stores so that the GPR
32557 store comes before the VSX stores to allow the power6 to work.
32558 This is due to the power6 not having a 32-bit integer store
32559 instruction from a FPR.
32560 (movsf_hardfloat): Likewise.
32561 (movsd_hardfloat): Likewise.
32563 2017-05-09 Martin Sebor <msebor@redhat.com>
32565 PR translation/80280
32566 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
32569 PR translation/80280
32570 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
32571 data member added in r247778.
32572 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
32574 2017-05-09 Nathan Sidwell <nathan@acm.org>
32576 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
32578 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
32581 2017-05-09 Marek Polacek <polacek@redhat.com>
32583 * doc/invoke.texi: Fix typo.
32585 2017-05-09 Richard Biener <rguenther@suse.de>
32587 * tree-vrp.c (vrp_val_is_max): Adjust comment.
32588 (vrp_val_is_min): Likewise.
32589 (set_value_range_to_value): Likewise.
32590 (set_value_range_to_nonnegative): Likewise.
32591 (gimple_assign_nonzero_p): Likewise.
32592 (gimple_stmt_nonzero_p): Likewise.
32593 (vrp_int_const_binop): Likewise. Remove unreachable case.
32594 (adjust_range_with_scev): Adjust comments.
32595 (compare_range_with_value): Likewise.
32596 (extract_range_from_phi_node): Likewise.
32597 (test_for_singularity): Likewise.
32599 2017-05-09 Richard Biener <rguenther@suse.de>
32601 * tree-vrp.c (get_single_symbol): Add assert that we don't
32602 get overflowed constants as invariant part.
32603 (compare_values_warnv): Add comment before the TREE_NO_WARNING
32604 checks. Use wi::cmp instead of recursing for integer constants.
32605 (compare_values): Just ignore whether we assumed undefined
32606 overflow instead of failing the compare.
32607 (extract_range_for_var_from_comparison_expr): Add comment before the
32608 TREE_NO_WARNING sets.
32609 (test_for_singularity): Likewise.
32610 (extract_range_from_comparison): Do not disable optimization
32611 when we assumed undefined overflow.
32612 (extract_range_basic): Remove init of unused var.
32614 2017-05-09 Richard Biener <rguenther@suse.de>
32616 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
32617 (extract_range_from_multiplicative_op_1): Adjust.
32618 (extract_range_from_binary_expr_1): Use int_const_binop.
32620 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
32623 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
32624 rs6000_store_data_bypass_p in seven define_bypass directives and
32625 in several comments.
32626 * config/rs6000/rs6000-protos.h: Add prototype for
32627 rs6000_store_data_bypass_p function.
32628 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
32629 function implements slightly different (rs6000-specific) semantics
32630 than store_data_bypass_p, returning false rather than aborting
32631 with assertion error when arguments do not satisfy the
32632 requirements of store data bypass.
32633 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
32634 rs6000_store_data_bypass_p.
32636 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
32638 * config/xtensa/xtensa-protos.h
32639 (xtensa_initial_elimination_offset): New declaration.
32640 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
32641 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
32642 macro definition, add case for FRAME_POINTER_REGNUM when
32643 FRAME_GROWS_DOWNWARD.
32644 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
32645 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
32646 xtensa_initial_elimination_offset.
32648 2017-05-08 Nathan Sidwell <nathan@acm.org>
32650 * doc/invoke.texi: Alphabetize -fdump options.
32652 2017-05-08 Martin Sebor <msebor@redhat.com>
32654 PR translation/80280
32655 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
32657 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
32659 * target.def (compute_frame_layout): New optional target hook.
32660 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
32661 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
32662 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
32664 * reload1.c (verify_initial_elim_offsets): Likewise.
32665 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
32666 (use_simple_return_p): Call arm_compute_frame_layout if needed.
32667 (arm_get_frame_offsets): Split up into this ...
32668 (arm_compute_frame_layout): ... and this function.
32670 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
32672 * config/aarch64/constraints.md (Usa): New constraint.
32673 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
32675 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
32677 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
32678 with_multilib_list after it has been checked.
32680 2017-05-08 Richard Biener <rguenther@suse.de>
32682 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
32683 (bitmap_set_subtract_values): Likewise.
32685 2017-05-08 Richard Biener <rguenther@suse.de>
32687 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
32688 (gimple_assign_nonzero): ... this and remove strict_overflow_p
32690 (gimple_stmt_nonzero_warnv_p): Rename to ...
32691 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
32693 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
32694 (extract_range_basic): Adjust, do not disable propagation on
32695 strict overflow sensitive simplification.
32696 (vrp_visit_cond_stmt): Likewise.
32698 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
32700 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
32701 body size unconditionally.
32703 2017-05-07 Jeff Law <law@redhat.com>
32706 2017-05-06 Jeff Law <law@redhat.com>
32707 PR tree-optimization/78496
32708 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32711 PR tree-optimization/78496
32712 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32713 (simplify_stmt_using_ranges): Call it.
32714 (vrp_dom_walker::before_dom_children): Extract equivalences
32715 from an ASSERT_EXPR with an equality comparison against a
32718 2017-05-06 Jeff Law <law@redhat.com>
32720 PR tree-optimization/78496
32721 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32724 PR tree-optimization/78496
32725 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32726 (simplify_stmt_using_ranges): Call it.
32727 (vrp_dom_walker::before_dom_children): Extract equivalences
32728 from an ASSERT_EXPR with an equality comparison against a
32731 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32733 * lra-constraints.c (lra_copy_reg_equiv): New function.
32734 (split_reg): Use it to copy equivalence information from the
32735 original register to the spill register.
32737 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32739 PR rtl-optimization/75964
32740 * simplify-rtx.c (simplify_const_relational_operation): Remove
32741 invalid handling of comparisons of integer ABS.
32743 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
32745 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
32746 initialize to zero.
32747 (init_regs): Remove declaration.
32748 (function_arg_advance_32): Initialize error_p as boolean variable.
32750 2017-05-05 Nathan Sidwell <nathan@acm.org>
32752 * store-motion.c (remove_reachable_equiv_notes): Reformat long
32753 lines. Use for (;;).
32755 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32757 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
32758 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
32759 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
32760 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
32761 VF=2 that require versioning.
32763 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32765 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
32768 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32770 * diagnostic.h (diagnostic_override_option_index): Convert from
32771 macro to inline function.
32773 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32775 * diagnostic.c (last_module_changed_p): New function.
32776 (set_last_module): New function.
32777 (diagnostic_report_current_module): Convert macro usage to
32778 the above functions.
32779 * diagnostic.h (diagnostic_context::last_module): Strengthen
32780 from const line_map * to const line_map_ordinary *.
32781 (diagnostic_last_module_changed): Delete macro.
32782 (diagnostic_set_last_module): Delete macro.
32784 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32786 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
32787 with diagnostic_report_diagnostic.
32788 (diagnostic_n_impl_richloc): Likewise.
32789 * diagnostic.h (report_diagnostic): Delete macro.
32790 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
32791 with diagnostic_report_diagnostic.
32792 * substring-locations.c (format_warning_va): Likewise.
32794 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32796 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
32797 save/restor of format_spec. Move option-printing code to...
32798 (print_option_information): ...this new function, and
32799 reimplement by simply printing to the pretty_printer,
32800 rather than appending to the format string.
32802 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32804 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
32805 handling logic into...
32806 (update_effective_level_from_pragmas): ...this new function.
32808 2017-05-04 Andrew Waterman <andrew@sifive.com>
32810 * config/riscv/riscv.opt (mstrict-align): New option.
32811 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
32812 (SLOW_UNALIGNED_ACCESS): Define.
32813 (riscv_slow_unaligned_access): Declare.
32814 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
32816 (riscv_slow_unaligned_access): New variable.
32817 (rocket_tune_info): Set slow_unaligned_access to true.
32818 (optimize_size_tune_info): Set slow_unaligned_access to false.
32819 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
32820 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
32821 (riscv_option_override): Set riscv_slow_unaligned_access.
32822 * doc/invoke.texi: Add -mstrict-align to RISC-V.
32824 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
32826 * config/riscv/riscv.md: Unify indentation.
32828 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
32833 * config/rs6000/rs6000.md (u code attribute): Add FIX and
32835 (extendsi<mode>2): Add support for doing sign extension via
32836 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
32837 don't have ISA 3.0 instructions.
32838 (extendsi<mode>2 splitter): Likewise.
32839 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
32840 generate the normal insns since SImode can now go in vector
32841 registers. Disallow the special UNSPECs needed for previous
32842 machines to hide SImode being used. Add new insns
32843 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
32844 (fix_trunc<mode>si2_stfiwx): Likewise.
32845 (fix_trunc<mode>si2_internal): Likewise.
32846 (fixuns_trunc<mode>si2): Likewise.
32847 (fixuns_trunc<mode>si2_stfiwx): Likewise.
32848 (fctiw<u>z_<mode>_smallint): Likewise.
32849 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32850 of floating point to 32-bit integer from doing a direct move to
32851 the GPR registers to do a store.
32852 (fctiwz_<mode>): Break long line.
32854 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32856 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32857 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32858 (addr_list, addr_offset_valid_p): New.
32859 (split_address_groups): Check offset validity with above function.
32860 (gt-tree-ssa-loop-ivopts.h): Include header file.
32862 2017-05-05 Nathan Sidwell <nathan@acm.org>
32864 * config.gcc (arm*-*-*): Add missing 'fi'.
32866 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32868 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32869 included in -fopt-info does not matter.
32870 * doc/optinfo.texi (-fopt-info): Fix description of default
32871 behavour. Explicitly say order of options included in -fopt-info
32874 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32876 * config.gcc: Allow combinations of aprofile and rmprofile values for
32877 --with-multilib-list.
32878 * config/arm/t-multilib: New file.
32879 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32880 variables. Remove setting of ISA and floating-point ABI in
32881 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32882 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32883 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32884 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32886 * config/arm/t-rmprofile: Likewise except for the matches changes.
32887 * doc/install.texi (--with-multilib-list): Document the combination of
32888 aprofile and rmprofile values and warn about pitfalls in doing that.
32890 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32892 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32893 (movdi_aarch64): Likewise.
32895 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32897 PR tree-optimization/80632
32898 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32900 (build_arrays): Initialize it for virtual phis.
32901 (fix_phi_nodes): Use it for virtual phis.
32903 PR tree-optimization/80558
32904 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32905 [x, y] op z into [x op, y op z] for op & or | if conditions
32908 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32909 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32912 * config/arm/arm.md (use_literal_pool): Remove.
32913 (64-bit immediate split): No longer takes cost into consideration
32914 if arm_disable_literal_pool is enabled.
32915 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32916 used when arm_disable_literal_pool is enabled.
32917 (arm_max_const_double_inline_cost): Remove use of
32918 arm_disable_literal_pool.
32919 (push_minipool_fix): Add assert.
32920 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32921 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32922 (no_literal_pool_sf_immediate): New.
32924 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32926 PR tree-optimization/80613
32927 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32928 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32930 2017-05-05 Richard Biener <rguenther@suse.de>
32932 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32934 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32936 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32937 of this flag from insn conditions due to removal from r247495.
32939 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32941 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32943 (arm_early_store_addr_dep_ptr): Likewise.
32944 * config/arm/aarch-common-protos.h
32945 (arm_early_load_addr_dep_ptr): Add prototype.
32946 (arm_early_store_addr_dep_ptr): Likewise.
32947 * config/arm/cortex-a53.md: Add new bypasses.
32949 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32951 * tree.c (next_type_uid): Change type to unsigned.
32952 (type_hash_canon): Decrement back next_type_uid if
32953 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32954 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32957 2017-05-04 Martin Sebor <msebor@redhat.com>
32959 * builtins.c: Fix a trivial typo in a comment.
32961 PR middle-end/79234
32962 * builtins.c (check_sizes): Adjust to handle reading past the end.
32963 Avoid printing excessive upper bound of ranges. Use %E to print
32964 tree nodes instead of converting them to %wu.
32965 (expand_builtin_memchr): New function.
32966 (compute_dest_size): Rename...
32967 (compute_objsize): ...to this.
32968 (expand_builtin_memcpy): Adjust.
32969 (expand_builtin_mempcpy): Adjust.
32970 (expand_builtin_strcat): Adjust.
32971 (expand_builtin_strcpy): Adjust.
32972 (check_strncat_sizes): Adjust.
32973 (expand_builtin_strncat): Adjust.
32974 (expand_builtin_strncpy): Adjust and simplify.
32975 (expand_builtin_memset): Adjust.
32976 (expand_builtin_bzero): Adjust.
32977 (expand_builtin_memcmp): Adjust.
32978 (expand_builtin): Handle memcmp.
32979 (maybe_emit_chk_warning): Check strncat just once.
32981 2017-05-04 Martin Sebor <msebor@redhat.com>
32983 PR preprocessor/79214
32984 PR middle-end/79222
32985 PR middle-end/79223
32986 * builtins.c (check_sizes): Add inlining context and issue
32987 warnings even when -Wno-system-headers is set.
32988 (check_strncat_sizes): Same.
32989 (expand_builtin_strncat): Same.
32990 (expand_builtin_memmove): New function.
32991 (expand_builtin_stpncpy): Same.
32992 (expand_builtin): Handle memmove and stpncpy.
32994 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32996 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32997 which is not used any more.
32999 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
33001 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
33003 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
33005 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
33006 (cortexa53_tunings): Likewise.
33007 (cortexa57_tunings): Likewise.
33008 (cortexa72_tunings): Likewise.
33009 (cortexa73_tunings): Likewise.
33011 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
33013 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
33014 Set loop alignment to 8.
33016 2017-05-04 Martin Sebor <msebor@redhat.com>
33018 PR translation/80280
33019 * builtins.c (expand_builtin_object_size): Add missing quoting to
33020 %D and like directives.
33021 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
33022 (hsa_type_for_tree_type): Same.
33023 (verify_function_arguments): Same.
33024 * symtab.c (symbol_table::change_decl_assembler_name): Same.
33025 * varasm.c (get_section): Same.
33028 2017-05-04 Martin Sebor <msebor@redhat.com>
33030 PR translation/80280
33031 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
33033 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
33035 * config/aarch64/aarch64.c (generic_addrcost_table):
33036 Change HI/TI mode setting.
33038 2017-05-04 Martin Jambor <mjambor@suse.cz>
33040 PR tree-optimization/80622
33041 * tree-sra.c (comes_initialized_p): New function.
33042 (build_accesses_from_assign): Only set write lazily when
33043 comes_initialized_p is false.
33044 (analyze_access_subtree): Use comes_initialized_p.
33045 (propagate_subaccesses_across_link): Assert !comes_initialized_p
33046 instead of testing for PARM_DECL.
33048 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33050 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
33051 constraint on operand 0 to allow more general addressing modes.
33052 Adjust output template.
33053 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
33055 * config/aarch64/aarch64-protos.h
33056 (aarch64_address_valid_for_prefetch_p): Declare prototype.
33057 * config/aarch64/constraints.md (Dp): New address constraint.
33058 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
33061 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
33063 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
33064 update use of estimate_ipcp_clone_size_and_time.
33065 (estimate_local_effects): Update use of
33066 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
33067 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
33068 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
33069 Return nonspecialized time.
33071 2017-05-04 Richard Biener <rguenther@suse.de>
33073 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
33074 for the last VUSE which def dominates the PHI. Directly call
33076 (get_continuation_for_phi_1): Remove.
33078 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
33080 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
33081 to explain the use of truncating division. Cap the number of
33082 iterations to the maximum given by nb_iterations_upper_bound,
33085 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33087 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
33088 * configure: Regenerate.
33089 * config.in: Regenerate.
33090 * config/i386/driver-mingw32.c: new file.
33091 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
33092 * config.host: Link driver-mingw32.o on MinGW host.
33093 * doc/install.texi: Document new --enable-mingw-wildcard configure
33096 2017-05-04 Marek Polacek <polacek@redhat.com>
33098 PR tree-optimization/80612
33099 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
33101 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
33102 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
33104 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
33105 (movt splitter): Likewise.
33106 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
33107 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
33108 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
33109 block for Thumb-1 with MOVT.
33110 (thumb2_legitimate_address_p): Move code block ...
33111 (can_avoid_literal_pool_for_label_p): ... into this new function.
33112 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
33114 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
33115 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
33116 "M-profile targets with the MOVT instruction".
33118 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
33120 * config/arm/arm-builtins.c (arm_init_builtins): Rename
33121 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
33122 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
33124 2017-05-04 Martin Liska <mliska@suse.cz>
33126 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
33127 variable cond_code.
33129 2017-05-04 Richard Biener <rguenther@suse.de>
33131 * tree.c (array_at_struct_end_p): Handle arrays at struct
33132 end with flexarrays more conservatively. Refactor and treat
33133 arrays of arrays or aggregates more strict. Fix
33134 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
33135 * tree.c (array_at_struct_end_p): Adjust prototype.
33136 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
33137 * gimple-fold.c (get_range_strlen): Likewise.
33138 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
33140 2017-05-04 Richard Biener <rguenther@suse.de>
33142 PR tree-optimization/31130
33143 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
33145 (supports_overflow_infinity): Likewise.
33146 (is_negative_overflow_infinity): Likewise.
33147 (is_positive_overflow_infinity): Likewise.
33148 (is_overflow_infinity): Likewise.
33149 (stmt_overflow_infinity): Likewise.
33150 (overflow_infinity_range_p): Likewise.
33151 (usable_range_p): Remove as always returning true.
33152 (make_overflow_infinity): Remove.
33153 (negative_overflow_infinity): Likewise.
33154 (positive_overflow_infinity): Likewise.
33155 (avoid_overflow_infinity): Likewise.
33156 (set_value_range): Adjust accordingly.
33157 (set_value_range_to_nonnegative): Likewise, remove now unused
33158 overflow_infinity arg.
33159 (vrp_operand_equal_p): Adjust.
33160 (update_value_range): Likewise.
33161 (range_int_cst_singleton_p): Likewise.
33162 (operand_less_p): Likewise.
33163 (compare_values_warnv): Likewise.
33164 (extract_range_for_var_from_comparison_expr): Likewise.
33165 (vrp_int_const_binop): Likewise.
33166 (zero_nonzero_bits_from_vr): Likewise.
33167 (extract_range_from_multiplicative_op_1): Likewise.
33168 (extract_range_from_binary_expr_1): Likewise.
33169 (extract_range_from_unary_expr): Likewise.
33170 (extract_range_from_comparison): Likewise.
33171 (extract_range_basic): Likewise.
33172 (adjust_range_with_scev): Likewise.
33173 (compare_ranges): Likewise.
33174 (compare_range_with_value): Likewise.
33175 (dump_value_range): Likewise.
33176 (test_for_singularity): Likewise, remove strict_overflow_p parameter
33178 (simplify_cond_using_ranges): Adjust.
33180 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
33182 * brig-builtins.def: Added a builtin for class_f64.
33183 * builtin-types.def: Added a builtin type needed by class_f64.
33185 2017-05-03 Jason Merrill <jason@redhat.com>
33187 * timevar.def: Add TV_CONSTEXPR.
33189 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33191 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
33193 2017-05-03 Martin Jambor <mjambor@suse.cz>
33195 * ipa-prop.c (ipa_update_after_lto_read): Removed.
33196 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
33197 * ipa-cp.c (ipcp_propagate_stage): Do not call
33198 ipa_update_after_lto_read.
33199 * ipa-inline.c (ipa_inline): Likewise.
33201 2017-05-03 Martin Jambor <mjambor@suse.cz>
33203 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
33204 tag. Added a default constructor and a destructor.
33205 (ipa_edge_args_sum_t): New class;
33206 (ipa_edge_args_sum): Declare.
33207 (ipa_edge_args_vector): Remove declaration.
33208 (IPA_EDGE_REF): Use ipa_edge_args_sum.
33209 (ipa_free_edge_args_substructures): Remove declaration.
33210 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
33211 (ipa_edge_args_info_available_for_edge_p): Likewise.
33212 * ipa-prop.c (ipa_edge_args_vector): Removed.
33213 (edge_removal_hook_holder): Likewise.
33214 (edge_duplication_hook_holder): Likewise.
33215 (ipa_edge_args_sum): New variable.
33216 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
33217 ipa_edge_args_vector.
33218 (ipa_free_edge_args_substructures): Likewise.
33219 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
33220 ipa_edge_args_vector.
33221 (ipa_edge_removal_hook): Turned into method
33222 ipa_edge_args_sum_t::remove.
33223 (ipa_edge_duplication_hook): Turned into method
33224 ipa_edge_args_sum_t::duplicate.
33225 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
33226 registering edge hooks.
33227 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
33228 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
33229 ipa_edge_args_sum instead of ipa_edge_args_vector.
33230 * ipa-profile.c (ipa_profile): Likewise.
33232 2017-05-03 Martin Jambor <mjambor@suse.cz>
33234 * symbol-summary.h (function_summary): New method exists.
33235 (function_summary::symtab_removal): Deallocate through release.
33236 (call_summary): New class.
33237 (gt_ggc_mx): New overload.
33238 (gt_pch_nx): Likewise.
33239 (gt_pch_nx): Likewise.
33241 2017-05-03 Jeff Law <law@redhat.com>
33243 PR tree-optimization/78496
33244 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
33245 from simplify_cond_using_ranges. Split off code to walk
33246 backwards through casts into ...
33247 (simplify_cond_using_ranges_2): New function.
33248 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
33249 (execute_vrp): After identifying jump threads, call
33250 simplify_cond_using_ranges_2.
33252 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
33255 * ipa-inline.h (inline_summary): Add ctor.
33256 (create_ggc): Do not use ggc_cleared_alloc.
33258 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
33259 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33261 * gcc.c (handle_braces): Support escaping in switch matching
33263 * doc/invoke.texi (Spec Files): Document it.
33264 Remove superfluous @code markup in items.
33266 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33268 * diagnostic-show-locus.c (struct column_range): New struct.
33269 (get_affected_columns): New function.
33270 (get_printed_columns): New function.
33271 (struct correction): New struct.
33272 (correction::ensure_capacity): New function.
33273 (correction::ensure_terminated): New function.
33274 (struct line_corrections): New struct.
33275 (line_corrections::~line_corrections): New dtor.
33276 (line_corrections::add_hint): New function.
33277 (layout::print_trailing_fixits): Reimplement in terms of the new
33279 (selftest::test_overlapped_fixit_printing): New function.
33280 (selftest::diagnostic_show_locus_c_tests): Call it.
33282 2017-05-03 Nathan Sidwell <nathan@acm.org>
33284 Canonicalize canonical type hashing
33285 * tree.h (type_hash_canon_hash): Declare.
33286 * tree.c (type_hash_list, attribute_hash_list): Move into
33287 type_hash_canon_hash.
33288 (build_type_attribute_qual_variant): Break out hash code calc into
33289 type_hash_canon_hash.
33290 (type_hash_canon_hash): New. Generic type hash computation.
33291 (build_range_type_1, build_array_type_1, build_function_type,
33292 build_method_type_directly, build_offset_type, build_complex_type,
33293 make_vector_type): Call it.
33295 2017-05-03 Richard Biener <rguenther@suse.de>
33297 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
33298 When all DRs have unknown misaligned do not always peel
33299 when there is a store but apply the same costing model as if
33300 there were only loads.
33302 2017-05-03 Richard Biener <rguenther@suse.de>
33305 PR tree-optimization/80492
33306 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33307 compare_base_decls returning dont-know properly.
33309 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
33311 * config/arm/iterators.md (CCSI): New mode iterator.
33312 (arch): New mode attribute.
33313 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
33314 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
33315 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
33316 code iterator for success result mode.
33317 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
33318 the corresponding new insn generators.
33320 2017-05-03 Bin Cheng <bin.cheng@arm.com>
33323 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33324 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33326 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
33328 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
33329 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
33330 (DDR_A): Wrap DDR argument in brackets.
33331 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
33332 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
33333 (DDR_REVERSED_P): Likewise.
33335 2017-05-03 Jakub Jelinek <jakub@redhat.com>
33337 PR tree-optimization/79472
33338 * tree-switch-conversion.c (struct switch_conv_info): Add
33339 contiguous_range and default_case_nonstandard fields.
33340 (collect_switch_conv_info): Compute contiguous_range and
33341 default_case_nonstandard fields, don't clear final_bb if
33342 contiguous_range and only the default case doesn't have the required
33344 (check_all_empty_except_final): Set default_case_nonstandard instead
33345 of failing if contiguous_range and the default case doesn't have empty
33347 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
33348 and only the default case doesn't have the required constants. Skip
33350 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
33351 if default_case_nonstandard.
33352 (build_constructors): Build constant 1 just once. Assert that default
33353 values aren't inserted in between cases if contiguous_range. Skip
33355 (build_arrays): Skip virtual phis.
33356 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
33357 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
33358 Handle virtual phis.
33359 (gen_inbound_check): Handle default_case_nonstandard case.
33360 (process_switch): Adjust check_final_bb caller. Call
33361 gather_default_values with the first non-default case instead of
33362 default case if default_case_nonstandard.
33364 2017-05-02 Nathan Sidwell <nathan@acm.org>
33366 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
33367 check. Fix formatting.
33369 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
33371 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
33372 errors when comparing specialized and unspecialized times.
33374 2017-05-02 David Malcolm <dmalcolm@redhat.com>
33376 * diagnostic-show-locus.c
33377 (layout::should_print_annotation_line_p): Make private.
33378 (layout::print_annotation_line): Make private.
33379 (layout::annotation_line_showed_range_p): Make private.
33380 (layout::show_ruler): Make private.
33381 (layout::print_source_line): Make private. Pass in line and
33382 line_width, rather than calling location_get_source_line. Drop
33384 (layout::print_leading_fixits): New method.
33385 (layout::print_any_fixits): Rename to...
33386 (layout::print_trailing_fixits): ...this, and make private.
33387 Don't print newline fixits.
33388 (diagnostic_show_locus): Move logic for printing one row into...
33389 (layout::print_line): ...this new function. Move the
33390 location_get_source_line call and error-handling from
33391 print_source_line to here. Call print_leading_fixits, and rename
33392 print_any_fixits to print_trailing_fixits.
33393 (selftest::test_fixit_insert_containing_newline): Update now that
33394 newlines are partially supported.
33395 (selftest::test_fixit_insert_containing_newline_2): New test.
33396 (selftest::test_fixit_replace_containing_newline): Update comments.
33397 (selftest::diagnostic_show_locus_c_tests): Call the new test.
33398 * edit-context.c (class added_line): New class.
33399 (class edited_line): Describe newline handling in comment.
33400 (edited_line::actually_edited_p): New method.
33401 (edited_line::print_content): Delete redundant decl.
33402 (edited_line::m_predecessors): New field.
33403 (edited_file::print_content): Call edited_line::print_content.
33404 (edited_file::print_diff): Update to support newlines.
33405 (edited_file::print_diff_hunk): Likewise.
33406 (edited_file::print_run_of_changed_lines): New function.
33407 (edited_file::print_diff_line): Convert to...
33408 (print_diff_line): ...this.
33409 (edited_file::get_effective_line_count): New function.
33410 (edited_line::edited_line): Initialize new field m_predecessors.
33411 (edited_line::~edited_line): Clean up m_predecessors.
33412 (edited_line::apply_fixit): Handle newlines.
33413 (edited_line::get_effective_line_count): New function.
33414 (edited_line::print_content): New function.
33415 (edited_line::print_diff_lines): New function.
33416 (selftest::test_applying_fixits_insert_containing_newline): New
33418 (selftest::test_applying_fixits_replace_containing_newline): New
33420 (selftest::insert_line): New function.
33421 (selftest::test_applying_fixits_multiple_lines): Add example of
33423 (selftest::edit_context_c_tests): Call the new tests.
33425 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33427 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
33428 parameter cand. Update dump information.
33429 (get_computation_cost): Update uses.
33431 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33433 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
33434 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
33435 (get_computation_at, rewrite_use_address): Update use of
33436 get_computation_aff.
33438 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33440 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
33441 (get_computation): Delete.
33442 (get_computation_cost): Implement like get_computation_cost_at.
33443 Use get_computation_at.
33444 (get_computation_cost_at): Delete.
33445 (rewrite_use_nonlinear_expr): Use get_computation_at.
33446 (rewrite_use_compare, remove_unused_ivs): Ditto.
33448 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33450 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
33452 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33454 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
33455 (ivopts_global_cost_for_size): Rename parameter and update uses.
33456 (iv_ca_recount_cost): Update uses.
33457 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
33458 candidates seperately in n_invs and n_cands.
33459 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
33461 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33463 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
33464 (find_inv_vars_cb): New.
33465 (find_depends): Renamed to ...
33466 (find_inv_vars): ... this.
33467 (add_candidate_1, force_var_cost): Call find_inv_vars.
33468 (split_address_cost, determine_group_iv_cost_cond): Ditto.
33470 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33472 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
33473 inv_vars. Add inv_exprs.
33474 (struct iv_cand): Rename depends_on to inv_vars.
33475 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
33476 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
33477 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
33478 (dump_cand): Dump inv_vars.
33479 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
33480 (record_invariant, find_depends, add_candidate_1): Ditto.
33481 (set_group_iv_cost, force_var_cost): Ditto.
33482 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
33483 (get_computation_cost_at, get_computation_cost): Ditto.
33484 (determine_group_iv_cost_generic): Ditto.
33485 (determine_group_iv_cost_address): Ditto.
33486 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
33487 (determine_group_iv_costs): Ditto.
33488 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
33489 (iv_ca_set_remove_invariants): Renamed to ...
33490 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
33491 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
33492 (iv_ca_set_add_invariants): Renamed to ...
33493 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
33494 (iv_ca_set_cp): Use iv_ca_set_add_invs.
33495 (iv_ca_has_deps): Support inv_vars and inv_exprs.
33496 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
33497 (create_new_ivs): Remove useless dump.
33499 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33501 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
33503 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
33504 (iv_ca_set_no_cp, create_new_iv): Ditto.
33506 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33508 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33510 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33512 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
33513 function tree_check2.
33515 2017-05-02 Martin Liska <mliska@suse.cz>
33517 * doc/gcov.texi: Add missing preposition.
33518 * gcov.c (function_info::function_info): Properly fill up
33519 all member variables.
33521 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33523 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
33525 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33527 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
33529 2017-05-02 Martin Liska <mliska@suse.cz>
33532 * lto-streamer-in.c (lto_read_tree_1): Remove
33533 LTO_STREAMER_DEBUG.
33534 * lto-streamer.c (struct tree_hash_entry): Likewise.
33535 (struct tree_entry_hasher): Likewise.
33536 (tree_entry_hasher::hash): Likewise.
33537 (tree_entry_hasher::equal): Likewise.
33538 (lto_streamer_init): Likewise.
33539 (lto_orig_address_map): Likewise.
33540 (lto_orig_address_get): Likewise.
33541 (lto_orig_address_remove): Likewise.
33542 * lto-streamer.h: Likewise.
33543 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
33544 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
33546 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
33548 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
33549 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
33550 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
33551 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
33552 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
33553 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
33554 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
33555 (mm_maskz_sub_ss): New intrinsics.
33556 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33557 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
33558 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
33559 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
33560 (__builtin_ia32_subss_mask_round): New builtins.
33561 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33562 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
33563 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
33565 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
33566 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
33568 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
33571 2017-05-02 Martin Jambor <mjambor@suse.cz>
33573 PR tree-optimization/78687
33574 * tree-sra.c (access): New field parent.
33575 (process_subtree_disqualification): New function.
33576 (disqualify_candidate): Call it.
33577 (build_accesses_from_assign): Reset write flag if creating an
33579 (build_access_subtree): Fill in parent field and also prpagate
33580 down grp_write flag.
33581 (create_artificial_child_access): New parameter set_grp_write, set
33582 grp_write to its value.
33583 (propagate_subaccesses_across_link): Also propagate grp_write flag
33585 (propagate_all_subaccesses): Push the closest parent back to work
33586 queue if add_access_to_work_queue returned true.
33588 2017-05-02 Richard Biener <rguenther@suse.de>
33590 * common.opt (fstrict-overflow): Alias negative to fwrapv.
33591 * doc/invoke.texi (fstrict-overflow): Remove all traces of
33592 -fstrict-overflow documentation.
33593 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
33594 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
33595 flag_strict_overflow.
33596 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
33597 * lto-opts.c (lto_write_options): Do not stream it.
33598 * lto-wrapper.c (merge_and_complain): Do not handle it.
33599 * opts.c (default_options_table): Do not set -fstrict-overflow.
33600 (finish_options): Likewise do not clear it when sanitizing.
33601 * simplify-rtx.c (simplify_const_relational_operation): Do not
33602 test flag_strict_overflow.
33604 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33606 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
33607 using enabled attribute.
33608 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
33609 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
33610 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
33611 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
33612 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
33613 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
33614 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
33615 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
33616 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
33617 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
33619 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33621 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
33623 2017-05-02 Richard Biener <rguenther@suse.de>
33625 PR tree-optimization/80591
33627 2017-04-10 Richard Biener <rguenther@suse.de>
33629 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33632 2017-05-02 Richard Biener <rguenther@suse.de>
33634 PR tree-optimization/80549
33635 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
33636 (cleanup_tree_cfg_noloop): Create forwarders to known loop
33637 headers if they do not have a preheader.
33639 2017-05-02 Martin Liska <mliska@suse.cz>
33642 * common.opt: Fix typo.
33643 * doc/invoke.texi: Likewise.
33645 2017-05-01 Jan Beulich <jbeulich@suse.com>
33647 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
33648 swapping, add (x,x,m,x,n) alternative.
33650 2017-05-01 Nathan Sidwell <nathan@acm.org>
33652 * calls.c (combine_pending_stack_adjustment_and_call): Remove
33653 unnecessary unadjusted_alignment check.
33655 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33658 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
33660 * gimplify.c (gimplify_cilk_detach): New function.
33661 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
33662 * tree-core.h: Document EXPR_CILK_SPAWN.
33663 * tree.h (EXPR_CILK_SPAWN): Define.
33665 2017-05-01 David Malcolm <dmalcolm@redhat.com>
33667 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
33668 to use new fixit_hint representation, using the "replace" logic.
33669 (get_line_span_for_fixit_hint): Likewise.
33670 (layout::print_any_fixits): Likewise.
33671 (selftest::test_one_liner_many_fixits): Rename to...
33672 (selftest::test_one_liner_many_fixits_1): ...this, and update
33673 comment and expected output to reflect that the multiple fix-it
33674 hints are now consolidated into one insertion.
33675 (selftest::test_one_liner_many_fixits_2): New test.
33676 (selftest::test_diagnostic_show_locus_one_liner): Update for
33678 (selftest::test_fixit_consolidation): Update for fix-it API
33680 * diagnostic.c (print_parseable_fixits): Likewise.
33681 * edit-context.c (edited_line::m_line_events): Convert from
33682 auto_vec <line_event *> to auto_vec <line_event>.
33683 (class line_event): Convert from abstract base class to a concrete
33684 class, taking over the role of replace_event.
33685 (class insert_event): Delete.
33686 (class replace_event): Rename to class line_event. Convert to
33688 (edit_context::add_fixits): Reimplement.
33689 (edit_context::apply_insert): Delete.
33690 (edit_context::apply_replace): Rename to...
33691 (edit_context::apply_fixit): ...this. Convert to half-open range.
33692 (edited_file::apply_insert): Delete.
33693 (edited_file::apply_replace): Rename to...
33694 (edited_file::apply_fixit): ...this.
33695 (edited_line::~edited_line): Drop deletion of events.
33696 (edited_line::apply_insert): Delete.
33697 (edited_line::apply_replace): Rename to...
33698 (edited_line::apply_fixit): ...this. Convert to half-open range.
33699 Update for change to type of m_line_events.
33700 * edit-context.h (edit_context::apply_insert): Delete.
33701 (edit_context::apply_replace): Rename to...
33702 (edit_context::apply_fixit): ...this.
33704 2017-05-01 Martin Sebor <msebor@redhat.com>
33706 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
33709 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
33712 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
33713 __get_cpuid_max returns 0.
33714 (__get_cpuid_count): Ditto.
33716 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
33718 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
33719 replacement expression is another instance of one of its arguments.
33721 2017-05-01 Jakub Jelinek <jakub@redhat.com>
33724 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
33725 check for stack push/pop autoinc.
33726 * config/i386/i386.c (ix86_agi_dependent): Return false
33727 if the only reason why modified_in_p returned true is that
33728 addr is SP based and set_insn is a push or pop.
33730 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33732 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
33735 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33738 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
33739 (account_size_time): Use two predicates - exec_pred and
33741 (evaluate_conditions_for_known_args): Compute both clause and
33743 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
33744 (inline_summary_t::duplicate): Update.
33745 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
33747 (compute_inline_parameters): Likewise.
33748 (estimate_edge_size_and_time): Update caluclation of time.
33749 (estimate_node_size_and_time): Compute both time and nonspecialized
33751 (estimate_ipcp_clone_size_and_time): Update.
33752 (inline_merge_summary): Update.
33753 (do_estimate_edge_time): Update.
33754 (do_estimate_edge_size): Update.
33755 (do_estimate_edge_hints): Update.
33756 (inline_read_section, inline_write_summary): Stream both new predicates.
33757 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
33759 (compute_inlined_call_time): Cleanup.
33760 (big_speedup_p): Update.
33761 (edge_badness): Update.
33762 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
33763 (size_time_entry): Replace predicate by exec_predicate and
33764 nonconst_predicate.
33765 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
33766 (estimate_edge_time): Return also nonspec_time.
33767 (reset_edge_growth_cache): Update.
33769 2017-04-29 Jakub Jelinek <jakub@redhat.com>
33771 PR rtl-optimization/80491
33772 * ifcvt.c (noce_process_if_block): When looking for x setter
33773 with missing else_bb, don't check only the insn right before
33774 cond_earliest, but look for the last insn that x is modified in
33775 within the same bb.
33777 PR rtl-optimization/80491
33778 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
33780 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
33782 PR tree-optimization/80487
33783 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
33785 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33787 PR tree-optimization/79697
33788 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
33789 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
33790 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
33792 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
33793 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
33795 2017-04-28 Martin Sebor <msebor@redhat.com>
33797 PR tree-optimization/80523
33798 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
33799 (init_target_to_host_charmap, target_to_host, target_strtol10): New
33801 (maybe_warn, format_directive, parse_directive): Use new functions.
33802 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
33804 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
33806 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
33808 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
33810 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
33811 target_header_dir): Set correctly.
33812 * configure: Regenerated.
33813 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33814 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33815 instead of SYSTEM_HEADER_DIR.
33817 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
33819 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
33820 (estimate_local_effects): Likewise.
33821 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
33822 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
33823 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
33824 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
33825 do_estimate_edge_time, estimate_edge_time): Likewise.
33826 * ipa-inline-analysis.c (estimate_node_size_and_time,
33827 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
33828 (estimate_time_after_inlining): Remove.
33830 2017-04-28 Martin Liska <mliska@suse.cz>
33832 * doc/gcov.texi: Enhance documentation of gcov.
33834 2017-04-28 Martin Liska <mliska@suse.cz>
33836 * doc/gcov.texi: Sort options in alphabetic order.
33837 * doc/gcov-dump.texi: Likewise.
33838 * doc/gcov-tool.texi: Likewise.
33839 * gcov.c (print_usage): Likewise.
33840 * gcov-dump.c (print_usage): Likewise.
33841 * gcov-tool.c (print_merge_usage_message): Likewise.
33842 (print_rewrite_usage_message): Likewise.
33843 (print_overlap_usage_message): Likewise.
33845 2017-04-28 Martin Liska <mliska@suse.cz>
33847 PR gcov-profile/53915
33848 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33850 2017-04-28 Martin Liska <mliska@suse.cz>
33852 PR gcov-profile/79891
33853 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33854 is marked by compiler as living on a line.
33855 (get_cycles_count): Remove usage of the union.
33856 (output_intermediate_file): Likewise.
33857 (find_source): Fix GNU coding style.
33858 (accumulate_line_counts): Remove old non-all block mode.
33859 (output_lines): Remove usage of the union.
33860 * profile.c (output_location): Include all BBs, even if
33861 belonging to a same line (and file) as a previous BB.
33863 2017-04-28 Martin Liska <mliska@suse.cz>
33865 * gcov.c (process_args): Handle new argument 'w'.
33866 (read_graph_file): Assign ID to BBs.
33867 (output_branch_count): Display BB # if verbose flag is set.
33868 (output_lines): Likewise for arcs.
33869 (print_usage): Add '--verbose' option help.
33870 * doc/gcov.texi: Document --verbose (-w) option.
33872 2017-04-28 Martin Liska <mliska@suse.cz>
33874 * gcov.c (struct block_location_info): New struct.
33875 (process_file): Fill up the new structure.
33876 (read_graph_file): Replace usage of encoding by the newly added
33878 (add_line_counts): Likewise.
33879 (accumulate_line_counts): Remove usage of the union.
33880 (function_info::function_info): New function.
33881 (function_info::~function_info): Likewise.
33882 (process_file): Call delete instead of release_function.
33883 (release_function): Release the function.
33884 (release_structures): Call delete instead of release_function.
33885 (solve_flow_graph): Replace usage of num_blocks.
33886 (find_exception_blocks): Likewise.
33887 (output_lines): Fix GNU coding style.
33889 2017-04-28 Martin Liska <mliska@suse.cz>
33892 * coverage.c (coverage_remove_note_file): New function.
33893 * coverage.h: Declare the function.
33894 * toplev.c (finalize): Clean if an error has been seen.
33896 2017-04-28 Martin Liska <mliska@suse.cz>
33898 PR gcov-profile/80031
33899 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33900 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33901 * gcov.c (read_graph_file): Read just number of blocks.
33902 * profile.c (branch_prob): Do not stream 0 flags per a basic
33905 2017-04-28 Martin Liska <mliska@suse.cz>
33907 * gcov-dump.c (tag_*): Add new argument to declarations.
33908 (dump_gcov_file): Likewise.
33909 (tag_blocks): Add and use new argument depth.
33910 (tag_arcs): Likewise.
33911 (tag_lines): Likewise.
33912 (tag_counters): Likewise.
33913 (tag_summary): Likewise.
33914 (dump_working_sets): Use depth to do a proper indentation.
33916 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33919 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33920 * symtab.c (symtab_node::debug_symtab): Move definition here.
33922 2017-04-28 Richard Biener <rguenther@suse.de>
33924 * lto-streamer.h (LTO_major_version): Bump to 7.
33926 2017-04-28 Richard Biener <rguenther@suse.de>
33928 * tree-vrp.c (assert_info): New struct.
33929 (add_assert_info): New helper.
33930 (register_edge_assert_for_2): Refactor to add asserts to a vector
33932 (register_edge_assert_for_1): Likewise.
33933 (register_edge_assert_for): Likewise.
33934 (finish_register_edge_assert_for): New helper actually registering
33935 asserts where live on edge.
33936 (find_conditional_asserts): Adjust.
33937 (find_switch_asserts): Likewise.
33938 (evrp_dom_walker::try_find_new_range): Generalize.
33939 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33941 2017-04-27 Marek Polacek <polacek@redhat.com>
33944 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33945 arg10 and arg11 to itype.
33947 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33949 * doc/extend.texi (Object Size Checking): Improve grammar.
33951 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33954 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33955 that the logic for permitting reciprocal estimates matches that
33958 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33961 * tree.c (type_cache_hasher::equal): Only compare
33962 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33963 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33964 non-aggregate element types.
33965 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33966 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33968 2017-04-27 Richard Biener <rguenther@suse.de>
33970 PR middle-end/80533
33971 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33972 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33973 keeping a reference to a trailing array.
33975 2017-04-27 Richard Biener <rguenther@suse.de>
33977 PR middle-end/80539
33978 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33979 being in loop-closed SSA form conservatively.
33980 (chrec_fold_multiply_poly_poly): Likewise.
33982 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33984 PR middle-end/79665
33985 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33986 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33988 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33991 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33992 (aarch64_function_arg_alignment): Return unsigned int again, but still
33993 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33994 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33995 Don't emit -Wpsabi note.
33996 (aarch64_function_arg_boundary): Likewise.
33997 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
34000 2017-04-26 Nathan Sidwell <nathan@acm.org>
34002 * tree.h (crc32_unsigned_n): Declare.
34003 (crc32_unsigned, crc32_unsigned): Make inline.
34004 * tree.c (crc32_unsigned_bits): Replace with ...
34005 (crc32_unsigned_n): ... this.
34006 (crc32_unsigned, crc32_byte): Remove.
34007 (crc32_string): Remove unnecessary braces.
34009 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
34011 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
34012 * ipa-inline-analysis.c (MAX_TIME): Remove.
34013 (account_size_time): Use sreal for time.
34014 (dump_inline_summary): Update.
34015 (estimate_function_body_sizes): Update.
34016 (estimate_edge_size_and_time): Update.
34017 (estimate_calls_size_and_time): Update.
34018 (estimate_node_size_and_time): Update.
34019 (inline_merge_summary): Update.
34020 (inline_update_overall_summary): Update.
34021 (estimate_time_after_inlining): Update.
34022 (inline_read_section): Update.
34023 (inline_write_summary): Update.
34024 * ipa-inline.c (compute_uninlined_call_time): Update.
34025 (compute_inlined_call_time): Update.
34026 (recursive_inlining): Update.
34027 (inline_small_functions): Update.
34028 (dump_overall_stats): Update.
34029 * ipa-inline.h: Include sreal.h.
34030 (size_time_entry): Turn time to sreal.
34031 (inline_summary): Turn self_time nad time to sreal.
34033 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
34035 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
34037 (sreal::stream_out, sreal::stream_in): New.
34038 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
34040 2017-04-25 Jakub Jelinek <jakub@redhat.com>
34042 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
34045 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
34048 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
34050 (dimode_scalar_chain::compute_convert_gain): Ditto.
34051 (dimode_scalar_chain::make_vector_copies): Ditto.
34052 (dimode_scalar_chain::convert_reg): Ditto.
34053 (dimode_scalar_chain::convert_insn): Ditto.
34054 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
34055 (VI248_AVX512BW_1): New mode iterator.
34056 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
34057 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
34060 2017-04-25 Martin Sebor <msebor@redhat.com>
34062 PR tree-optimization/80497
34063 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
34064 constants are representable in HOST_WIDE_INT.
34065 (parse_directive): Ditto.
34067 2017-04-25 Martin Sebor <msebor@redhat.com>
34070 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
34071 (new_zero_array): Adjust signature.
34072 (dom_info::dom_init): Used unsigned rather that size_t.
34073 (dom_info::dom_info): Same.
34075 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
34076 Jakub Jelinek <jakub@redhat.com>
34079 * config/arm/arm.c: Include gimple.h.
34080 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
34081 returns negative, increment ncrn only if it returned positive.
34082 (arm_needs_doubleword_align): Return int instead of bool,
34083 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
34084 members, but if there is any such non-FIELD_DECL
34085 > PARM_BOUNDARY aligned decl, return -1 instead of false.
34086 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
34087 returns negative, increment nregs only if it returned positive.
34088 (arm_setup_incoming_varargs): Likewise.
34089 (arm_function_arg_boundary): Emit -Wpsabi note if
34090 arm_needs_doubleword_align returns negative, return
34091 DOUBLEWORD_ALIGNMENT only if it returned positive.
34093 2017-04-25 Marek Polacek <polacek@redhat.com>
34096 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34097 first argument to type.
34099 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
34102 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
34103 type checks to test for compatibility instead of equality.
34105 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
34106 Jakub Jelinek <jakub@redhat.com>
34109 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
34111 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
34112 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
34113 the alignment computation, but return their maximum in warn_alignment.
34114 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
34115 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
34117 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
34118 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
34121 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34123 * config/arc/simdext.md (dmpyh): Fix typo.
34125 2017-04-25 Richard Biener <rguenther@suse.de>
34127 PR tree-optimization/80492
34128 * alias.c (compare_base_decls): Handle registers with asm
34129 specification conservatively.
34130 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
34131 compare_base_decls returning dont-know properly.
34133 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34135 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
34136 (legitimate_offset_address_p): New function.
34137 (arc_legitimate_address_p): Use above function.
34139 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34141 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
34143 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34145 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
34146 ACCH registers whenever they are available.
34148 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34150 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
34151 double regs fix when not used.
34153 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34155 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
34157 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
34158 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
34160 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34162 * config/arc/arc.c (arc_output_addsi): Check for h-register class
34163 when emitting short ADD instructions.
34165 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34167 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
34169 (cmpsi_cc_c_insn): Likewise.
34170 (cbranchsi4_scratch): Compute proper instruction length using
34171 compact_hreg_operand.
34172 * config/arc/predicates.md (compact_hreg_operand): New predicate.
34174 2017-04-25 Richard Biener <rguenther@suse.de>
34176 PR middle-end/80509
34177 * passes.c (pass_manager::pass_manager): Initialize
34178 m_name_to_pass_map.
34180 2017-04-25 Richard Biener <rguenther@suse.de>
34182 PR tree-optimization/79201
34183 * tree-ssa-sink.c (statement_sink_location): Handle calls.
34185 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34188 * config/s390/vector.md: Split MEM->GPR vector moves for
34189 non-s_operand addresses.
34191 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34194 * config/s390/predicates.md (reload_const_wide_int_operand): New
34196 * config/s390/s390.md ("movti"): Remove d/P alternative.
34197 ("movti_bigconst"): New pattern definition.
34199 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34202 * s390-protos.h (s390_expand_cs_hqi): Removed.
34203 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
34204 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
34205 modes as well as CCZ1mode and CCZmode.
34206 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
34207 signature of s390_emit_compare_and_swap.
34208 (s390_expand_cs_hqi): Likewise, make static.
34209 (s390_expand_cs_tdsi): Generate an explicit compare before trying
34210 compare-and-swap, in some cases.
34211 (s390_expand_cs): Wrapper function.
34212 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
34214 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
34215 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
34216 patterns for small and large integers. Forbid symref memory operands.
34217 Move expander to s390.c. Require cc register.
34218 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
34219 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
34220 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
34221 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
34222 symref memory operands. Remove CC mode and call s390_match_ccmode
34224 ("atomic_exchange<mode>"): Allow and implement all integer modes.
34226 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34228 * config/s390/s390.md (define_peephole2): New peephole to help
34229 combining the load-and-test pattern with volatile memory.
34231 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34233 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
34234 with CCZmode for TARGET_Z196.
34236 2017-04-25 Jakub Jelinek <jakub@redhat.com>
34238 PR rtl-optimization/80501
34239 * combine.c (make_compound_operation_int): Set subreg_code to SET
34240 even for AND with mask of the sign bit of mode.
34242 PR rtl-optimization/80500
34243 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
34244 sum's initial value.
34246 2017-04-25 Julian Brown <julian@codesourcery.com>
34247 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34249 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
34251 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
34253 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
34255 2017-04-25 Julian Brown <julian@codesourcery.com>
34256 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34258 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
34259 (thunderx2t99_sha): New Reservation.
34261 2017-04-25 Julian Brown <julian@codesourcery.com>
34262 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34264 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
34265 type for 1-element load.
34267 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
34269 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
34271 2017-04-24 Martin Jambor <mjambor@suse.cz>
34273 PR tree-optimization/80293
34274 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
34275 char arrays not totally scalarizable if it is false.
34276 (analyze_all_variable_accesses): Pass correct value in the new
34277 parameter. Add a statistics counter.
34279 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
34281 PR middle-end/79931
34282 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
34284 2017-04-24 Richard Biener <rguenther@suse.de>
34286 PR tree-optimization/80494
34287 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
34288 out for complex types.
34290 2017-04-24 Richard Biener <rguenther@suse.de>
34292 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
34293 * tree-ssa-sccvn.c (print_scc): Print SCC size.
34294 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
34295 (DFS): Adjust and never fail.
34296 (sccvn_dom_walker::fail): Remove.
34297 (sccvn_dom_walker::before_dom_children): Adjust.
34298 (run_scc_vn): Likewise and never fail.
34299 * tree-ssa-pre.c (pass_pre::execute): Adjust.
34300 (pass_fre::execute): Likewise.
34302 2017-04-24 Richard Biener <rguenther@suse.de>
34304 PR tree-optimization/79725
34305 * tree-ssa-sink.c (statement_sink_location): Return whether
34306 failure reason was zero uses. Move that check later.
34307 (sink_code_in_bb): Deal with zero uses by removing the stmt
34310 2017-04-24 Richard Biener <rguenther@suse.de>
34313 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
34314 pointer-based references.
34316 2017-04-24 Richard Biener <rguenther@suse.de>
34319 * pass_manager.h (pass_manager::operator new): Remove.
34320 (pass_manager::operator delete): Likewise.
34321 * passes.c (pass_manager::operator new): Remove.
34322 (pass_manager::operator delete): Likewise.
34323 (pass_manager::pass_manager): Zero individual pass members.
34325 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
34328 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
34329 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
34330 Check "XEXP (src, 1)" operand here.
34331 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
34332 Check "XEXP (src, 1)" operand here.
34333 (dimode_scalar_chain::make_vector_copies): Detect count register
34334 of a shift instruction. Zero extend count register from QImode
34335 to DImode to satisfy vector shift pattern count operand predicate.
34336 Substitute vector shift count operand with a DImode copy.
34337 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
34340 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
34342 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
34343 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34344 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
34345 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34346 (UNSPEC_NOREX_MEM): Remove definition.
34348 2017-04-21 Richard Biener <rguenther@suse.de>
34350 PR tree-optimization/79547
34351 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
34352 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
34353 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
34354 without any constraints.
34356 2017-04-21 Richard Biener <rguenther@suse.de>
34358 PR tree-optimization/78847
34359 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
34361 2017-04-21 Richard Biener <rguenther@suse.de>
34363 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
34364 (build_distinct_type_copy): Likewise.
34365 (build_variant_type_copy): Likewise.
34366 * tree.c (build_qualified_type): Pass down mem-stat info.
34367 (build_distinct_type_copy): Likewise.
34368 (build_variant_type_copy): Likewise.
34370 2017-04-21 Richard Biener <rguenther@suse.de>
34372 PR tree-optimization/80237
34373 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
34375 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
34376 for a simplified result.
34378 2016-04-21 Richard Biener <rguenther@suse.de>
34380 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
34381 sth as strict as a simple_iv but a chrec without symbols and an
34382 operand defined in the loop we are peeling (and not some subloop).
34383 (propagate_constants_for_unrolling): Propagate all constants.
34385 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34388 * config/i386/i386.c (print_reg): Remove assert for disalowed
34389 regno values, call output_operand_lossage instead.
34391 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34394 * config/i386/constraints.md (Yc): New register constraint.
34395 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
34396 Use Yc constraint for alternative 2 of operand 0. Remove
34397 preferred_for_speed attribute.
34399 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34401 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
34402 lastprivate clauses in SIMT case.
34404 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
34406 * doc/invoke.texi (-Wextra-semi): Document new warning option.
34408 2017-04-20 Richard Biener <rguenther@suse.de>
34410 PR tree-optimization/57796
34411 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
34412 as N scalar stores.
34413 (vect_model_load_cost): Cost gathers as N scalar loads.
34415 2017-04-20 Richard Biener <rguenther@suse.de>
34417 * ggc-page.c (ggc_allocated_p): Rename to ...
34418 (safe_lookup_page_table_entry): ... this and return the lookup
34420 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
34422 2017-04-20 Richard Biener <rguenther@suse.de>
34424 PR tree-optimization/80453
34425 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
34426 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
34427 from the conditions.
34428 (vn_phi_eq): Pass them down.
34429 (vn_phi_lookup): Record them.
34430 (vn_phi_insert): Likewise.
34432 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
34434 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
34435 uninitialized variable warning to avoid buffer overrun.
34437 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34440 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
34441 is suppressed for '{ 0 }' in C.
34443 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34445 * BASE-VER: Set to 8.0.0.
34447 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
34449 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
34450 priority .init_array and .fini_array section with SECTION_NOTYPE
34453 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34455 PR middle-end/80423
34456 * tree.h (build_array_type): Add typeless_storage default argument.
34457 * tree.c (type_cache_hasher::equal): Also compare
34458 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
34459 (build_array_type): Add typeless_storage argument, set
34460 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
34462 (build_nonshared_array_type): Adjust build_array_type_1 caller.
34463 (build_array_type): Likewise. Add typeless_storage argument.
34465 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34466 Jakub Jelinek <jakub@redhat.com>
34468 PR tree-optimization/80426
34469 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
34470 operation on symbolic operands, also compute the overflow for the
34471 invariant part when the operation degenerates into a negation.
34473 2017-04-19 Jakub Jelinek <jakub@redhat.com>
34476 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
34477 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
34480 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
34482 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
34485 * config/avr/avr.c (tree.h): Include it.
34486 (cgraph.h): Include it.
34487 (avr_encode_section_info): Don't warn for uninitialized progmem
34488 variable if it's just an alias.
34490 2017-04-19 Richard Biener <rguenther@suse.de>
34493 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
34494 when needed by AutoPGO.
34496 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
34499 * doc/lto.texi: Remove an extra 'that'.
34501 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
34503 PR rtl-optimization/80429
34504 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
34505 are only used in debug insns.
34507 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34508 Vladimir Makarov <vmakarov@redhat.com>
34510 * config/sparc/predicates.md (input_operand): Add comment. Return
34511 true for any memory operand when LRA is in progress.
34512 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
34514 2017-04-18 Jeff Law <law@redhat.com>
34517 * mips.md ({return,simple_return}_internal): Do not overwrite
34520 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34522 PR tree-optimization/80443
34523 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
34524 instead of adding 1, subtract -1 and similarly instead of subtracting
34527 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
34529 PR rtl-optimization/80357
34530 * haifa-sched.c (tmp_bitmap): New variable.
34531 (model_recompute): Handle duplicate use records.
34532 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
34533 (free_global_sched_pressure_data): Free it.
34535 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
34538 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34539 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34540 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34541 instead of SYSTEM_HEADER_DIR.
34543 2017-04-18 Jeff Law <law@redhat.com>
34545 PR middle-end/80422
34546 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
34547 predecessors after walking up the insn chain.
34549 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34552 * dwarf2out.c (modified_type_die): Try harder not to emit internal
34553 sizetype type into debug info.
34555 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34558 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
34559 unneeded test for TARGET_UPPER_REGS_SF.
34560 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
34562 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34565 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
34566 instead of gsi_after_labels.
34568 2017-04-18 Jeff Law <law@redhat.com>
34570 * regcprop.c (maybe_mode_change): Avoid creating copies of the
34574 2017-04-13 Jeff Law <law@redhat.com>
34575 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34576 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34578 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
34581 * config/avr/avr.c (intl.h): Include it.
34582 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
34584 2017-04-18 Martin Liska <mliska@suse.cz>
34586 PR gcov-profile/78783
34587 * gcov-tool.c (gcov_output_files): Validate that destination
34588 file is either removed by the tool or by a user.
34590 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34591 Guy Benyei <guybe@mellanox.com>
34593 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
34594 block, and do not negate it, the stored id is already negative.
34596 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34598 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
34600 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34603 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
34604 masks of options that should be turned off if the VSX vector
34605 options are turned off.
34606 (OTHER_P8_VECTOR_MASKS): Likewise.
34607 (OTHER_VSX_VECTOR_MASKS): Likewise.
34608 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
34609 rs6000_disable_incompatible_switches to validate no type switches
34611 (rs6000_incompatible_switch): New function to disallow turning on
34612 other vector options if -mno-vsx, -mno-power8-vector, or
34613 -mno-power9-vector are specified.
34615 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34617 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
34619 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34621 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
34622 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
34623 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
34624 (ARG_POINTER_CFA_OFFSET): Likewise.
34626 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34628 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
34629 conditions to take advantage of various optimizations.
34631 2017-04-13 Jeff Law <law@redhat.com>
34633 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34634 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34635 (zero_extendsidi2_dext): Likewise.
34637 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34640 * fold-const.c (fold_ternary_loc): Revert
34641 use op0 instead of fold_convert_loc (loc, type, arg0) part of
34644 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
34646 PR rtl-optimization/80343
34647 * lra-remat.c (update_scratch_ops): Assign original hard reg to
34648 new scratch pseudo.
34650 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
34653 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
34654 to ubsan_encode_value.
34656 2017-04-13 Jeff Law <law@redhat.com>
34658 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
34659 appearing in DEBUG_INSNs.
34661 2017-04-13 Martin Liska <mliska@suse.cz>
34663 PR gcov-profile/80413
34664 * gcov-io.c (gcov_write_string): Copy to buffer just when
34665 allocated size is greater than zero.
34667 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34670 * dwarf2out.c (decls_for_scope): Ignore declarations of
34671 current_function_decl in BLOCK_NONLOCALIZED_VARS.
34673 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
34676 * ipa-visibility.c (non_local_p): Fix typos.
34677 (localize_node): When localizing symbol in same comdat group,
34678 dissolve the group only when we know external symbols are going
34680 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
34682 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34684 PR tree-optimization/79390
34685 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
34686 order does not result in usable sequence, retry with reversed operand
34692 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
34693 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
34694 op0 instead of fold_convert_loc (loc, type, arg0).
34696 2017-04-12 Jeff Law <law@redhat.com>
34698 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
34699 has a delay slot in the generated code.
34701 * config/cris/cris.md (cris_preferred_reload_class): Return
34702 GENNONACR_REGS rather than GENERAL_REGS.
34704 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34707 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
34708 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
34709 signedness of the result type.
34711 2017-04-12 Richard Biener <rguenther@suse.de>
34712 Jeff Law <law@redhat.com>
34714 PR tree-optimization/80359
34715 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
34716 trim stores to TARGET_MEM_REFs.
34718 2017-04-12 Richard Biener <rguenther@suse.de>
34720 PR tree-optimization/79390
34721 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
34722 threading case even more.
34724 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
34727 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
34728 for quad_address_p for TImode, instead of just not indexed_address.
34730 2017-04-12 Richard Biener <rguenther@suse.de>
34731 Bernd Edlinger <bernd.edlinger@hotmail.de>
34733 PR middle-end/79671
34734 * alias.c (component_uses_parent_alias_set_from): Handle
34735 TYPE_TYPELESS_STORAGE.
34736 (get_alias_set): Likewise.
34737 * tree-core.h (tree_type_common): Add typeless_storage flag.
34738 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
34739 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
34740 for types containing members with TYPE_TYPELESS_STORAGE.
34741 (place_field): Likewise.
34742 (layout_type): Likewise for ARRAY_TYPE.
34743 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
34744 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
34745 TYPE_TYPELESS_STORAGE.
34746 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
34748 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34751 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34752 first argument to type.
34754 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34758 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
34759 CONST0_RTX (mode) rather than const0_rtx where appropriate.
34760 (rs6000_expand_binop_builtin): Likewise.
34761 (rs6000_expand_ternop_builtin): Likewise; also add missing
34762 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
34763 vshasigma built-ins.
34764 * doc/extend.texi: Document that vec_xxpermdi's third argument
34765 must be a constant.
34767 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
34769 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
34770 Use shift_const cost parameter when calculating gain of STV shifts.
34772 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
34774 PR rtl-optimization/70478
34775 * lra-constraints.c (process_alt_operands): Check memory for
34776 disfavoring memory insn operand.
34778 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34780 PR middle-end/80100
34781 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
34782 left shift in unsigned HOST_WIDE_INT type.
34784 PR rtl-optimization/80385
34785 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
34786 (not (neg X)) into (plus X -1) for complex or non-integral modes.
34789 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
34790 if they have any depend clauses.
34792 2017-04-11 Martin Liska <mliska@suse.cz>
34795 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
34796 * ipa-split.c (split_function): Create a local comdat symbol
34797 if caller is in a comdat group.
34799 2017-04-11 Martin Liska <mliska@suse.cz>
34802 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
34805 2017-04-11 Martin Sebor <msebor@redhat.com>
34807 PR middle-end/80364
34808 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
34809 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
34811 (directive::set_width, directive::set_precision, format_character):
34813 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
34816 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
34819 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
34820 conflict, set target->arch_name instead of target->cpu_name.
34822 2017-04-11 Richard Biener <rguenther@suse.de>
34824 PR tree-optimization/80374
34825 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
34826 build_zero_cst, remove fold_convertible_p check again.
34828 2017-04-11 Martin Liska <mliska@suse.cz>
34831 * ubsan.c (instrument_object_size): Do not instrument register
34834 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34837 * config/i386/i386-builtin-types.def
34838 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
34839 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
34840 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
34841 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
34842 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
34843 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
34844 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
34845 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
34846 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
34847 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
34848 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34849 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34850 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34851 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34852 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34853 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34854 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34855 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34856 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34857 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34858 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34859 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34860 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34861 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34862 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34863 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34864 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34865 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34866 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34867 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34868 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34869 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34870 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34871 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34872 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34873 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34874 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34875 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34876 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34877 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34878 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34879 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34880 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34881 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34882 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34883 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34885 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34886 flag to second_arg_count, handle 4 argument function type _COUNT
34887 aliases, handle second_arg_count on second argument rather than last.
34889 2017-04-10 Jeff Law <law@redhat.com>
34891 PR tree-optimization/80374
34892 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34893 record anything if we can not convert integer_zero_node to the
34896 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34899 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34900 Enhance special handling given to the TARGET_P9_MINMAX option in
34901 relation to certain other options.
34903 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34905 PR tree-optimization/80153
34906 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34907 remove POINTER_PLUS_EXPR's base part directly, rather than through
34910 2017-04-10 Richard Biener <rguenther@suse.de>
34911 Bin Cheng <bin.cheng@arm.com>
34913 PR tree-optimization/80153
34914 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34915 the first element of pointer type aff_tree. Build result expr in
34917 (add_elt_to_tree): Convert to type unconditionally. Remove other
34918 fold_convert calls.
34919 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34920 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34922 2017-04-10 Richard Biener <rguenther@suse.de>
34924 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34927 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34929 PR rtl-optimization/70478
34930 * lra-constraints.c (curr_small_class_check): New.
34931 (update_and_check_small_class_inputs): New.
34932 (process_alt_operands): Update curr_small_class_check. Disfavor
34933 alternative insn memory operands. Check available regs for small
34936 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34939 * config/mips/mips.opt (-mvirt): Update description.
34940 * doc/invoke.texi (-mvirt): Likewise.
34942 2017-04-10 Richard Biener <rguenther@suse.de>
34944 PR middle-end/80362
34945 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34946 looking for NEGATE_EXPR in -A / -B to A / B folding.
34948 2017-04-10 Martin Liska <mliska@suse.cz>
34950 PR gcov-profile/80224
34951 * gcov.c (print_usage): Fix usage string.
34952 (get_gcov_intermediate_filename): Remove.
34953 (output_gcov_file): Use both for normal and intermediate format.
34954 (generate_results): Do not initialize special file for
34955 intermediate format.
34957 2017-04-10 Richard Biener <rguenther@suse.de>
34959 PR tree-optimization/80304
34960 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34963 2017-04-10 Nathan Sidwell <nathan@acm.org>
34966 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34967 (rs6000_init_builtins): Use it.
34969 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34971 * config/arm/arm.md (<mrc>): Add mode to SET source.
34972 (<mrrc>): Likewise.
34974 2017-04-10 Richard Biener <rguenther@suse.de>
34976 PR middle-end/80344
34977 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34979 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34982 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34983 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34984 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34985 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34986 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34987 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34988 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34989 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34990 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34991 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34992 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34993 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34994 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34995 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34996 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34997 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34998 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34999 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
35000 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
35001 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
35002 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
35003 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
35004 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
35006 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
35008 PR rtl-optimization/70478
35009 * lra-constraints.c: Reverse the last patch.
35011 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
35013 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
35014 Add comment for WCHAR_T.
35016 2017-04-08 Martin Liska <mliska@suse.cz>
35019 2017-04-07 Martin Liska <mliska@suse.cz>
35022 * ipa-split.c (split_function): Add function part to a same comdat
35025 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35028 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
35030 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
35032 * rs6000/rs6000.c (vec_load_pendulum): Rename...
35033 (vec_pairing): ...to this.
35034 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
35035 (rs6000_sched_init): Adjust for name change.
35036 (struct rs6000_sched_context): Likewise.
35037 (rs6000_init_sched_context): Likewise.
35038 (rs6000_set_sched_context): Likewise.
35040 2017-04-07 Jakub Jelinek <jakub@redhat.com>
35046 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
35048 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
35049 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
35050 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
35052 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
35054 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
35056 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
35058 PR rtl-optimization/70703
35059 * ira-color.c (update_conflict_hard_regno_costs): Use
35060 int64_t instead of HOST_WIDE_INT.
35062 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
35064 PR rtl-optimization/70478
35065 * lra-constraints.c (process_alt_operands): Disfavor alternative
35066 insn memory operands.
35068 2017-04-07 Jeff Law <law@redhat.com>
35070 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
35071 CALL and NOTE_INSN_CALL_ARG_LOCATION.
35073 2017-04-07 Martin Liska <mliska@suse.cz>
35076 * config/aarch64/aarch64.c (aarch64_process_target_attr):
35077 Show error message instead of an ICE.
35079 2017-04-07 Martin Liska <mliska@suse.cz>
35082 * ipa-split.c (split_function): Add function part to a same comdat
35085 2017-04-07 Richard Biener <rguenther@suse.de>
35087 PR middle-end/80341
35088 * tree.c (get_unwidened): Also handle ! for_type case for
35090 * convert.c (do_narrow): Split out from ...
35091 (convert_to_integer_1): ... here. Do not pass final truncation
35092 type to get_unwidened for TRUNC_DIV_EXPR.
35094 2017-04-07 Richard Biener <rguenther@suse.de>
35096 * tree-affine.c (wide_int_ext_for_comb): Take type rather
35098 (aff_combination_const): Adjust.
35099 (aff_combination_scale): Likewise.
35100 (aff_combination_add_elt): Likewise.
35101 (aff_combination_add_cst): Likewise.
35102 (aff_combination_convert): Likewise.
35103 (add_elt_to_tree): Likewise. Remove unused argument.
35104 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
35106 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
35108 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
35110 * config/arm/arm.c (arm_default_short_enums): Use
35111 ARM_DEFAULT_SHORT_ENUMS.
35112 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
35114 2017-04-06 Jakub Jelinek <jakub@redhat.com>
35117 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
35118 members with redundant out-of-class redeclaration.
35120 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
35123 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
35124 * config/i386/i386.md (*zero_extendsidi2):
35125 Add (?*x,*x) and (?*v,*v) alternatives.
35127 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
35130 * config/i386/i386.c (ix86_expand_builtin)
35131 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
35132 mode from insn data. Convert operands to insn operand mode.
35133 Copy operands that don't satisfy insn predicate to a register.
35135 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
35137 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
35140 2017-04-06 Richard Biener <rguenther@suse.de>
35142 PR tree-optimization/80334
35143 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
35144 preserve alignment of accesses.
35146 2017-04-06 Richard Biener <rguenther@suse.de>
35148 PR tree-optimization/80262
35149 * tree-sra.c (build_ref_for_offset): Preserve address-space
35151 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
35152 Drop useless address-space information on MEM_REF offsets.
35154 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
35156 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
35158 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
35160 PR rtl-optimization/70703
35161 * ira-color.c (update_conflict_hard_regno_costs): Use
35162 HOST_WIDE_INT instead of long.
35164 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
35167 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
35168 not defined for x86_64 target. Add -mmmx target option when __SSE2__
35170 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
35171 for x86_64 target. Handle -m3dnowa option.
35173 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
35175 PR rtl-optimization/70703
35176 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
35177 (update_conflict_hard_regno_costs): Use long instead of unsigned
35178 arithmetic for cost calculation.
35180 2017-04-05 Jakub Jelinek <jakub@redhat.com>
35181 Bernd Edlinger <bernd.edlinger@hotmail.de>
35184 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
35187 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
35190 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
35191 ptr_mode with Pmode throughout.
35192 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
35193 into probe_stack_range and use DImode.
35195 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
35198 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
35199 call_eh_return is true.
35201 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35203 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
35204 Initialize last_match_fntype_index.
35206 2017-04-05 Jakub Jelinek <jakub@redhat.com>
35209 * tree-nvr.c: Include internal-fn.h.
35210 (pass_return_slot::execute): Ignore internal calls without
35213 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35214 Richard Biener <rguenther@suse.de>
35217 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
35218 captures used multiple times, except for the last use.
35219 * generic-match-head.c: Include gimplify.h.
35221 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35223 PR tree-optimization/79390
35224 * target.h (struct noce_if_info): Declare.
35225 * targhooks.h (default_noce_conversion_profitable_p): Declare.
35226 * target.def (noce_conversion_profitable_p): New target hook.
35227 * ifcvt.h (struct noce_if_info): New type, moved from ...
35228 * ifcvt.c (struct noce_if_info): ... here.
35229 (noce_conversion_profitable_p): Renamed to ...
35230 (default_noce_conversion_profitable_p): ... this. No longer
35232 (noce_try_store_flag_constants, noce_try_addcc,
35233 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
35234 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
35235 instead of noce_conversion_profitable_p.
35236 * config/i386/i386.c: Include ifcvt.h.
35237 (ix86_option_override_internal): Don't override
35238 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
35239 (ix86_noce_conversion_profitable_p): New function.
35240 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
35241 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
35242 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
35243 * doc/tm.texi: Regenerated.
35245 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35247 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
35250 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
35253 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
35254 instructions for small multiply cores.
35256 2017-04-04 Jeff Law <law@redhat.com>
35258 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
35260 (mips_expand_vec_perm_const): Initialize elements in orig_perm
35261 that are not set by the loop over the elements.
35263 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35266 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
35267 int mode, convert_modes it to mode as unsigned, otherwise use
35268 lowpart_subreg to mode rather than SImode.
35269 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
35270 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
35271 Use DImode instead of SImode for the shift count operand.
35272 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
35275 2017-04-04 Richard Biener <rguenther@suse.de>
35277 PR middle-end/80281
35278 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
35279 arithmetic done for the negate or the plus. Simplify.
35280 (A - (-B) -> A + B): Likewise.
35281 * fold-const.c (split_tree): Make sure to not negate pointers.
35283 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
35285 PR rtl-optimization/60818
35286 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
35287 a compare of comparisons with the thing compared if this results
35288 in a different machine mode.
35290 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
35292 * alias.c (base_alias_check): Fix typo in comment.
35293 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
35294 * cgraphunit.c (symbol_table::compile): Likewise.
35295 * collect2.c (maybe_run_lto_and_relink): Likewise.
35296 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
35297 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
35298 * config/avr/avr.c (avr_map_op_t): Likewise.
35299 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
35300 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
35301 * config/epiphany/epiphany.md (movcc): Likewise.
35302 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
35303 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
35305 * config/mips/mips.c (mips_save_restore_reg): Likewise.
35306 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
35307 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
35308 * config/sh/sh.c (sh_rtx_costs): Likewise.
35309 * fold-const.c (fold_truth_andor): Likewise.
35310 * genautomata.c (collapse_flag): Likewise.
35311 * gengtype.h (struct type::u::s): Likewise.
35312 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
35313 * input.c (FORMAT_AMOUNT): Likewise.
35314 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
35315 (known_aggs_to_agg_replacement_list): Likewise.
35316 * ipa-inline-analysis.c: Likewise.
35317 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
35318 * ipa-polymorphic-call.c
35319 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
35320 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
35321 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
35323 * modulo-sched.c (apply_reg_moves): Likewise.
35324 * omp-expand.c (build_omp_regions_1): Likewise.
35325 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
35326 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
35327 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
35328 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
35329 * value-prof.c: Likewise.
35330 * var-tracking.c (val_reset): Likewise.
35332 2017-04-03 Richard Biener <rguenther@suse.de>
35334 PR tree-optimization/80275
35335 * fold-const.c (split_address_to_core_and_offset): Handle
35338 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
35340 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
35341 descriptors is at least equal to that of functions.
35343 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35345 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
35347 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35350 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
35351 (mov<IMOD4:mode>): New expander.
35352 (*mov<IMOD4:mode>_internal): New insn and split pattern.
35354 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
35356 PR rtl-optimization/79405
35357 * fwprop.c (propagations_left): New variable.
35358 (forward_propagate_into): Decrement it.
35359 (fwprop_init): Initialize it.
35360 (fw_prop): If the variable has reached zero, stop propagating.
35361 (fwprop_addr): Ditto.
35363 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35366 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
35367 a FUNCTION_DECL, pass it as decl instead of origin to
35370 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
35372 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
35375 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
35378 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
35379 TARGET_VSX_SMALL_INTEGER.
35381 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35383 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
35384 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
35386 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
35388 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
35389 extraction from odd-numbered MSA register.
35391 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35393 PR middle-end/80173
35394 * expmed.c (store_bit_field_1): Don't attempt to create
35395 a word subreg out of hard registers wider than word if they
35396 have HARD_REGNO_NREGS of 1 for their mode.
35398 PR middle-end/80163
35399 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
35400 conversions to integer types wider than word and pointer.
35403 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
35404 (rtx_equal_for_cselib_p): Pass 0 to it.
35405 * cselib.c (cselib_hasher::equal): Likewise.
35406 (rtx_equal_for_cselib_1): Add depth argument. If depth
35407 is 128, don't look up VALUE locs and punt. Increment
35408 depth in recursive calls when walking VALUE locs.
35410 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
35412 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
35413 (make_gcov_file_name): Use the canonical path name for generating
35415 (read_line): Fix handling of files with ascii null bytes.
35417 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
35419 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
35420 to initialise a vector register instead
35421 of using a const_int.
35423 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35425 PR translation/80189
35426 * gimplify.c (omp_default_clause): Use %qs instead of %s in
35427 diagnostic messages.
35429 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
35432 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
35433 (dfp_diex_<mode>): Update mode of operand 1.
35434 * doc/extend.texi (dxex, dxexq): Document change to return type.
35435 (diex, diexq): Document change to argument type.
35437 2017-03-30 Martin Jambor <mjambor@suse.cz>
35440 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
35441 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
35442 it reflects the signature changes performed at the callee side.
35443 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
35444 to cgraph_build_function_type_skip_args.
35445 (build_function_decl_skip_args): Adjust call to the above function.
35447 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35450 * config/i386/sse.md
35451 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
35452 register as dest whenever it is a MEM not rtx_equal_p to the
35453 corresponding dup operand, and when forcing into reg move the
35454 reg into the memory afterwards.
35455 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
35456 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
35457 for the force_reg mode.
35458 (avx512vl_vextractf128<mode>): Use register as dest either
35459 always when a MEM, or when it is a MEM not rtx_equal_p to the
35460 corresponding dup operand, or even not when it is a CONST_VECTOR
35461 depending on the mode and lo vs. hi.
35462 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
35464 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
35465 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
35466 Likewise. Require that operands[2] is even.
35467 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
35468 Remove extraneous parens. Require that operands[2] is a multiple
35470 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
35471 operands[0] is a MEM if <mask_applied>, the predicates/constraints
35472 disallow memory then.
35474 2017-03-30 Richard Biener <rguenther@suse.de>
35476 PR tree-optimization/77498
35477 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
35478 to non-constants over backedges.
35480 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
35482 PR rtl-optimization/80233
35483 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
35484 as last_combined_insn. Do not test for BARRIER_P separately.
35486 2017-03-29 Andreas Schwab <schwab@suse.de>
35489 * calls.c (prepare_call_address): Convert funexp to Pmode before
35490 copying to temp reg.
35492 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35494 PR tree-optimization/80158
35495 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
35496 Handle possible future case of more than one alternate
35498 (replace_rhs_if_not_dup): Likewise.
35499 (replace_one_candidate): Likewise.
35501 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
35503 PR rtl-optimization/80193
35504 * ira.c (ira): Do not check allocation for LRA.
35506 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
35508 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
35509 (nvptx_output_simt_exit): Declare.
35510 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
35511 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
35512 (init_softstack_frame): Move initialization of crtl->is_leaf to...
35513 (nvptx_declare_function_name): ...here. Emit declaration of local
35514 memory space buffer for omp_simt_enter insn.
35515 (nvptx_output_unisimt_switch): New.
35516 (nvptx_output_softstack_switch): New.
35517 (nvptx_output_simt_enter): New.
35518 (nvptx_output_simt_exit): New.
35519 * config/nvptx/nvptx.h (struct machine_function): New fields
35520 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
35521 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
35522 (UNSPECV_SIMT_EXIT): Ditto.
35523 (omp_simt_enter_insn): New insn.
35524 (omp_simt_enter): New expansion.
35525 (omp_simt_exit): New insn.
35526 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
35528 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
35529 (expand_GOMP_SIMT_ENTER_ALLOC): New.
35530 (expand_GOMP_SIMT_EXIT): New.
35531 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
35532 (GOMP_SIMT_ENTER_ALLOC): Ditto.
35533 (GOMP_SIMT_EXIT): Ditto.
35534 * target-insns.def (omp_simt_enter): New insn.
35535 (omp_simt_exit): Ditto.
35536 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
35538 (lower_rec_simd_input_clauses): Implement SIMT privatization.
35539 (lower_rec_input_clauses): Likewise.
35540 (lower_lastprivate_clauses): Handle SIMT privatization.
35542 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
35543 (ompdevlow_adjust_simt_enter): New.
35544 (find_simtpriv_var_op): New.
35545 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
35546 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
35548 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
35549 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
35550 (copy_decl_for_dup_finish): Ditto.
35552 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
35554 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
35557 * config/i386/i386.c (ix86_option_override_internal): Always
35558 allow -mpreferred-stack-boundary=3 for 64-bit targets.
35560 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35562 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
35564 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35566 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
35567 mark new edge's irreducible flag accordign to it.
35568 (vect_do_peeling): Check loop preheader edge's irreducible flag
35569 and pass it to function slpeel_add_loop_guard.
35571 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
35573 PR tree-optimization/80218
35574 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35575 Update block frequencies and counts.
35577 2017-03-28 Richard Biener <rguenther@suse.de>
35579 PR tree-optimization/78644
35580 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
35581 of a simplification result we may not use it at all.
35583 2017-03-28 Richard Biener <rguenther@suse.de>
35586 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
35587 without arguments, generate default definition of a SSA name.
35589 2017-03-28 Richard Biener <rguenther@suse.de>
35591 PR middle-end/80222
35592 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
35593 TYPE_REF_CAN_ALIAS_ALL references.
35594 * fold-const.c (fold_indirect_ref_1): Likewise.
35596 2017-03-28 Martin Liska <mliska@suse.cz>
35599 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
35600 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
35602 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35603 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
35605 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
35606 (EXTRA_SPECS): Define.
35607 (SUBTARGET_EXTRA_SPECS): Likewise.
35608 (SUBTARGET_CPP_SPEC): Likewise.
35609 * config/arc/elf.h (EXTRA_SPECS): Renamed to
35610 SUBTARGET_EXTRA_SPECS.
35611 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
35613 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35615 * config/arc/simdext.md (vst64_insn): Update pattern.
35616 (vld32wh_insn): Likewise.
35617 (vld32wl_insn): Likewise.
35618 (vld64_insn): Likewise.
35619 (vld32_insn): Likewise.
35621 2017-03-28 Marek Polacek <polacek@redhat.com>
35624 * fold-const.c (fold_comparison): Use protected_set_expr_location
35625 instead of SET_EXPR_LOCATION.
35627 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
35629 * tree.c (add_expr): Avoid name lookup warning.
35631 2017-03-27 Jeff Law <law@redhat.com>
35633 PR tree-optimization/80216
35634 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
35635 function name. Limit recursion depth.
35636 (record_temporary_equivalences): Corresponding changes.
35638 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
35640 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
35643 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35646 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
35648 * cfgcleanup.c (reg_note_cfa_p): New array.
35649 (insns_have_identical_cfa_notes): New function.
35650 (old_insns_match_p): Don't cross-jump in between /f
35651 and non-/f instructions. If both i1 and i2 are frame related,
35652 verify all CFA notes, their order and content.
35654 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
35657 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
35658 HImode and SImode with zero extend to DImode to one insn.
35659 (bswap<mode>2_extenddi): Likewise.
35660 (bswapsi2_extenddi): Likewise.
35661 (bswaphi2_extendsi): Likewise.
35662 (bswaphi2): Combine bswap HImode and SImode into one insn.
35663 Separate memory insns from swapping register.
35664 (bswapsi2): Likewise.
35665 (bswap<mode>2): Likewise.
35666 (bswaphi2_internal): Delete, no longer used.
35667 (bswapsi2_internal): Likewise.
35668 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
35669 store, and gpr<-gpr swap insns.
35670 (bswap<mode>2_store): Likewise.
35671 (bswaphi2_reg): Register only splitter, combine with the splitter.
35672 (bswaphi2 splitter): Likewise.
35673 (bswapsi2_reg): Likewise.
35674 (bswapsi2 splitter): Likewise.
35675 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
35676 the insns into load, store, and register/register insns.
35677 (bswapdi2_ldbrx): Likewise.
35678 (bswapdi2_load): Likewise.
35679 (bswapdi2_store): Likewise.
35680 (bswapdi2_reg): Likewise.
35682 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
35684 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
35685 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
35687 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
35690 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
35692 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35693 special handling for target option conflicts between dform
35694 options (-mpower9-dform, -mpower9-dform-vector,
35695 -mpower9-dform-scalar) and -mno-direct-move.
35697 2017-03-27 Richard Biener <rguenther@suse.de>
35699 PR tree-optimization/80181
35700 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
35702 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35704 * config/arc/predicates.md (move_double_src_operand): Replace the
35705 call to move_double_src_operand with a call to address_operand.
35707 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35709 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
35710 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
35711 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
35713 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35715 * config/arc/predicates.md (long_immediate_loadstore_operand):
35716 Consider scaled addresses cases.
35718 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35720 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
35721 restored when in interrupt.
35722 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
35723 doesn't have delay slot.
35725 2017-03-27 Richard Biener <rguenther@suse.de>
35728 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
35729 inlined thunk clones.
35731 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35734 * asan.c (instrument_derefs): Copy over last operand from
35735 original COMPONENT_REF to the new COMPONENT_REF with
35736 DECL_BIT_FIELD_REPRESENTATIVE.
35737 * ubsan.c (instrument_object_size): Likewise.
35739 2017-03-27 Richard Biener <rguenther@suse.de>
35741 PR tree-optimization/80170
35742 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
35743 sure DR/SCEV didnt fold in constants we do not see when looking
35744 at the reference base alignment.
35746 2017-03-27 Richard Biener <rguenther@suse.de>
35748 PR middle-end/80171
35749 * gimple-fold.c (fold_ctor_reference): Properly guard against
35750 NULL return value from canonicalize_constructor_val.
35752 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
35755 * config/i386/i386.c (ix86_expand_builtin)
35756 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
35757 flags reg setting and flags reg using instructions.
35758 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
35759 clobbering instructions to zero extend op2.
35761 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
35763 * doc/install.texi (Configuration) <--with-aix-soname>:
35764 Update link to AIX ld.
35766 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
35768 PR rtl-optimization/80160
35769 PR rtl-optimization/80159
35770 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
35771 reg_alternate_class into account.
35773 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
35776 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
35777 to consider in curr_insn_transform.
35779 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35781 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
35782 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
35783 and emit_mode_inner.
35785 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35787 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
35788 argument to the overloaded builtin variants. Use the new flag to
35789 deprecate certain builtin variants.
35790 * config/s390/s390-builtin-types.def: Add new builtin types.
35791 * config/s390/s390-builtins.h: Support new flags field for
35792 overloaded builtins.
35793 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
35794 (s390_macro_to_expand): Enable vector float data type.
35795 (s390_cpu_cpp_builtins_internal): Indicate support of the new
35796 builtins by incrementing the __VEC__ version number.
35797 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
35799 (s390_resolve_overloaded_builtin): Emit error messages depending
35800 on the builtin flags.
35801 * config/s390/s390.c (s390_expand_builtin): Support additional
35802 flags argument. Change error message to match the messages
35803 emitted in s390-c.c.
35804 * config/s390/s390.md: New UNSPEC_* constants.
35805 (op_type): Add new instruction types.
35806 * config/s390/vecintrin.h: Add new builtins and test data class
35808 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
35809 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
35810 (VEC_INEXACT, VEC_NOINEXACT): New constants.
35811 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
35812 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
35813 ("vec_mergel<mode>"): V_HW -> VEC_HW.
35815 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
35816 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
35817 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
35818 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
35820 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
35821 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
35822 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
35823 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
35825 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
35826 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
35827 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
35828 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
35829 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
35830 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
35831 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
35833 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
35834 ("vec_scatter_element<V_HW_4:mode>_DI")
35835 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
35836 ("vec_fpint<mode>", "vflls")
35837 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
35838 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
35839 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
35840 ("*vec_cmphe<mode>_cc"): ... these.
35842 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
35843 mode constant instead of magic value.
35845 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35847 * config/s390/s390.c (s390_expand_vec_compare): Support other
35848 vector floating point modes than just V2DF.
35849 (s390_expand_vcond): Likewise.
35850 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35851 (s390_cannot_change_mode_class): Prevent mode changes between TF
35852 and V1TF in vector registers.
35853 * config/s390/s390.md (DF, SF): New mode attributes.
35854 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35855 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35856 SFmode support for VRs.
35857 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35859 (VFT, VF_HW): New mode iterators.
35860 (vw, sdx): New mode attributes.
35861 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35862 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35863 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35864 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35865 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35866 also the new vector floating point modes. Renaming to ...
35868 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35869 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35870 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35871 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35872 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35873 ("vec_unordered<mode>"): ... these.
35875 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35876 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35877 ("*vec_extendv2df"): New insn definitions.
35879 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35881 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35882 ("mulditi3_2", "*muldi3_sign"): New patterns.
35883 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35884 rename the pattern definition.
35886 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35888 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35890 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35892 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35894 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35895 instruction if possible.
35896 * config/s390/vector.md (vec_halfnumelts): New mode
35898 ("*vec_vllezlf<mode>"): New pattern.
35900 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35902 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35903 ("popcountv4si2", "popcountv2di2"): Rename to ...
35904 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35905 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35907 ("popcount<mode>2_vxe"): New pattern.
35909 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35911 * common/config/s390/s390-common.c (processor_flags_table): Add
35913 * config.gcc: Add arch12.
35914 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35915 Default to arch12 for unknown CPU model numbers.
35916 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35917 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35918 PROCESSOR_max sanity check.
35919 * config/s390/s390-opts.h (enum processor_type): Add
35921 * config/s390/s390.c (processor_table): Add arch12.
35922 (s390_expand_builtin): Add check for B_VXE flag.
35923 (s390_issue_rate): Add PROCESSOR_ARCH12.
35924 (s390_get_sched_attrmask): Likewise.
35925 (s390_get_unit_mask): Likewise.
35926 (s390_sched_score): Enable z13 scheduling for arch12.
35927 (s390_sched_reorder): Likewise.
35928 (s390_sched_variable_issue): Likewise.
35929 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35931 (s390_tune_attr): Use z13 scheduling also for arch12.
35932 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35933 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35934 (TARGET_VXE_P): New macros.
35935 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35936 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35937 * config/s390/s390.opt: Add arch12 as processor_type.
35939 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35941 * config/s390/s390.md
35942 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35943 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35944 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35946 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35947 Rename expanders to ...
35949 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35950 ("fixuns_truncdddi2_emu"): ... these.
35952 ("fixuns_trunc<mode>si2_emu"): New expander.
35954 ("*fixuns_truncdfdi2_z13"): Rename to ...
35955 ("*fixuns_truncdfdi2_vx"): ... this.
35957 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35959 * config/s390/2964.md: Remove the single element vector compare
35960 instructions which are no longer used.
35961 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35963 (s390_canonicalize_comparison): Remove handling of DFmode
35965 (s390_expand_vec_compare_scalar): Remove function.
35966 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35967 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35969 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35971 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35973 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35975 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35976 will anyway by matched by mov<mode>_64dfp.
35978 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35980 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35981 vlef/vstef. Add missing operand to vleif.
35983 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35985 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35986 pair for all vector types with 64 bit elements.
35987 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35988 * config/s390/vector.md (V_HW_64): ... here.
35989 (V_128_NOSINGLE): New mode iterator.
35990 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35991 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35992 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35993 ("*vec_load_pairv2di"): Change to ...
35994 ("*vec_load_pair<mode>"): ... this one.
35996 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35998 * config/s390/constraints.md: Add comments.
35999 (jKK): Reject element sizes > 8 bytes.
36000 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
36002 * config/s390/s390.md: Add the s_operand checks formerly in
36003 s390_split_ok_p to various splitters where they are still
36005 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
36006 for 128 bit vectors. Plus two splitters.
36008 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36010 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
36013 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36016 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
36017 error if the boundary argument is not constant.
36019 2017-03-24 Jakub Jelinek <jakub@redhat.com>
36021 PR rtl-optimization/80112
36022 * loop-doloop.c (doloop_condition_get): Don't check condition
36023 if cmp isn't SET with IF_THEN_ELSE src.
36025 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36027 PR tree-optimization/80158
36028 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
36029 replacing a candidate statement, also replace it for the
36030 candidate's alternate interpretation.
36031 (replace_rhs_if_not_dup): Likewise.
36032 (replace_one_candidate): Likewise.
36034 2017-03-24 Richard Biener <rguenther@suse.de>
36036 PR tree-optimization/80167
36037 * graphite-isl-ast-to-gimple.c
36038 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
36040 (translate_isl_ast_to_gimple::get_rename): Likewise.
36042 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
36044 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
36045 handling of certain combinations of target options, including the
36046 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
36047 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
36049 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36052 * config/arm/arm.md (*load_multiple): Add reload_completed to
36053 matching condition.
36055 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36056 Richard Biener <rguenth@suse.de>
36058 PR tree-optimization/79908
36059 PR tree-optimization/80136
36060 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
36061 been cast away, gimplify_and_add suffices.
36063 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
36065 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
36067 2017-03-23 Richard Biener <rguenther@suse.de>
36069 PR tree-optimization/80032
36070 * gimplify.c (gimple_push_cleanup): Forced unconditional
36071 cleanups still have to go to the conditional_cleanups
36074 2017-03-22 Jakub Jelinek <jakub@redhat.com>
36076 PR tree-optimization/80072
36077 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
36079 (next_operand_entry_id): Change type to unsigned int.
36080 (sort_by_operand_rank): Make sure to return the right return value
36081 even if unsigned fields are bigger than INT_MAX.
36082 (struct oecount): Change cnt and id type to unsigned int.
36083 (oecount_hasher::equal): Formatting fix.
36084 (oecount_cmp): Make sure to return the right return value
36085 even if unsigned fields are bigger than INT_MAX.
36086 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
36089 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
36090 TREE_READONLY on result if writing it more than once.
36093 * doc/invoke.texi (-fsanitize=thread): Document that with
36094 -fnon-call-exceptions atomics are not able to throw
36098 * tsan.c: Include tree-eh.h.
36099 (instrument_builtin_call): Call maybe_clean_eh_stmt or
36100 maybe_clean_or_replace_eh_stmt where needed.
36101 (instrument_memory_accesses): Add cfg_changed argument.
36102 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
36103 if it returned true.
36104 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
36106 PR rtl-optimization/63191
36107 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
36108 wrapper function, moved the whole old content into ...
36109 (ix86_delegitimize_address_1): ... this. New inline function.
36110 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
36111 true as last argument instead of ix86_delegitimize_address.
36113 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
36115 * config/aarch64/aarch64.c (generic_branch_cost): Copy
36116 cortexa57_branch_cost.
36118 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
36120 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
36122 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36125 * doc/md.texi (Constraints): Document wA constraint.
36126 * config/rs6000/constraints.md (wA): New.
36127 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
36128 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
36129 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
36130 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
36132 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
36135 * gimplify.c (is_oacc_declared): New function.
36136 (oacc_default_clause): Use it to set default flags for acc declared
36137 variables inside parallel regions.
36138 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
36139 declared variables.
36140 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
36141 declare attribute to any decl as necessary.
36143 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
36146 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
36147 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
36148 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
36149 (arm_arch_lpae): This.
36150 * config/arm/arm.c (arm_arch7ve): Rename into ...
36151 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
36152 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
36155 2017-03-22 Martin Liska <mliska@suse.cz>
36158 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
36159 error message instead of an ICE.
36161 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36163 * doc/extend.texi (6.11 Additional Floating Types): Revise.
36165 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
36167 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36169 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
36172 2017-03-21 Martin Sebor <msebor@redhat.com>
36174 * doc/extend.texi: Use "cannot" instead of "can't."
36175 * doc/hostconfig.texi: Same.
36176 * doc/install.texi: Same.
36177 * doc/invoke.texi: Same.
36178 * doc/loop.texi: Same.
36179 * doc/md.texi: Same.
36180 * doc/objc.texi: Same.
36181 * doc/rtl.texi: Same.
36182 * doc/tm.texi: Same.
36183 * doc/tm.texi.in: Same.
36184 * doc/trouble.texi: Same.
36186 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
36189 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
36190 (collect_checksum_attributes): Set it.
36191 (die_checksum_ordered): Use it.
36193 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36195 PR tree-optimization/79908
36196 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
36197 change: For a VA_ARG whose LHS has been cast away, use
36198 force_gimple_operand to construct the side effects.
36200 2017-03-21 David Malcolm <dmalcolm@redhat.com>
36202 PR translation/80001
36203 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
36204 more amenable to translation.
36205 (oacc_loop_auto_partitions): Likewise.
36207 2017-03-21 Marek Polacek <polacek@redhat.com>
36208 Martin Sebor <msebor@redhat.com>
36210 PR tree-optimization/80109
36211 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
36212 on INTEGRAL_TYPE_P.
36214 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36215 Segher Boessenkool <segher@kernel.crashing.org>
36218 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
36219 check reg_used_between_p between insn and one of succ or succ2
36220 depending on if succ is artificial insn not inserted into insn
36223 2017-03-21 Martin Liska <mliska@suse.cz>
36225 PR gcov-profile/80081
36226 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
36227 * doc/gcc.texi: Include gcov-dump stuff.
36228 * doc/gcov-dump.texi: New file.
36230 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
36232 PR rtl-optimization/79150
36233 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
36234 conditional jump, if the jump is the last insn of the loop.
36236 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36237 Richard Biener <rguenth@suse.de>
36239 PR tree-optimization/79908
36240 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
36241 been cast away, use force_gimple_operand to construct the side
36244 2017-03-21 Martin Liska <mliska@suse.cz>
36246 PR libfortran/79956
36247 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
36250 2017-03-21 Brad Spengler <spender@grsecurity.net>
36253 * plugin.c (htab_hash_plugin): New function.
36254 (add_new_plugin): Use it and adjust.
36255 (parse_plugin_arg_opt): Adjust.
36256 (init_one_plugin): Likewise.
36258 2017-03-21 Richard Biener <rguenther@suse.de>
36260 PR tree-optimization/80032
36261 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
36262 if set force the cleanup to happen unconditionally.
36263 (gimplify_target_expr): Push inserted clobbers with force_uncond
36264 to avoid them being removed by control-dependent DCE.
36266 2017-03-21 Richard Biener <rguenther@suse.de>
36268 PR tree-optimization/80122
36269 * tree-inline.c (copy_bb): Do not expans va-arg packs or
36270 va_arg_pack_len when the inlined call stmt requires pack
36272 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
36274 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36277 * tsan.c (instrument_builtin_call): If the memory model argument
36278 is not a constant, assume it is valid.
36281 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
36284 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36286 PR rtl-optimization/79910
36287 * combine.c (can_combine_p): Do not allow combining an I0 or I1
36288 if its dest is used by an insn before I2 (other than the combined
36289 insns themselves, which are properly handled already).
36291 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36294 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36296 * combine.c (record_used_regs): New static function.
36297 (try_combine): Handle situations where there is an additional
36298 instruction between I2 and I3 which needs to have a LOG_LINK
36302 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36304 * combine.c (try_combine): Delete redundant i1 test. Call
36305 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36307 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36310 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
36311 alternatives 13/14.
36313 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36315 PR tree-optimization/80054
36316 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
36317 the optimization if a PHI or any of its arguments is not dominated
36318 by the candidate's basis. Use gphi* rather than gimple* as
36320 (replace_profitable_candidates): Clean up a gimple* variable that
36321 should be a gphi* variable.
36323 2017-03-20 Martin Sebor <msebor@redhat.com>
36326 * doc/extend.texi (attribute constructor): Document present limitation.
36328 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
36331 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
36332 __POWER9_VECTOR__ #ifdef control, change template definition to
36333 use Power9-specific built-in function.
36334 (vec_any_eq): Likewise.
36335 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
36336 to control outcomes from this test.
36337 (vector_ae_<mode>p): For VEC_F modes, likewise.
36339 2017-03-20 Ian Lance Taylor <iant@google.com>
36341 * config/i386/i386.c (ix86_function_regparm): Save an extra
36342 register for -fsplit-stack with DECL_STATIC_CHAIN.
36344 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36347 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
36348 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
36350 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36352 * config/riscv/riscv.c (riscv_print_operand): Use "fence
36354 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
36357 2017-03-20 Marek Polacek <polacek@redhat.com>
36360 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
36362 2017-03-20 Richard Biener <rguenther@suse.de>
36364 PR tree-optimization/80113
36365 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
36366 allocate extra SSA name for PHI def.
36367 (add_close_phis_to_outer_loops): Likewise.
36368 (add_close_phis_to_merge_points): Likewise.
36369 (copy_loop_close_phi_args): Likewise.
36370 (copy_cond_phi_nodes): Likewise.
36372 2017-03-20 Martin Liska <mliska@suse.cz>
36374 PR middle-end/79753
36375 * tree-chkp.c (chkp_build_returned_bound): Do not build
36376 returned bounds for a LHS that's not a BOUNDED_P type.
36378 2017-03-20 Martin Liska <mliska@suse.cz>
36382 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
36383 COMPLEX_CST and VECTOR_CST.
36385 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36388 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
36389 target operand. A new splitter adds the clobber statement in case
36390 the target operand is dead anyway.
36392 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
36394 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
36395 to age-old versions of binutils and glibc.
36397 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
36399 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
36401 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36403 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
36405 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36407 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
36408 requirement for binutils 2.13.
36410 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36412 * combine.c (try_combine): Delete redundant i1 test. Call
36413 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36415 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
36417 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
36418 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
36420 <riscv64-*-elf>: Re-arrange section
36421 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
36422 <riscv32-*-linux>: Likewise.
36423 <riscv64-*-elf>: Likewise
36424 <riscv64-*-linux>: Likewise.
36426 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
36429 * aarch64.opt(verbose-cost-dump): Fix typo.
36431 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
36434 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
36435 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
36437 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36439 * reload.c (find_reloads): When reloading a nonoffsettable address,
36440 use RELOAD_OTHER for it and its address reloads.
36442 PR rtl-optimization/79910
36443 * combine.c (record_used_regs): New static function.
36444 (try_combine): Handle situations where there is an additional
36445 instruction between I2 and I3 which needs to have a LOG_LINK
36448 2017-03-17 Jeff Law <law@redhat.com>
36450 PR tree-optimization/71437
36451 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
36452 conditional in the hash table first.
36453 (vrp_dom_walker::before_dom_children): Extract condition from
36454 ASSERT_EXPR. Record condition, its inverion and any implied
36455 conditions as well.
36457 2017-03-17 Marek Polacek <polacek@redhat.com>
36458 Markus Trippelsdorf <markus@trippelsdorf.de>
36460 PR tree-optimization/80079
36461 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
36464 2017-03-17 Richard Biener <rguenther@suse.de>
36466 PR middle-end/80075
36467 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
36468 Properly verify the LHS before the RHS possibly claims to be
36470 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
36473 2017-03-17 Martin Jambor <mjambor@suse.cz>
36475 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
36476 (List of -O2 options): Likewise.
36477 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
36480 2017-03-17 Tom de Vries <tom@codesourcery.com>
36482 * gcov-dump.c (print_usage): Print bug_report_url.
36484 2017-03-17 Richard Biener <rguenther@suse.de>
36486 PR middle-end/80050
36487 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
36488 (parser::peek): Likewise.
36490 2017-03-17 Richard Biener <rguenther@suse.de>
36492 PR tree-optimization/80048
36493 * sese.c (free_sese_info): Properly release rename_map and
36494 copied_bb_map elements.
36496 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
36498 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
36499 Add linked-list forward and backlinks. Insert on
36500 construction, remove on destruction.
36501 (class pass_store_merging): Add m_stores_head field.
36502 (pass_store_merging::terminate_and_process_all_chains):
36503 Iterate over m_stores_head list.
36504 (pass_store_merging::terminate_all_aliasing_chains):
36506 (pass_store_merging::execute): Check for debug stmts first.
36507 Push new chains onto the m_stores_head stack.
36509 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
36512 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
36513 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
36514 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
36516 2017-03-16 Jeff Law <law@redhat.com>
36518 PR tree-optimization/71437
36519 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
36520 member function. Implementation moved into after_dom_children
36521 member function and into the threader's thread_outgoing_edges
36523 (dom_opt_dom_walker::after_dom_children): Simplify by moving
36524 some code into new thread_outgoing_edges.
36525 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
36526 definition. Simplify marker handling (do it here). Assume we always
36527 have the available expression and the const/copies tables.
36528 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
36530 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
36531 * tree-vrp.c (equiv_stack): No longer file scoped.
36532 (vrp_dom_walker): New class.
36533 (vrp_dom_walker::before_dom_children): New member function.
36534 (vrp_dom_walker::after_dom_children): Likewise.
36535 (identify_jump_threads): Setup domwalker. Use it rather than
36536 walking edges in a random order by hand. Simplify setup/finalization.
36537 (finalize_jump_threads): Remove.
36538 (vrp_finalize): Do not call identify_jump_threads here.
36539 (execute_vrp): Do it here instead and call thread_through_all_blocks
36542 PR tree-optimization/71437
36543 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
36545 (simplify_stmt_for_jump_threading): Add basic_block argument. All
36547 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
36548 (dom_opt_dom_walker::thread_across_edge): Remove
36549 handle_dominating_asserts argument. All callers changed.
36550 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
36551 changes. Remove calls to lhs_of_dominating_assert. Other
36552 uses of handle_dominating_asserts turn into unconditional code
36553 (simplify_control_stmt_condition_1): Likewise.
36554 (simplify_control_stmt_condition): Likewise.
36555 (thread_through_normal_block, thread_across_edge): Likewise.
36556 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
36557 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
36558 object if it is not an SSA_NAME.
36559 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
36560 before calling into the VRP specific simplifiers.
36561 (identify_jump_threads): Remove handle_dominating_asserts
36564 2017-03-16 Jakub Jelinek <jakub@redhat.com>
36567 * tree-diagnostic.c (default_tree_printer): No longer static.
36568 * tree-diagnostic.h (default_tree_printer): New prototype.
36570 2017-03-16 Tamar Christina <tamar.christina@arm.com>
36572 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
36573 Change ins into fmov.
36575 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36577 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
36578 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
36579 Use h_con constraint for operand 1.
36580 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
36581 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
36583 2017-03-15 Jeff Law <law@redhat.com>
36585 PR tree-optimization/71437
36586 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
36587 (record_temporary_equivalences): Use it.
36589 PR tree-optimization/71437
36590 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
36591 tree-ssa-scopedtables.
36592 (lookup_avail_expr, build_and_record_new_cond): Likewise.
36593 (record_conditions, record_cond, vuse_eq): Likewise.
36594 (record_edge_info): Adjust to API tweak of record_conditions.
36595 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
36596 (record_temporary_equivalences, optimize_stmt): Likewise.
36597 (eliminate_redundant_computations): Likewise.
36598 (record_equivalences_from_stmt): Likewise.
36599 * tree-ssa-scopedtables.c: Include options.h and params.h.
36600 (vuse_eq): New function, moved from tree-ssa-dom.c
36601 (build_and_record_new_cond): Likewise.
36602 (record_conditions): Likewise. Accept vector of conditions rather
36603 than edge_equivalence structure for first argument.
36604 for the first argument.
36605 (avail_exprs_stack::lookup_avail_expr): New member function, moved
36606 from tree-ssa-dom.c.
36607 (avail_exprs_stack::record_cond): Likewise.
36608 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
36609 from tree-ssa-dom.c.
36610 (avail_exprs_stack): Add new member functions lookup_avail_expr
36612 (record_conditions): Declare.
36614 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
36617 * lra-constraints.c (process_alt_operands): Increase reject for
36618 reloading an input/output operand.
36620 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
36623 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
36624 insns to convert from signed/unsigned char/short to IEEE 128-bit
36626 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
36628 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
36631 * config/i386/i386.c (ix86_vector_duplicate_value): Create
36632 subreg of inner mode for values already in registers.
36634 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
36636 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
36637 iteration reg is used after the loop.
36639 2017-03-14 Martin Sebor <msebor@redhat.com>
36641 PR tree-optimization/79800
36642 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
36643 precision in negative-positive range.
36644 (format_floating): Call non-const overload with adjusted precision.
36646 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
36649 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
36652 2017-03-14 Martin Sebor <msebor@redhat.com>
36654 PR middle-end/80020
36655 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
36656 * builtins.def (aligned_alloc): Use it.
36659 * Makefile.in (GTFILES): Add calls.c.
36660 * calls.c: Include "gt-calls.h".
36662 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
36664 PR rtl-optimization/79728
36665 * regs.h (struct target_regs): New field
36666 x_contains_allocatable_regs_of_mode.
36667 (contains_allocatable_regs_of_mode): New macro.
36668 * reginfo.c (init_reg_sets_1): Initialize it, and change
36669 contains_reg_of_mode so it includes global regs as well.
36670 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
36671 rather than contains_regs_of_mode.
36673 2017-03-14 Martin Liska <mliska@suse.cz>
36675 * doc/invoke.texi: Document options that can't be combined with
36676 -fcheck-pointer-bounds.
36678 2017-03-14 Martin Liska <mliska@suse.cz>
36680 PR middle-end/79831
36681 * doc/invoke.texi (-Wchkp): Document the option.
36683 2017-03-14 Martin Liska <mliska@suse.cz>
36685 * Makefile.in: Install gcov-dump.
36687 2017-03-14 Martin Liska <mliska@suse.cz>
36689 * multiple_target.c (expand_target_clones): Bail out for
36690 an invalid attribute.
36692 2017-03-14 Richard Biener <rguenther@suse.de>
36694 * alias.c (struct alias_set_entry): Pack properly.
36695 * cfgloop.h (struct loop): Likewise.
36696 * cse.c (struct set): Likewise.
36697 * ipa-utils.c (struct searchc_env): Likewise.
36698 * loop-invariant.c (struct invariant): Likewise.
36699 * lra-remat.c (struct cand): Likewise.
36700 * recog.c (struct change_t): Likewise.
36701 * rtl.h (struct address_info): Likewise.
36702 * symbol-summary.h (function_summary): Likewise.
36703 * tree-loop-distribution.c (struct partition): Likewise.
36704 * tree-object-size.c (struct object_size_info): Likewise.
36705 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
36706 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
36707 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
36708 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
36709 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
36710 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
36711 (struct _stmt_vec_info): Likewise.
36713 2017-03-14 Martin Liska <mliska@suse.cz>
36716 * multiple_target.c (create_dispatcher_calls): Check that
36717 a target can create a function dispatcher.
36719 2017-03-14 Martin Liska <mliska@suse.cz>
36722 * multiple_target.c (expand_target_clones): Drop local.local
36723 flag for default implementation.
36725 2017-03-14 Richard Biener <rguenther@suse.de>
36727 PR tree-optimization/80030
36728 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
36730 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
36732 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
36733 gcc_fallthrough() instead of __attribute__((fallthrough));
36735 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36737 * doc/gcc.texi: Remove "up" link to (DIR).
36738 * doc/gccint.texi: Ditto.
36740 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36742 * doc/install.texi (Specific) <avr>: Remove reference to
36745 2017-03-13 Jeff Law <law@redhat.com>
36747 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
36748 attribute rather than comments.
36750 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
36751 match_scratch operand is highest.
36753 2017-03-13 Martin Liska <mliska@suse.cz>
36755 PR middle-end/78339
36756 * ipa-pure-const.c (warn_function_noreturn): If the declarations
36757 is a CHKP clone, use original declaration.
36759 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36761 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
36762 (arc_conditional_register_usage): Use a different allocation order
36763 when optimizing for size.
36764 * common/config/arc/arc-common.c (arc_option_optimization_table):
36765 Section anchors default on when optimizing for size.
36767 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36769 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
36771 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36773 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
36774 * config/arc/arc.md (cpu_facility): Add cd variant.
36775 (*movqi_insn): Add code density variant.
36776 (*movhi_insn): Likewise.
36777 (*movqi_insn): Likewise.
36778 (*addsi3_mixed): Likewise.
36779 (subsi3_insn): Likewise.
36781 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36783 * config/arc/arc.md (movsi_cond_exec): Update constraint.
36785 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36787 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
36788 expressions with MINUS and UNARY ops.
36790 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36793 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
36795 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
36796 between vec_select and vector argument.
36797 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
36798 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
36799 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
36800 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
36801 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
36802 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
36804 2017-03-13 Richard Biener <rguenther@suse.de>
36807 * params.def (vect-max-peeling-for-alignment): Fix typo.
36809 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36811 * doc/install.texi (Specific) <mips-*-*>: Remove description of
36812 issue that only occurred with binutils below 2.18.
36814 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36816 * doc/install.texi (Specific) <cris-axis-elf>: No longer
36817 refer to binutils 2.11/2.12 minimum.
36819 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36821 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
36822 ftp.kernel.org and simplify binutils requirement.
36824 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
36826 * doc/invoke.texi (Warning Options): Fix spelling of link-time
36828 (Optimize Options): Ditto. Also remove redundancy.
36830 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36832 PR translation/79848
36833 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
36835 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
36836 to G_ to avoid double translation.
36838 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36840 PR translation/79923
36841 * auto-profile.c (get_combined_location): Convert leading
36842 character of diagnostics to lower case and remove trailing period.
36843 (read_profile): Likewise for various diagnostics.
36844 * config/arm/arm.c (arm_option_override): Remove trailing period
36845 from various diagnostics.
36846 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
36847 (msp430_expand_delay_cycles): Likewise.
36849 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36852 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36853 full command-line argument, rather than just "str".
36854 (aarch64_validate_march): Likewise.
36855 (aarch64_validate_mtune): Likewise.
36857 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36859 PR rtl-optimization/78911
36860 * lra-assigns.c (must_not_spill_p): New function.
36861 (spill_for): Use it.
36863 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36865 PR tree-optimization/79981
36866 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36867 ATOMIC_COMPARE_EXCHANGE ifn result.
36868 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36869 IFN_ATOMIC_COMPARE_EXCHANGE.
36871 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36874 * opts.c (parse_sanitizer_options): Add missing question mark to
36875 "did you mean" message.
36877 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36879 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36881 (VMULEUH_UNS): Likewise.
36882 (VMULOUB_UNS): Likewise.
36883 (VMULOUH_UNS): Likewise.
36884 * config/rs6000/rs6000.c (builtin_function_type): Remove
36885 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36887 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36890 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36891 x with result of extra_parsing_for_operand_code_0.
36892 (function_reader::extra_parsing_for_operand_code_0): Convert
36893 return type from void to rtx, returning x. When reading
36894 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36895 larger size containing struct block_symbol.
36897 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36899 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36900 -mfloat128-hardware without -m64.
36902 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36905 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36906 entries to the case statement that marks unsigned arguments to
36907 overloaded functions.
36909 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36911 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36912 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36914 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36917 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36918 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36920 2017-03-10 Martin Liska <mliska@suse.cz>
36924 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36925 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36928 2017-03-10 Olivier Hainque <hainque@adacore.com>
36930 * tree-switch-conversion (array_value_type): Start by resetting
36931 candidate type to it's main variant.
36933 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36935 PR rtl-optimization/79909
36936 * combine.c (try_combine): Use simplify_replace_rtx on individual
36937 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36938 of the whole CALL_INSN_FUNCTION_USAGE.
36940 PR tree-optimization/79972
36941 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36942 get_range_info on SSA_NAMEs. Formatting fixes.
36944 2017-03-10 Richard Biener <rguenther@suse.de>
36945 Jakub Jelinek <jakub@redhat.com>
36947 PR tree-optimization/77975
36948 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36949 edge to be constant.
36950 (get_val_for): For constant x return it. Formatting fix.
36951 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36952 would use the same bases as the current one.
36954 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36956 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36957 instead of vec_select for V1TImode.
36958 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36960 (VSX_LE_128): Add V1TI to this mode iterator.
36961 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36962 (*vsx_le_perm_store_<mode>): Likewise.
36963 (pre-reload splitter for VSX stores): Likewise.
36964 (post-reload splitter for VSX stores): Likewise.
36965 (*vsx_xxpermdi2_le_<mode>): Likewise.
36966 (*vsx_lxvd2x2_le_<mode>): Likewise.
36967 (*vsx_stxvd2x2_le_<mode>): Likewise.
36969 2017-03-09 Michael Eager <eager@eagercon.com>
36971 Correct failures with --enable-checking=yes,rtl.
36973 * config/microblaze/microblaze.c (microblaze_expand_shift):
36974 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36975 test for const0_rtx.
36976 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36977 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36979 2017-03-09 Richard Biener <rguenther@suse.de>
36981 PR tree-optimization/79977
36982 * graphite-scop-detection.c (scop_detection::merge_sese):
36983 Handle the case of extra exits to blocks dominating the entry.
36985 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36987 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36990 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36992 PR rtl-optimization/79949
36993 * lra-constraints.c (process_alt_operands): Check memory when
36994 trying to predict a cycle. Print about the overall increase.
36996 2017-03-09 Richard Biener <rguenther@suse.de>
36998 PR middle-end/79971
36999 * gimple-expr.c (useless_type_conversion_p): Preserve
37000 TYPE_SATURATING for fixed-point types.
37002 2017-03-09 Richard Biener <rguenther@suse.de>
37005 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
37006 alignment of BLKmode params.
37008 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37011 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
37012 (VALL_NO_V2Q): Likewise.
37014 * config/aarch64/aarch64-simd.md
37015 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
37017 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
37018 VALL_NO_V2Q mode iterator.
37019 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
37021 2017-03-09 Martin Liska <mliska@suse.cz>
37023 PR tree-optimization/79631
37024 * tree-chkp-opt.c (chkp_is_constant_addr): Call
37025 tree_int_cst_sign_bit just for INTEGER constants.
37027 2017-03-09 Martin Liska <mliska@suse.cz>
37031 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
37034 2017-03-09 Marek Polacek <polacek@redhat.com>
37037 * tree.c (inchash::add_expr): Handle TREE_VEC.
37039 2017-03-09 Martin Liska <mliska@suse.cz>
37042 (chkp_narrow_size_and_offset): New function.
37043 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
37044 (void chkp_parse_bit_field_ref): New function.
37045 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
37046 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
37048 2017-03-09 Martin Liska <mliska@suse.cz>
37051 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
37052 (chkp_find_bounds_1): Remove gcc_unreachable.
37054 2017-03-09 Jakub Jelinek <jakub@redhat.com>
37057 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
37058 BUILT_IN_SYNC*, determine the access type from the size suffix and
37059 always build a MEM_REF with that type. Handle forgotten
37060 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
37063 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
37064 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
37065 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
37066 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
37067 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
37068 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
37069 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
37070 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
37071 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
37072 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
37073 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
37074 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
37075 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
37076 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
37077 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
37078 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
37079 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
37080 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
37081 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
37082 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
37083 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
37084 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
37085 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
37086 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
37087 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
37088 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
37089 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
37090 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
37091 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
37092 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
37093 definitions outside of __OPTIMIZE__ guarded section.
37096 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
37097 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
37098 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
37099 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
37102 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37104 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
37105 ("vfenez<mode>"): Add missing constraints.
37107 2017-03-08 Martin Sebor <msebor@redhat.com>
37110 * config/nds32/nds32.c (nds32_option_override):
37111 Fix misspelled diagnostic.
37113 2017-03-08 Jakub Jelinek <jakub@redhat.com>
37116 * gimplify.c (gimplify_omp_for): Replace index var in outer
37117 taskloop statement with an artificial variable and add
37118 OMP_CLAUSE_PRIVATE clause for it.
37120 2017-03-08 Richard Biener <rguenther@suse.de>
37122 PR tree-optimization/79955
37123 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
37124 for accesses that are completely outside of the variable.
37126 2017-03-08 Andrew Haley <aph@redhat.com>
37128 PR tree-optimization/79943
37129 * tree-ssa-loop-split.c (compute_new_first_bound): When
37130 calculating the new upper bound, (END-BEG) should be added, not
37133 2017-03-08 Jakub Jelinek <jakub@redhat.com>
37135 * config/avr/avr.md (setmemhi): Make sure match_dup
37136 operand number comes before match_scratch.
37138 2017-03-08 Richard Biener <rguenther@suse.de>
37140 PR tree-optimization/79920
37141 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
37142 with ncopies == 1 to ...
37143 (vect_transform_slp_perm_load): ... here. Properly compute
37144 all element loads by iterating VF times over the group. Do
37145 not handle ncopies (computed in a broken way) in
37146 vect_create_mask_and_perm.
37148 2017-03-08 Jakub Jelinek <jakub@redhat.com>
37151 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
37152 is a uniform vector, use uniform_vector_p return value instead of
37153 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
37155 2017-03-07 Marek Polacek <polacek@redhat.com>
37157 PR middle-end/79809
37158 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
37159 (alloca_call_type): Likewise.
37161 2017-03-07 Martin Liska <mliska@suse.cz>
37163 * gcov.c (process_args): Put comment to correct location.
37165 2017-03-07 Martin Liska <mliska@suse.cz>
37167 PR middle-end/68270
37168 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
37169 Use array_at_struct_end_p instead of DECL_CHAIN (field).
37170 (chkp_narrow_bounds_for_field): Likewise.
37171 (chkp_parse_array_and_component_ref): Pass one more argument to
37174 2017-03-07 Richard Biener <rguenther@suse.de>
37176 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
37179 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
37181 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
37182 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
37184 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37187 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
37188 to end of description.
37189 (PARAM_MAX_STORES_TO_MERGE): Likewise.
37191 2017-03-07 Jakub Jelinek <jakub@redhat.com>
37193 PR rtl-optimization/79901
37194 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
37196 (*avx512f_<code><mode>3<mask_name>): ... this.
37197 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
37198 iterator instead of VI8_AVX2_AVX512BW.
37200 PR rtl-optimization/79901
37201 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
37202 min/max expander, expand it using expand_vec_cond_expr.
37205 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
37208 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37211 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
37212 to void * for PCH reasons.
37213 * dwarf2out.c (output_loc_operands, output_die): Cast
37214 v.val_vec.array to unsigned char *.
37216 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
37219 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
37222 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
37224 PR rtl-optimization/79571
37225 * lra-constraints.c (process_alt_operands): Calculate static
37226 reject and subtract it from overall when only addresses will be
37229 2017-03-06 Julia Koval <julia.koval@intel.com>
37232 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
37233 incoming stack boundary to 128 for 64-bit targets.
37235 2017-03-06 Richard Biener <rguenther@suse.de>
37237 PR tree-optimization/79894
37238 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
37239 to NULL after folding it.
37241 2017-03-06 Richard Biener <rguenther@suse.de>
37243 PR tree-optimization/79824
37244 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
37245 check disabling peeling for gaps.
37247 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
37249 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
37250 attributes): Document gettimeofday.
37252 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37254 * config/s390/s390.c (s390_option_override_internal): Set
37255 PARAM_MIN_VECT_LOOP_BOUND
37257 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37259 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
37260 * config/s390/s390.md: Likewise.
37262 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37265 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
37266 (<avx2_avx512>_perm<mode>): Rename to ...
37267 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
37269 (<avx512>_perm<mode>_mask): Rename to ...
37270 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
37272 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
37273 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
37274 instead of VI8F_256_512.
37275 (avx512f_perm<mode>): New define_expand.
37276 (avx512f_perm<mode>_mask): Likewise.
37277 (avx512f_perm<mode>_1<mask_name>): New define_insn.
37278 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
37280 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37282 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
37283 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
37285 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
37287 2017-03-06 Martin Liska <mliska@suse.cz>
37290 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
37291 when having a SSA NAME w/o VAR_DECL assigned to it.
37293 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37295 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
37296 msa_dpsub_<su>_d): Fix MODE for vec_select.
37298 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37300 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
37302 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
37304 2017-03-06 Richard Biener <rguenther@suse.de>
37306 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
37307 * plugin.c (register_plugin_info): Likewise.
37308 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
37310 2017-03-05 Jakub Jelinek <jakub@redhat.com>
37312 * config/i386/sse.md (sse_storehps, sse_storelps,
37313 avx_<castmode><avxsizesuffix>_<castmode>,
37314 avx512f_<castmode><avxsizesuffix>_<castmode>,
37315 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
37316 in condition that at least one operand is not a MEM.
37318 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37320 PR middle-end/79805
37321 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
37322 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
37324 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
37325 gimple_call_nothrow_p flag based on whether original builtin can throw.
37326 If it can, emit following stmts on the fallthrough edge.
37327 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
37328 don't create new bb if inserting just debug stmts on the edge, try to
37329 insert them on the fallthru bb or just reset debug stmts.
37331 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
37334 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
37335 restore recog_data (including the operand rtxes inside it) around
37336 the call to get_insn_template.
37338 2017-03-03 Martin Sebor <msebor@redhat.com>
37340 PR tree-optimization/79699
37341 * context.c (context::~context): Free MPFR caches to avoid
37342 a memory leak on program exit.
37344 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37346 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
37347 Use wide_int::ulow () instead of .elt (0).
37349 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37351 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
37352 (*pushxf): Limit oF constraint to 32bit targets and add oC
37353 constraint for 64bit targets.
37354 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
37355 (*pushdf): Change rmF constraint to rmC.
37357 2017-03-03 Martin Liska <mliska@suse.cz>
37359 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
37360 Remove unused variable.
37362 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37365 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
37366 is a memory operand, increase num_memory.
37367 (ix86_expand_args_builtin): Likewise.
37369 2017-03-03 Jan Hubicka <jh@suse.cz>
37372 * ipa-devirt.c (maybe_record_node): Properly handle
37373 __cxa_pure_virtual visibility.
37375 2017-03-03 Martin Liska <mliska@suse.cz>
37377 PR tree-optimization/79803
37378 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
37380 (pass_loop_prefetch::execute): Disabled optimization if an
37381 assumption about L1 cache size is not met.
37383 2017-03-03 Martin Liska <mliska@suse.cz>
37385 PR rtl-optimization/79574
37386 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
37387 (hash_scan_set): Likewise.
37388 (dump_hash_table): Likewise.
37389 (hoist_code): Likewise.
37391 2017-03-03 Richard Biener <rguenther@suse.de>
37393 * fixed-value.c (fixed_from_string): Restore use of elt (1)
37394 in place of uhigh ().
37395 (fixed_convert_from_real): Likewise.
37397 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37400 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
37402 2017-03-03 Richard Biener <rguenther@suse.de>
37404 PR middle-end/79818
37405 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
37406 TYPE_OVERFLOW_UNDEFINED check.
37408 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37410 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
37412 (vector_ae_<mode>_p): Likewise.
37413 (vector_nez_<mode>_p): Likewise.
37414 (vector_ne_v2di_p): Likewise.
37415 (vector_ae_v2di_p): Likewise.
37416 (vector_ne_<mode>_p): Likewise.
37417 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
37419 (vsx_tsqrt<mode>2_fe): Likewise.
37421 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
37424 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
37426 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37428 PR rtl-optimization/79780
37429 * cprop.c (one_cprop_pass): When second and further conditional trap
37430 in a single basic block is turned into an unconditional trap, turn it
37431 into a deleted note to avoid RTL verification failures.
37433 2017-03-02 Richard Biener <rguenther@suse.de>
37435 * fold-const.c (const_binop): Use ulow () instead of elt (0).
37437 2017-03-02 Richard Biener <rguenther@suse.de>
37439 PR tree-optimization/79345
37441 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
37442 param and abort the walk, returning -1 if it is hit.
37443 (walk_aliased_vdefs): Take a limit param and pass it on.
37444 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
37445 defaulting to 0 and return a signed int.
37446 * tree-ssa-uninit.c (struct check_defs_data): New struct.
37447 (check_defs): New helper.
37448 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
37449 about uninitialized memory.
37450 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
37451 bogus uninitialized warning.
37452 (fixed_convert_from_real): Likewise.
37454 2017-03-02 Bin Cheng <bin.cheng@arm.com>
37456 PR tree-optimization/66768
37457 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
37458 iv_use if base object can't be determined.
37460 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37462 PR tree-optimization/79345
37463 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
37464 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
37465 (get_pattern_stats): Initialize it.
37466 * genemit.c (gen_expand): Verify match_scratch numbers come after
37467 match_operand/match_dup numbers.
37468 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
37469 match_scratch numbers.
37470 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
37472 * config/s390/s390.md (trunctdsd2): Likewise.
37474 2017-03-02 Richard Biener <rguenther@suse.de>
37476 * wide-int.h (wide_int_storage::operator=): Implement in terms
37479 2017-03-02 Richard Biener <rguenther@suse.de>
37481 PR tree-optimization/79777
37482 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
37483 the to insert expression to sth existing.
37485 2017-03-01 Martin Sebor <msebor@redhat.com>
37487 PR middle-end/79692
37488 * gimple-ssa-sprintf.c
37489 (directive::known_width_and_precision): New function.
37490 (format_integer): Use it.
37491 (get_mpfr_format_length): Consider the full range of precision
37492 when computing %g output with the # flag. Set the likely byte
37493 count to 3 rather than 1 when precision is indeterminate.
37494 (format_floating): Correct the lower bound of precision.
37496 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37498 * doc/invoke.texi: Document default code model for 64-bit Linux.
37500 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37503 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
37504 udiv rather than div since input pattern is unsigned.
37506 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
37508 * config/i386/i386.c (print_reg): Warn for values of
37509 unsupported size in integer register.
37511 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
37514 * config/rs6000/predicates.md (current_file_function_operand): Do
37515 not allow self calls to be local if the function is replaceable.
37517 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
37520 * config/rs6000/altivec.h (vec_ctz and others): Change the
37521 preprocessor macro that controls conditional compilation from
37522 _ARCH_PWR9 to __POWER9_VECTOR__.
37523 (vec_all_ne): Change parameterization of __altivec_scalar_pred
37524 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
37525 control (instead of _ARCH_PWR9 control) so that template
37526 definition uses power9-specific function.
37527 (vec_any_eq): Likewise.
37528 (vec_all_ne): Change macro definition to use a power9-specific
37529 expansion under #ifdef __POWER9_VECTOR__ control (instead of
37530 _ARCH_PWR9 control).
37531 (vec_any_eq) Likewise.
37532 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
37533 expansion for CMPNEF to remove support for xvcmpnesp instruction.
37534 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
37535 support for xvcmpnedp instruction.
37536 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
37537 macro expansion so that Power9 implementation of vec_all_ne does
37538 not use the AltiVec predicate framework.
37539 (VCMPNEH_P): Likewise.
37540 (VCMPNEW_P): Likewise.
37541 (VCMPNED_P): Likewise.
37542 (VCMPNEFP_P): Likewise.
37543 (VCMPNEDP_P): Likewise.
37544 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
37545 implementation of vec_any_eq to not use AltiVec predicate
37547 (VCMPAEH_P): Likewise.
37548 (VCMPAEW_P): Likewise.
37549 (VCMPAED_P): Likewise.
37550 (VCMPAEFP_P): Likewise.
37551 (VCMPAEDP_P): Likewise.
37552 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
37553 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
37554 not use the AltiVec predicate framework.
37555 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
37556 of vec_any_eq to not use AltiVec predicate framework.
37557 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
37558 support for predefined __POWER9_VECTOR__ macro to indicate that
37559 Power9 instruction selection is enabled.
37560 (altivec_overloaded_builtins): Remove extraneous
37561 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
37562 function argument types RS6000_BTI_bool_V16QI and
37563 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
37564 entry for overloaded function argument types RS6000_BTI_bool_V4SI
37565 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
37566 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
37567 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
37568 Power9 for implementations of vec_cmpne. Change the signature for
37569 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
37570 (representing vec_all_ne) to remove the previously described first
37571 argument of type RS6000_BTI_INTSI, as this was an artifact of
37572 reliance on the AltiVec predicate framework, which is no longer
37573 used in the implementation of these functions. Add
37574 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
37575 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
37576 since, unlike the AltiVec predicate framework implementation, we
37577 do not share function descriptors between vec_alle and vec_anyeq.
37578 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
37579 set of modes that receive special treatment even when
37580 TARGET_P9_VECTOR is true. The special treatment emits code that
37581 does not depend on Power9 instructions.
37582 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
37583 define_expand to not rely on AltiVec predicate framework.
37584 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37586 (vector_ne_v2di_p): Change this define_expand to not rely on
37587 AltiVec predicate framework.
37588 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
37590 (vector_ne_<mode>_p): Change this define_expand to not rely on
37591 AltiVec predicate framework.
37592 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37594 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
37595 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
37596 define_insn pattern.
37597 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
37598 define_insn pattern because the xvcmpne<VSs>. instruction is not
37600 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
37601 instruction is not supported.
37603 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37605 * config/nvptx/nvptx.c: Include intl.h.
37607 2017-03-01 Martin Jambor <mjambor@suse.cz>
37610 * ipa-prop.h (ipa_bits): Removed field known.
37611 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
37612 to pointers. Adjusted their comments to warn about their sharing.
37613 (ipcp_transformation_summary): Change bits to a vector of pointers.
37614 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
37615 (ipa_get_ipa_bits_for_value): Declare.
37616 * tree-vrp.h (value_range): Mark as GTY((for_user)).
37617 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
37618 (ipa_bits_hash_table): Likewise.
37619 (ipa_vr_ggc_hash_traits): Likewise.
37620 (ipa_vr_hash_table): Likewise.
37621 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
37622 being pointers and vr_known being removed.
37623 (ipa_set_jf_unknown): Likewise.
37624 (ipa_get_ipa_bits_for_value): New function.
37625 (ipa_set_jfunc_bits): Likewise.
37626 (ipa_get_value_range): New overloaded functions.
37627 (ipa_set_jfunc_vr): Likewise.
37628 (ipa_compute_jump_functions_for_edge): Use the above functions to
37629 construct bits and vr parts of jump functions.
37630 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
37631 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37633 (ipcp_grow_transformations_if_necessary): Also allocate
37634 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37636 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
37637 them. Fix too long lines.
37638 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
37639 vr_known being removed.
37640 (ipa_read_jump_function): Use new setter functions to construct bits
37641 and vr parts of jump functions or set them to NULL.
37642 (write_ipcp_transformation_info): Adjust for bits being pointers.
37643 (read_ipcp_transformation_info): Likewise.
37644 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
37646 Include gt-ipa-prop.h.
37647 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
37649 (ipcp_store_bits_results): Likewise.
37650 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
37651 Do not write to existing jump functions but use a temporary instead.
37653 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37656 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
37657 attempt to use its first operand as BIT_FIELD_REF base.
37659 2017-03-01 Richard Biener <rguenther@suse.de>
37661 PR middle-end/79721
37662 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
37663 interpolating formula in wrapping arithmetic.
37664 (chrec_apply): Convert chrec_evaluate return value to wanted type.
37666 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37668 PR tree-optimization/79734
37669 * tree-vect-generic.c (expand_vector_condition): Optimize
37670 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
37671 Handle VEC_COND_EXPR where comparison has different inner width from
37672 type's inner width.
37674 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
37676 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
37677 markup, and similar issues. Remove @opindex entries for things
37678 that aren't options. Add missing -mmpy-option entries.
37680 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37682 PR tree-optimization/79737
37683 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
37684 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
37685 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
37686 instead of byte_size. Formatting fix.
37687 (shift_bytes_in_array_right): Formatting fix.
37689 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
37692 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
37693 condition on optimize for the leaf function test.
37695 2017-02-28 Martin Liska <mliska@suse.cz>
37698 * read-rtl-function.c (function_reader::handle_unknown_directive):
37699 Bail out when one uses -flto.
37701 2017-02-28 Martin Liska <mliska@suse.cz>
37703 * common.opt: Replace space with tabular for options of <number>
37705 * config/i386/i386.opt: Show <number> value for
37706 -mlarge-data-threshold.
37707 * opts.c (print_filtered_help): Do not display number in hexadecimal
37710 2017-02-28 Martin Liska <mliska@suse.cz>
37712 * common.opt: Fix --help=option -Q for options which are of
37715 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
37717 * config/i386/i386.c (print_reg): Error out for values
37718 of 8-bit size in invalid integer register.
37720 2017-02-28 Martin Sebor <msebor@redhat.com>
37722 PR tree-optimization/79691
37723 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
37725 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37728 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
37729 gcc_unreachable with output_operand_lossage.
37731 2017-02-28 Richard Biener <rguenther@suse.de>
37733 PR tree-optimization/79740
37734 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
37736 (visit_nary_op): Insert the nary into the hashtable if we
37737 pattern-matched sth.
37738 * tree-ssa-pre.c (eliminate_insert): Robustify.
37740 2017-02-28 Richard Biener <rguenther@suse.de>
37742 PR middle-end/79731
37743 * fold-const.c (decode_field_reference): Reject out-of-bound
37746 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37748 * config/i386/i386.c: Include intl.h.
37749 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
37750 instead of just cond ? "..." : "...".
37751 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
37752 * coverage.c (read_counts_file): Likewise.
37753 * omp-offload.c: Include intl.h.
37754 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
37755 of just cond ? "..." : "...".
37756 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
37757 of just cond ? "..." : "...".
37759 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
37762 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
37764 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
37765 'tune for' CPU name.
37766 * config/arm/arm-cpu-data.h: Regenerated.
37768 2017-02-28 Richard Biener <rguenther@suse.de>
37770 PR tree-optimization/79732
37771 * tree-inline.c (expand_call_inline): Do not shadow var.
37773 2017-02-28 Richard Biener <rguenther@suse.de>
37775 PR tree-optimization/79723
37776 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
37777 address-space properly.
37779 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
37781 * doc/optinfo.texi (Optimization groups): Fix option used for
37783 * doc/invoke.texi (-fopt-info): Document "omp".
37784 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
37785 (OPTGROUP_ALL): Add OPTGROUP_OMP.
37786 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
37787 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
37788 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
37790 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
37792 * dumpfile.c (optgroup_options): Instead of "openmp", associate
37793 OPTGROUP_OMP with "omp".
37795 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
37798 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
37799 for arithmetic shift of unsigned V2DI.
37801 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
37803 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
37804 arc/linux.h headers.
37805 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
37806 (LINK_SPEC): Likewise.
37807 (ARC_TLS_EXTRA_START_SPEC): Likewise.
37808 (EXTRA_SPECS): Likewise.
37809 (STARTFILE_SPEC): Likewise.
37810 (ENDFILE_SPEC): Likewise.
37811 (LIB_SPEC): Likewise.
37812 (TARGET_SDATA_DEFAULT): Likewise.
37813 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
37814 (MULTILIB_DEFAULTS): Likewise.
37815 (DWARF2_UNWIND_INFO): Likewise.
37816 * config/arc/big.h: New file.
37817 * config/arc/elf.h: Likewise.
37818 * config/arc/linux.h: Likewise.
37819 * config/arc/t-uClibc: Remove.
37821 2017-02-27 Bin Cheng <bin.cheng@arm.com>
37823 PR tree-optimization/77536
37824 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
37825 (tree_transform_and_unroll_loop): Use above function to compute the
37826 estimated niter of unrolled loop and use it when scaling profile.
37827 Also use count info rather than frequency if it's non-zero.
37828 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
37829 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
37830 (vect_transform_loop): Call above function.
37832 2017-02-27 Richard Biener <rguenther@suse.de>
37834 PR tree-optimization/45397
37835 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
37836 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
37837 (visit_nary_op): Add pattern matching for CSEing sign-changed
37838 or truncated operations with wider ones.
37840 2017-02-27 Richard Biener <rguenther@suse.de>
37842 PR tree-optimization/79690
37843 * tree-vect-stmts.c (vectorizable_store): Use vector type
37844 built from the DR with address-space.
37846 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
37848 * doc/invoke.texi (Optimize Options): Refine the description
37849 of asan-use-after-return.
37851 2017-02-25 Alan Modra <amodra@gmail.com>
37853 PR rtl-optimization/79584
37854 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37855 base, not ad->base_term, the reg within base. Remove assertion
37856 that ad->base == ad->base_term. Replace gen_int_mode using
37857 bogus mode with const0_rtx.
37859 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37861 PR middle-end/79396
37862 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37863 FMA_EXPR like tcc_binary or tcc_unary.
37865 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37868 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37870 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37871 (output_loc_operands): Handle DW_OP_call_ref and
37872 DW_OP_GNU_variable_value.
37873 (struct variable_value_struct): New type.
37874 (struct variable_value_hasher): Likewise.
37875 (variable_value_hash): New variable.
37876 (string_types): Remove.
37877 (copy_loc_descr): New function.
37878 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37879 (prepend_loc_descr_to_each): New function.
37880 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37881 instead of add_loc_descr_to_each if the first argument is single
37882 location list and the second has multiple.
37883 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37884 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37885 when looking for variable value which doesn't have other location info.
37886 (loc_list_from_tree): Formatting fix.
37887 (gen_array_type_die): Simplify DW_AT_string_length handling.
37888 (adjust_string_types): Remove.
37889 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37890 string_types. Call resolve_variable_values.
37891 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37892 (resolve_addr_in_expr): Likewise. Add A argument.
37893 (copy_deref_exprloc): Remove deref argument. Adjust for the
37894 original expression being DW_OP_GNU_variable_value with optionally
37895 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37896 optionally after it.
37897 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37898 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37899 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37900 (variable_value_hasher::hash, variable_value_hasher::equal): New
37902 (resolve_variable_value_in_expr, resolve_variable_value,
37903 resolve_variable_values, note_variable_value_in_expr,
37904 note_variable_value): New functions.
37905 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37908 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37911 * opts.h (handle_generated_option): Add GENERATED_P argument.
37912 * opts-common.c (handle_option): Adjust function comment.
37913 (handle_generated_option): Add GENERATED_P argument, pass it to
37915 (control_warning_option): Pass false to handle_generated_option
37917 * opts.c (maybe_default_option): Pass true to handle_generated_option
37919 * optc-gen.awk: Likewise.
37921 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37923 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37924 a REG, look at the REG it is a SUBREG of.
37925 (splitter for cmpeqsi_t): Ditto.
37927 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37929 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37930 the special USEs with the pattern of the insn, not the insn itself.
37932 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37935 * doc/invoke.texi: Document -mload-store-pairs.
37937 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37938 Sandra Loosemore <sandra@codesourcery.com>
37940 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37941 argument isn't a CONST_INT.
37942 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37943 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37944 (nios2_validate_compare): Bypass alternate compare logic if *op2
37945 is not a CONST_INT.
37946 (ldstwm_operation_p): Return false if first_base is not a REG or
37947 if first_offset is not a CONST_INT.
37949 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37951 * config/cris/cris.md: Use correct operand in a define_peephole2.
37953 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37955 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37957 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37959 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37960 this_insn if it is an INSN or JUMP_INSN.
37961 (force_offsettable): Look at base, not at addr.
37962 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37963 on things that aren't necessarily CONST_INTs.
37965 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37967 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37968 -mfpmath=sse is the default also for x86-32 targets with SSE2
37969 instruction set when @option{-ffast-math} is enabled
37971 2017-02-24 Jeff Law <law@redhat.com>
37973 PR rtl-optimizatoin/79286
37974 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37977 2017-02-24 Richard Biener <rguenther@suse.de>
37979 PR tree-optimization/79389
37980 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37983 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37985 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37986 function comment to reflect reality.
37987 (loop_exits_before_overflow): Fix typo in function description.
37989 2017-02-24 Richard Biener <rguenther@suse.de>
37991 PR tree-optimization/79389
37992 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37993 properly that a threading opportunity exists. Detect conditional
37994 copy/constant propagation opportunities.
37996 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37998 * config/visium/visium.md (type): Add trap.
37999 (b): New mode attribute.
38000 (*btst): Rename into...
38001 (*btst<mode>): ...this and adjust.
38002 (*cbranchsi4_btst_insn): Rename into...
38003 (*cbranch<mode>4_btst_insn): ...this and adjust.
38004 (trap): New define_insn.
38006 2017-02-23 Jakub Jelinek <jakub@redhat.com>
38008 PR tree-optimization/79389
38009 * ifcvt.c (struct noce_if_info): Add rev_cond field.
38010 (noce_reversed_cond_code): New function.
38011 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
38012 reversed_comparison_code. Formatting fix.
38013 (noce_try_store_flag): Test rev_cond != NULL in addition to
38014 reversed_comparison_code.
38015 (noce_try_store_flag_constants): Likewise.
38016 (noce_try_store_flag_mask): Likewise.
38017 (noce_try_addcc): Use rev_cond if non-NULL instead of
38018 reversed_comparison_code.
38019 (noce_try_cmove_arith): Likewise. Formatting fixes.
38020 (noce_try_minmax, noce_try_abs): Clear rev_cond.
38021 (noce_find_if_block): Initialize rev_cond.
38022 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
38023 instead of false as last argument never attempt to reverse it
38026 2017-02-23 Bin Cheng <bin.cheng@arm.com>
38028 PR tree-optimization/79663
38029 * tree-predcom.c (combine_chains): Process refs in reverse order
38030 only for ZERO length chains, and add explaining comment.
38032 2017-02-23 Jeff Law <law@redhat.com>
38034 PR tree-optimization/79578
38035 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
38036 in call to operand_equal_p.
38038 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
38041 * config/i386/cpuid.h: Fix another undefined behavior.
38043 2017-02-23 Richard Biener <rguenther@suse.de>
38045 PR tree-optimization/79683
38046 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
38047 vector types for data-refs.
38049 2017-02-23 Martin Liska <mliska@suse.cz>
38051 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
38053 2017-02-23 Jakub Jelinek <jakub@redhat.com>
38055 PR middle-end/79665
38056 * internal-fn.c (get_range_pos_neg): Moved to ...
38057 * tree.c (get_range_pos_neg): ... here. No longer static.
38058 * tree.h (get_range_pos_neg): New prototype.
38059 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
38060 are known to be in between 0 and signed maximum inclusive, try to
38061 expand both unsigned and signed divmod and use the cheaper one from
38064 2017-02-22 Jeff Law <law@redhat.com>
38066 PR tree-optimization/79578
38067 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
38068 to compare base operands.
38070 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
38073 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
38074 gpc_reg_operand instead of fpr_reg_operand.
38076 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
38078 * config/mips/mips.c (mips_return_in_memory): Force FP
38079 vector types to be returned in memory for o32 ABI.
38081 2017-02-22 Jakub Jelinek <jakub@redhat.com>
38083 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
38084 instead of DW_TAG_member for static data member declarations and don't
38085 set no_linkage_name for static inline data members.
38086 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
38089 2017-02-22 Martin Liska <mliska@suse.cz>
38091 * doc/invoke.texi: Replace inequality signs with square brackets
38094 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
38097 * lra-constraints.c (simplify_operand_subreg): Handle
38098 WORD_REGISTER_OPERATIONS targets.
38100 2017-02-22 Jakub Jelinek <jakub@redhat.com>
38103 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
38104 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
38105 elimination by swapping fld*.
38107 2017-02-22 Richard Biener <rguenther@suse.de>
38109 PR tree-optimization/79673
38110 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
38111 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
38112 irrelevant address-space qualifiers and avoiding a
38113 ADDR_SPACE_CONVERT_EXPR from fold_convert.
38115 2017-02-22 Richard Biener <rguenther@suse.de>
38117 PR tree-optimization/79666
38118 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
38119 to not symbolically negate if that may introduce undefined
38122 2017-02-22 Martin Liska <mliska@suse.cz>
38125 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
38126 * data-streamer-out.c (streamer_write_gcov_count_stream):
38128 * value-prof.c (stream_out_histogram_value): Make assert more
38129 precise based on type of counter.
38131 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
38134 * config/i386/i386.md (standard_x87sse_constant_load splitter):
38135 Use nonimmediate_operand instead of memory_operand for operand 1.
38136 (float-extend standard_x87sse_constant_load splitter): Ditto.
38138 2017-02-21 Jeff Law <law@redhat.com>
38140 PR tree-optimization/79621
38141 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
38142 blocks with edges to themselves.
38144 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38147 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
38148 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
38149 Use gimple_call_builtin_p.
38152 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
38153 on temporarily removed DEBUG_INSNs.
38155 PR tree-optimization/79649
38156 * tree-loop-distribution.c (classify_partition): Give up on
38157 non-generic address space loads/stores.
38159 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
38161 * doc/loop.texi (Loop manipulation): Remove nonexistent
38162 tree_ssa_loop_version from the documentation.
38163 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
38165 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38168 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
38169 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
38170 * config/rs6000/rs6000.c: Include except.h.
38171 (rs6000_expand_split_stack_prologue): Call
38172 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
38174 2017-02-21 Martin Jambor <mjambor@suse.cz>
38177 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
38178 have been analyzed.
38180 2017-02-21 Martin Jambor <mjambor@suse.cz>
38182 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
38183 for backward compatibility only.
38184 * doc/invoke.texi (Option Summary): Remove all references to
38185 -fipa-cp-alignment.
38187 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
38191 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38193 * lra-constraints.c (curr_insn_transform): Handle
38194 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38196 2017-02-21 Martin Liska <mliska@suse.cz>
38198 * config/i386/i386.opt: Replace -masm-dialect with -masm.
38200 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
38202 PR translation/79638
38203 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
38205 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
38208 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
38209 (arm_function_ok_for_sibcall): Return false for an indirect call by
38210 descriptor if all the argument registers are used.
38211 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
38212 alignment of the function.
38214 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38216 PR tree-optimization/61441
38217 * simplify-rtx.c (simplify_const_unary_operation): For
38218 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
38219 the sNaN unmodified.
38221 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
38223 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
38224 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
38225 instead of SYSTEM_HEADER_DIR.
38227 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
38228 Martin Liška <mliska@suse.cz>
38230 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
38231 Fix typos and grammar, use active voice, and clarify.
38233 2017-02-20 Marek Polacek <polacek@redhat.com>
38235 PR middle-end/79537
38236 * gimplify.c (gimplify_expr): Handle unused *&&L;.
38239 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
38241 2017-02-20 Jakub Jelinek <jakub@redhat.com>
38244 * config/i386/i386.c (ix86_expand_builtin): Handle
38245 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
38246 ix86_builtins_isa[fcode].isa as a requirement of those
38247 flags and any other flag in the bitmask.
38248 (ix86_init_mmx_sse_builtins): Use 0 instead of
38249 ~OPTION_MASK_ISA_64BIT as mask.
38250 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
38251 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
38252 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
38253 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
38255 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38258 * lra-constraints.c (split_reg): Check requested split mode
38259 is supported by the register.
38261 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38263 * lra-constraints.c (simplify_operand_subreg): Remove early
38266 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38269 * lra-constraints.c (curr_insn_transform): Tighten condition
38270 for converting SUBREG reloads from OP_OUT to OP_INOUT.
38272 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38275 * lra-constraints.c (curr_insn_transform): Handle
38276 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38278 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
38281 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
38283 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
38285 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
38288 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
38291 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38293 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
38296 2017-02-18 Jakub Jelinek <jakub@redhat.com>
38298 * final.c (last_columnnum, override_columnnum): New variables.
38299 (final_start_function): Set last_columnnum, pass it to begin_prologue
38300 hook and pass 0 to dwarf2out_begin_prologue.
38301 (final_scan_insn): Update override_columnnum. Pass last_columnnum
38302 to source_line debug hook.
38303 (notice_source_line): Compute last_columnnum and for debug_column_info
38304 return true on column changes.
38305 * debug.h (struct gcc_debug_hooks): Add column argument to
38306 source_line and begin_prologue hooks.
38307 (debug_nothing_int_charstar_int_bool): Remove prototype.
38308 (debug_nothing_int_int_charstar,
38309 debug_nothing_int_int_charstar_int_bool): New prototypes.
38310 (dwarf2out_begin_prologue): Add column argument.
38311 * debug.c (do_nothing_debug_hooks): Adjust source_line and
38312 begin_prologue hooks.
38313 (debug_nothing_int_charstar_int_bool): Remove.
38314 (debug_nothing_int_int_charstar,
38315 debug_nothing_int_int_charstar_int_bool): New functions.
38316 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
38317 through to dwarf2out_source_line.
38318 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
38319 (dwarf2out_source_line): Add column argument, emit it if requested.
38320 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
38322 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38323 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38324 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
38325 through to dwarf2out_begin_prologue.
38326 (vmsdbgout_source_line): Add column argument, pass it through to
38327 dwarf2out_source_line.
38328 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
38329 dbxout_source_line caller.
38330 (dbxout_source_line): Add column argument.
38332 * common.opt (gno-column-info, gcolumn-info): New options.
38333 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
38334 (check_die): Also test for multiple DW_AT_decl_column attributes.
38335 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
38336 DW_AT_decl_column if requested.
38337 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
38339 (gen_variable_die): Likewise.
38340 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
38341 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
38344 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
38345 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
38346 (ix86_handle_option): Handle OPT_m3dnowa.
38347 * doc/invoke.texi (-m3dnowa): Document.
38348 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
38349 -m3dnowa instead of -m3dnow -march=athlon.
38352 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
38353 instead of gcc_assert for K, r and R code checks. Formatting fixes.
38355 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38358 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
38359 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
38360 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
38361 generator for vsx_xxpermdi_<mode>_be.
38362 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
38363 force big-endian semantics.
38364 (vsx_xxpermdi_<mode>_be): New define_expand with same
38365 implementation as previous version of vsx_xxpermdi_<mode>.
38367 2017-02-17 Jakub Jelinek <jakub@redhat.com>
38369 PR tree-optimization/79327
38370 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
38371 variable, its initialization and use.
38373 2017-02-17 Julia Koval <julia.koval@intel.com>
38375 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
38376 (OPTION_MASK_ISA_PKU_UNSET): New.
38377 (ix86_handle_option): Handle -mrdpid.
38378 * config/i386/cpuid.h (bit_RDPID): New.
38379 * config/i386/driver-i386.c (host_detect_local_cpu):
38380 Detect RDPID feature.
38381 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
38382 * config/i386/i386-c.c (ix86_target_macros_internal):
38384 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
38385 (ix86_valid_target_attribute_inner_p): Add "rdpid".
38386 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
38387 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
38388 * config/i386/i386.md (define_insn "rdpid"): New.
38389 * config/i386/i386.opt Add -mrdpid.
38390 * config/i386/immintrin.h (_rdpid_u32): New.
38392 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
38394 PR rtl-optimization/79541
38395 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
38396 instead of transforming it into USE.
38398 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
38400 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
38401 If HONOR_SNANS (SFmode) force the input to a register.
38402 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
38403 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
38404 an frsp or similar insn.
38406 2017-02-17 Martin Liska <mliska@suse.cz>
38408 PR rtl-optimization/79577
38409 * params.def (selsched-max-sched-times): Increase minimum to 1.
38411 2017-02-17 Martin Liska <mliska@suse.cz>
38413 PR rtl-optimization/79574
38414 * gcse.c (want_to_gcse_p): Prevent integer overflow.
38416 2017-02-17 Martin Liska <mliska@suse.cz>
38418 PR tree-optimization/79529
38419 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
38420 ssa_defined_default_def_p to handle cases which are implicitly
38422 * tree-ssa.c (ssa_defined_default_def_p): New function.
38423 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
38424 which are implicitly defined.
38425 * tree-ssa.h (ssa_defined_default_def_p): Declare.
38427 2017-02-17 Richard Biener <rguenther@suse.de>
38429 PR middle-end/79576
38430 * params.def (max-ssa-name-query-depth): Limit to 10.
38432 2017-02-17 Richard Biener <rguenther@suse.de>
38434 PR tree-optimization/79552
38435 * tree-ssa-structalias.c (visit_loadstore): Properly verify
38438 2017-02-17 Richard Biener <rguenther@suse.de>
38441 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
38443 2017-02-17 Marek Polacek <polacek@redhat.com>
38445 PR middle-end/79536
38446 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
38447 (fold_negate_expr): New wrapper.
38449 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
38451 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
38452 Correct terminology and de-emphasize pre-standard behavior.
38454 2017-02-16 Alan Modra <amodra@gmail.com>
38456 PR rtl-optimization/79286
38457 * ira.c (def_dominates_uses): New function.
38458 (update_equiv_regs): Don't create an equivalence for insns that
38459 may trap where the register def does not dominate the use.
38461 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
38463 PR rtl-optimization/78127
38464 * lra.c (lra): Call lra_eliminate before finish the loop after
38467 2017-02-16 Richard Biener <rguenther@suse.de>
38469 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
38471 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
38472 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
38473 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
38474 (isl_val_int_from_wi): New function.
38475 (extract_affine_gmp): Rename to ...
38476 (extract_affine_wi): ... this, take a widest_int.
38477 (extract_affine_int): Just wrap extract_affine_wi.
38478 (add_param_constraints): Use isl_val_int_from_wi.
38479 (add_loop_constraints): Likewise, and extract_affine_wi.
38481 2017-02-15 Jeff Law <law@redhat.com>
38483 PR middle-end/79521
38484 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
38485 ira_init_register_move_cost_if_necessary.
38487 2017-02-15 Martin Sebor <msebor@redhat.com>
38489 PR middle-end/32003
38490 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
38491 removed in a prior commit.
38493 2017-02-15 Bin Cheng <bin.cheng@arm.com>
38495 PR tree-optimization/79347
38496 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
38497 counters during peeling.
38499 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
38501 * Makefile.in (site.exp): Remove "set ISLVER".
38503 2017-02-15 Jakub Jelinek <jakub@redhat.com>
38506 * real.c (real_from_integer): Call real_convert even for decimal.
38508 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
38511 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
38513 2017-02-14 Andrew Pinski <apinski@cavium.com>
38515 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
38516 cores and change the partno/implementer to be correct.
38517 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
38518 the 'B" as the implementer.
38519 * config/aarch64/aarch64-tune.md: Regenerate.
38521 2017-02-14 Carl Love <cel@us.ibm.com>
38523 * config/rs6000/rs6000.c: Add case statement entry to make the
38524 xvcvuxdsp built-in argument unsigned.
38525 * config/rs6000/vsx.md: Fix the source and return operand types so they
38526 match the instruction definitions from the ISA document. Fix typo
38527 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
38530 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
38533 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
38534 member early_clobber_alts.
38535 * lra-lives.c (reg_early_clobber_p): New.
38536 (process_bb_lives): Use it.
38537 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
38538 (debug_operand_data): Initialize early_clobber_alts.
38539 (setup_operand_alternative): Set up early_clobber_alts.
38540 (collect_non_operand_hard_regs): Ditto. Pass early clobber
38541 alternatives to new_insn_reg.
38542 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
38544 (lra_update_insn_regno_info): Pass the new arg.
38546 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38548 PR middle-end/79505
38549 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
38550 (new_oacc_loop_raw): Don't clear already cleared fields.
38553 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
38554 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
38555 _mm512_prefetch_i64gather_ps): New inline functions and macros.
38557 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
38560 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
38562 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
38565 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
38566 the extra instruction to the right place to store 128-bit constant
38569 2017-02-14 Martin Sebor <msebor@redhat.com>
38571 PR middle-end/79448
38572 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
38573 warning for strings of unknown length.
38575 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
38577 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
38579 2017-02-14 Jeff Law <law@redhat.com>
38582 * ira-costs.c (scan_one_insn): Initialize register move costs
38583 for pseudos seen in USE/CLOBBER insns.
38585 PR tree-optimization/79095
38586 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
38587 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
38588 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
38589 if the operands are known to be not equal, then the resulting range
38591 (intersect_ranges): If the new range is ~[0,0] and the old range is
38592 wide, then prefer ~[0,0].
38593 * tree-vrp.c (overflow_comparison_p_1): New function.
38594 (overflow_comparison_p): New function.
38595 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
38596 if NAME is used in an overflow test.
38597 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
38598 overflow check that can be expressed as an equality test, then adjust
38599 ops to be that equality test.
38601 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38603 * config/s390/s390-builtin-types.def: Remove flags argument.
38604 * config/s390/s390.c (s390_init_builtins): Likewise.
38606 2017-02-14 Martin Liska <mliska@suse.cz>
38608 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
38609 vector. Fix trailing white spaces.
38611 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
38613 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
38616 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38618 PR rtl-optimization/68664
38619 * config/arm/arm.c (arm_sched_can_speculate_insn):
38620 New function. Declare prototype.
38621 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38623 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38625 PR rtl-optimization/68664
38626 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
38628 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38630 2017-02-14 Amit Pawar <amit.pawar@amd.com>
38632 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
38633 max skip bytes for function, loop and jump.
38635 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38637 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
38638 ABS_EXPR for gimple dump.
38640 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38643 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
38645 PR tree-optimization/79408
38646 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
38647 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
38648 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
38649 also if rhs1 is INTEGER_CST.
38651 2017-02-14 Richard Biener <rguenther@suse.de>
38653 PR middle-end/79432
38654 * tree-into-ssa.c (insert_phi_nodes): When the function can
38655 have abnormal edges rewrite SSA names with broken use-def
38656 dominance out of SSA and register them for PHI insertion.
38658 2017-02-13 Martin Sebor <msebor@redhat.com>
38660 PR middle-end/79496
38661 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
38662 clearing info.nowrite flag when snprintf size argument is a range.
38664 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38666 * cprop.c (cprop_jump): Add missing space in string literal.
38667 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
38668 (get_constraint_for_component_ref): Likewise.
38669 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
38670 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
38671 * lra-constraints.c (process_alt_operands): Likewise.
38672 * ipa-inline.c (inline_small_functions): Likewise.
38673 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
38674 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
38675 * trans-mem.c (diagnose_tm_1_op): Likewise.
38676 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
38677 (grid_parallel_clauses_gridifiable): Likewise.
38679 * config/nvptx/mkoffload.c (process): Add space in between
38682 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
38683 "MOD4_SSE_REGS" and "ALL_REGS".
38685 * spellcheck.c (test_data): Add , in between "foo" and "food".
38687 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38690 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
38691 boundary crossing check and subsequent code generation agree.
38693 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38695 * config/aarch64/aarch64.c (has_memory_op): Delete.
38696 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
38699 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38701 PR rtl-optimization/79388
38702 PR rtl-optimization/79450
38703 * combine.c (distribute_notes): When removing TEM_INSN for which
38704 corresponding dest has last value recorded, invalidate that last
38707 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38709 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
38710 of explicit '@'. Add missing assembly comment marker on branch costs
38713 2017-02-13 Nathan Sidwell <nathan@acm.org>
38715 * gengtype-lex.l (<in_struct>): Add '/'.
38717 2017-02-13 Martin Liska <mliska@suse.cz>
38720 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
38722 2017-02-13 Richard Biener <rguenther@suse.de>
38724 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
38726 * configure: Re-generate.
38727 * config.in: Likewise.
38728 * graphite-dependences.c: Simplify as if
38729 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
38730 * graphite-isl-ast-to-gimple.c: Likewise.
38731 * graphite-optimize-isl.c: Likewise.
38732 * graphite-poly.c: Likewise.
38733 * graphite-sese-to-poly.c: Likewise.
38734 * graphite.h: Likewise.
38735 * toplev.c: Include isl/version.h and use isl_version () for
38736 printing the ISL version.
38737 * doc/install.texi: Update ISL requirement.
38739 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38741 * doc/standards.texi (Standards): Update reference to
38744 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38746 * doc/extend.texi (Named Address Spaces): sourceware.org now
38748 * doc/install.texi (Binaries): Ditto.
38751 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38753 * doc/cpp.texi: Replace "stringify"/"stringification" with C
38754 standard terminology "stringize"/"stringizing" throughout.
38755 * doc/cppinternals.texi: Likewise.
38757 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38759 * doc/extend.texi: Fix some spelling mistakes and typos.
38760 * doc/invoke.texi: Likewise.
38762 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38765 * params.def (inline-min-speedup) Change from 10 to 8.
38767 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38769 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
38772 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38775 * ipa-inline-analysis.c (get_minimal_bb): New function.
38776 (record_modified): Use it.
38777 (remap_edge_change_prob): Handle also ancestor functions.
38779 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
38781 * doc/contrib.texi (Contributors): Remove broken link into
38782 the Mauve CVS repository.
38784 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38786 PR middle-end/79454
38787 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
38788 result computation whenever lhs doesn't have vector mode, not
38789 just when it has BLKmode.
38791 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38793 * doc/makefile.texi (profiledbootstrap): Refer to the
38794 installation instructions only in textual form.
38796 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38799 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
38801 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38803 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
38804 (Specific): Update mingw-w64 reference.
38806 (Specific): Remove broken link to Renesas RX processor.
38808 2017-02-10 Richard Biener <rguenther@suse.de>
38810 * toplev.c (process_options): Do not mention obsolete graphite
38811 options when printing sorry message about missing graphite support.
38812 Mention -floop-nest-optimize.
38814 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
38816 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
38817 (vtst_p16): Likewise.
38818 (vtstq_p8): Likewise.
38819 (vtstq_p16): Likewise.
38821 (vtstq_p64): Likewise.
38822 * config/arm/arm_neon.h (vgetq_lane_p64): New.
38823 (vset_lane_p64): New.
38824 (vsetq_lane_p64): New.
38826 2017-02-10 Jakub Jelinek <jakub@redhat.com>
38828 PR tree-optimization/79411
38829 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
38830 stmt operands are SSA_NAMEs used in abnormal phis.
38831 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
38834 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38837 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
38840 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38842 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
38844 2017-02-09 Jakub Jelinek <jakub@redhat.com>
38846 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38850 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38851 not arbitrary TREE_CONSTANT.
38854 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38855 "omp declare target link" attribute unless is_global_var.
38856 * omp-offload.c (find_link_var_op): Likewise.
38858 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38859 Chung-Lin Tang <cltang@codesourcery.com>
38861 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38863 (gimplify_adjust_omp_clauses): Don't delete TILE.
38864 (gimplify_omp_for): Deal with TILE.
38865 * internal-fn.c (expand_GOACC_TILE): New function.
38866 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38868 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38869 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38871 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38872 avoid DIV for outermost collapse var.
38873 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38874 Remove out of date comments, fix whitespace.
38875 * omp-general.c (omp_extract_for_data): Deal with tiling.
38876 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38877 adjust OLF_DIM_BASE value.
38878 (struct omp_for_data): Add tiling field.
38879 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38880 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38881 for auto loops. Remove default auto determining, moved to
38882 oacc_loop_fixed_partitions.
38883 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38884 stmts, add e_mask field.
38885 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38886 (oacc_thread_numbers): Use oacc_dim_call.
38887 (oacc_xform_tile): New.
38888 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38889 (finish_oacc_loop): Adjust for ifns vector.
38890 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38891 add case for GOACC_TILE fns.
38892 (oacc_loop_xform_loop): Delete.
38893 (oacc_loop_process): Iterate over call list directly, and add
38894 handling for GOACC_TILE fns.
38895 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38897 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38898 vector partitioning to outer loops. Assign 2 partitions to loops
38899 when available. Add TILE handling.
38900 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38901 (execite_oacc_device_lower): Process GOACC_TILE fns,
38902 ignore unknown specs.
38903 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38904 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38905 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38907 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38909 * configure.ac (ACX_BUGURL): Update.
38910 * configure: Regenerate.
38912 2017-02-09 Richard Biener <rguenther@suse.de>
38914 PR tree-optimization/69823
38915 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38916 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38918 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38920 * config/arc/arc-c.def: Add __NPS400__ definition.
38921 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38922 (TARGET_NPS400): Define.
38924 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38926 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38928 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38929 pointer, arch_info.
38930 (arc_cpu_types): Fill the arch_info field with a pointer into the
38931 arc_arch_types table.
38932 (arc_selected_cpu): Declare.
38933 * config/arc/arc.c (arc_selected_cpu): Make global.
38934 (arc_selected_arch): Delete.
38935 (arc_base_cpu): Delete.
38936 (arc_override_options): Remove references to deleted variables,
38937 update access to arch information.
38938 (ARC_OPT): Update access to arch information.
38939 (ARC_OPTX): Likewise.
38940 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38941 (TARGET_ARC600): Update access to arch information.
38942 (TARGET_ARC601): Likewise.
38943 (TARGET_ARC700): Likewise.
38944 (TARGET_EM): Likewise.
38945 (TARGET_HS): Likewise.
38946 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38949 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38952 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38953 condition/operands for integer GE/LE/GEU/LEU operations.
38955 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38957 PR translation/79397
38958 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38961 2017-02-08 Martin Jambor <mjambor@suse.cz>
38964 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38965 whether allocation happened.
38966 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38967 nothing was allocated.
38969 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38971 PR tree-optimization/79408
38972 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38973 constant, but SSA_NAME with a known integer range, use the minimum
38974 of that range instead of op1 to determine if modulo can be replaced
38975 with its first operand.
38977 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38979 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38981 2017-02-08 Richard Biener <rguenther@suse.de>
38983 PR tree-optimization/71824
38984 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38985 Check all loops contained in the merged region.
38987 2017-02-07 Andrew Pinski <apinski@cavium.com>
38989 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38991 2017-02-07 Andrew Pinski <apinski@cavium.com>
38993 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38994 (thunderxt88): Likewise.
38995 (thunderxt81): Disable LSE and change v8.1 to v8.
38996 (thunderxt83): Likewise.
38998 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38999 Richard Biener <rguenther@suse.de>
39001 PR middle-end/79399
39002 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
39003 type from int to size_t.
39004 * ira-costs.c (struct_costs_size): Change type from int to size_t.
39006 2017-02-07 Jakub Jelinek <jakub@redhat.com>
39008 PR rtl-optimization/79386
39009 * cprop.c (bypass_conditional_jumps): Initialize
39010 bypass_last_basic_block already before splitting bbs after
39011 unconditional traps...
39012 (bypass_conditional_jumps): ... rather than here.
39015 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
39016 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
39017 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
39018 fix -masm=intel patterns.
39020 2017-02-07 Richard Biener <rguenther@suse.de>
39022 PR tree-optimization/79256
39023 PR middle-end/79278
39024 * builtins.c (get_object_alignment_2): Use min_align_of_type
39025 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
39026 and ADJUST_FIELD_ALIGN.
39028 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
39030 * doc/tm.texi: Regenerate.
39031 * stor-layout.c (layout_decl): Adjust.
39032 (update_alignment_for_field): Likewise.
39033 (place_field): Likewise.
39034 (min_align_of_type): Likewise.
39035 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
39036 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
39037 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
39038 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
39039 * config/frv/frv.c (frv_adjust_field_align): Likewise.
39040 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
39041 * config/i386/i386.c (x86_field_alignment): Likewise.
39042 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
39043 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
39044 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
39045 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
39046 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
39047 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
39051 2017-01-30 Richard Biener <rguenther@suse.de>
39053 PR tree-optimization/79256
39054 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
39055 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
39058 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
39060 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
39061 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
39062 builtins to SImode and emit a zero-extend, if necessary.
39064 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
39066 * docs/invoke.texi (RISC-V Options): Alphabetize.
39068 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
39070 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
39073 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
39075 * config/riscv/riscv.c: New file.
39076 * common/config/riscv/riscv-common.c: Likewise.
39077 * config.gcc: Likewise.
39078 * config/riscv/constraints.md: Likewise.
39079 * config/riscv/elf.h: Likewise.
39080 * config/riscv/generic.md: Likewise.
39081 * config/riscv/linux.h: Likewise.
39082 * config/riscv/multilib-generator: Likewise.
39083 * config/riscv/peephole.md: Likewise.
39084 * config/riscv/pic.md: Likewise.
39085 * config/riscv/predicates.md: Likewise.
39086 * config/riscv/riscv-builtins.c: Likewise.
39087 * config/riscv/riscv-c.c: Likewise.
39088 * config/riscv/riscv-ftypes.def: Likewise.
39089 * config/riscv/riscv-modes.def: Likewise.
39090 * config/riscv/riscv-opts.h: Likewise.
39091 * config/riscv/riscv-protos.h: Likewise.
39092 * config/riscv/riscv.h: Likewise.
39093 * config/riscv/riscv.md: Likewise.
39094 * config/riscv/riscv.opt: Likewise.
39095 * config/riscv/sync.md: Likewise.
39096 * config/riscv/t-elf-multilib: Likewise.
39097 * config/riscv/t-linux: Likewise.
39098 * config/riscv/t-linux-multilib: Likewise.
39099 * config/riscv/t-riscv: Likewise.
39100 * configure.ac: Likewise.
39101 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
39102 Waterman as RISC-V maintainers.
39103 * doc/install.texi: Add RISC-V entries.
39104 * doc/invoke.texi: Add RISC-V options section.
39105 * doc/md.texi: Add RISC-V constraints section.
39106 * configure: Regenerated.
39108 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
39111 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
39112 false values to be constant vectors with all 0 or all 1 bits set.
39113 (vcondu<mode><mode>): Likewise.
39114 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
39116 (fpmask_comparison_operator): Update comment.
39117 (vecint_comparison_operator): New predicate.
39118 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
39119 vector conditionals when the true and false values are constant
39120 vectors with all 0 bits or all 1 bits set.
39122 2017-02-06 Martin Sebor <msebor@redhat.com>
39124 PR tree-optimization/79376
39125 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
39127 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
39129 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
39130 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
39131 to simplify split condition.
39133 2017-02-06 Jakub Jelinek <jakub@redhat.com>
39135 * omp-expand.c (oxpand_omp_atomic_fetch_op,
39136 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
39139 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
39141 PR rtl-optimization/68664
39142 * target.def (can_speculate_insn): New hook.
39143 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
39144 * doc/tm.texi: Regenerate.
39145 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
39146 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
39147 (rs6000_sched_can_speculate_insn): New function.
39149 2017-02-06 Jakub Jelinek <jakub@redhat.com>
39151 PR tree-optimization/79284
39152 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
39153 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
39154 vectorizable_mask_load_store, vectorizable_operation,
39155 vect_is_simple_cond, get_same_sized_vectype): Use it instead
39156 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
39157 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
39158 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
39159 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
39160 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
39161 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
39162 is_gimple_assign (stmt). Replace another such test with
39163 is_gimple_assign (stmt).
39165 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
39168 * config/avr/avr.c (rtl-iter.h): Include it.
39169 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
39170 (avr_legitimate_combined_insn): ...and implementation.
39172 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
39174 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
39175 * config/s390/s390.c (s390_const_operand_ok)
39176 (s390_canonicalize_comparison, s390_extract_part)
39177 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
39178 (s390_contiguous_bitmask_p, s390_rtx_costs)
39179 (legitimize_pic_address): Likewise.
39180 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
39181 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
39182 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
39183 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
39184 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
39186 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
39188 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
39189 REGNO($0) == REGNO($1).
39191 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39193 * config/s390/linux.h(SIZE_TYPE): Add comment.
39195 2017-02-06 Julian Brown <julian@codesourcery.com>
39196 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39197 Virendra Pathak <virendra.pathak@broadcom.com>
39199 * config/aarch64/aarch64-cores.def: Change the scheduler
39201 * config/aarch64/aarch64.md: Include thunderx2t99.md.
39202 * config/aarch64/thunderx2t99.md: New file.
39204 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39206 * doc/standards.texi (Go Language): Update link to language
39209 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
39211 * tree-eh.c (lower_resx): Sanitize profile.
39212 (cleanup_empty_eh_move_lp): Likewise.
39214 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
39217 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
39219 * cfgloopmanip.h (loop_version): Update prototype.
39220 * modulo-sched.c (sms_schedule): Update call of loop_version.
39221 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
39222 * tree-parloops.c (gen_parallel_loop): Likewise.
39223 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
39224 * tree-ssa-loop-split.c (split_loop): Likewise.
39225 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
39226 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
39228 2017-02-05 Martin Liska <mliska@suse.cz>
39231 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
39233 (print_operand_address): Initialize a struct to zero.
39235 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39237 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
39238 garbage collector only in textual form.
39240 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39242 * doc/extend.texi (x86 specific memory model extensions for
39243 transactional memory): Simplify a phrase.
39245 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
39248 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
39249 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
39250 (atomic_storedi_1): Likewise.
39252 2017-02-04 Jakub Jelinek <jakub@redhat.com>
39254 PR tree-optimization/79338
39255 * tree-parloops.c (gather_scalar_reductions): Don't call
39256 vect_analyze_loop_form for loop->inner before destroying loop's
39259 2017-02-03 Martin Sebor <msebor@redhat.com>
39261 PR tree-optimization/79327
39262 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
39263 when precision has resulted in leading zeros.
39264 (format_integer): Adjust the likely counter to assume an unknown
39265 argument that may be zero is non-zero.
39267 2017-02-03 Jason Merrill <jason@redhat.com>
39270 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
39271 avoid copying non-taken branch.
39273 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39275 PR tree-optimization/79340
39276 * tree-vect-loop.c (vectorizable_reduction): Release
39277 vec_defs elements after safe_splicing them into other vectors.
39280 PR tree-optimization/79327
39281 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
39282 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
39284 (format_integer): Use wide_int_to_tree instead of build_int_cst
39285 + to_?hwi. If argmin is NULL, just set argmin and argmax to
39286 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
39287 of shortest and longest sequence.
39289 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
39291 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
39292 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
39294 2017-02-03 Walter Lee <walt@tilera.com>
39297 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
39298 after initial stackframe link reg save.
39299 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
39301 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39304 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
39305 wu for stxssp alternative.
39307 2017-02-03 Martin Sebor <msebor@redhat.com>
39309 PR tree-optimization/79352
39310 * gimple-fold.c (get_range_strlen): Add argument.
39311 (get_range_strlen): Change return type to bool.
39312 (get_maxval_strlen): Pass in a dummy argument.
39313 * gimple-fold.h (get_range_strlen): Change return type to bool.
39314 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
39315 * tree.h (array_at_struct_end_p): Add argument.
39316 * tree.c (array_at_struct_end_p): Handle it.
39318 2017-02-03 Martin Liska <mliska@suse.cz>
39321 * multiple_target.c (create_dispatcher_calls): Redirect edge
39322 from a caller of a dispatcher.
39323 (expand_target_clones): Make the clones local.
39324 (ipa_target_clone): Do both target clones and resolvers.
39325 (ipa_dispatcher_calls): Remove the pass.
39326 (pass_dispatcher_calls::gate): Likewise.
39327 (make_pass_dispatcher_calls): Likewise.
39328 * passes.def (pass_target_clone): Put as very first IPA early
39331 2017-02-03 Martin Liska <mliska@suse.cz>
39333 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
39334 in case of a function with ifunc attribute.
39336 2017-02-03 Martin Liska <mliska@suse.cz>
39338 * cgraph.c (cgraph_node::dump): Dump function version info.
39339 * symtab.c (symtab_node::dump_base): Add missing new line.
39341 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39343 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
39344 (ifcombine_ifandif): Use it.
39346 2017-02-03 Martin Liska <mliska@suse.cz>
39348 * doc/invoke.texi: Document default value for
39349 use-after-scope-direct-emission-threshold.
39351 2017-02-03 Martin Liska <mliska@suse.cz>
39353 PR tree-optimization/79339
39354 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
39355 (format_floating): Likewise.
39357 2017-02-03 Martin Liska <mliska@suse.cz>
39360 * ipa-prop.c (ipa_node_params_t::insert): Remove current
39362 (ipa_node_params_t::remove): Likewise.
39363 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
39364 initialization from removed ipa_node_params_t::insert.
39365 (ipa_node_params::~ipa_node_params): Move from removed
39366 ipa_node_params_t::release.
39367 * symbol-summary.h (symbol_summary::m_released): New member.
39368 Do not release a summary twice. Do not allow to call finalizer
39369 for types of a summary that live in GGC memory.
39371 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39373 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
39376 2017-02-02 Martin Sebor <msebor@redhat.com>
39378 PR middle-end/79275
39379 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
39380 (format_string): Tighten up the range of output for non-constant
39381 strings and correct the expected range for wide non-constant strings.
39383 2017-02-02 Martin Sebor <msebor@redhat.com>
39385 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
39387 PR middle-end/32003
39388 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
39390 (-fdump-tree-@var): Add to index and document how to come up
39391 with pass-specific option and dump file names.
39392 (-fdump-passes): Clarify where to look for output.
39394 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39396 PR middle-end/77445
39397 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
39398 statistics of the analyzed path; allow threading for speed when
39399 any of BBs along the path are optimized for speed.
39401 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
39403 PR middle-end/78468
39404 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
39405 settings of the virtual registers.
39408 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
39410 * explow.c (get_dynamic_stack_size): Take known alignment of stack
39411 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
39414 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39416 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
39417 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
39419 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39421 * config/s390/s390.md: Add missing comments with the expanded
39423 * config/s390/vector.md: Likewise.
39424 * config/s390/vx-builtins.md: Likewise.
39426 2017-02-02 Jakub Jelinek <jakub@redhat.com>
39429 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
39430 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
39431 conditions on a single line.
39433 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39435 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39436 __S390_VX__ to __VX__.
39438 2017-02-01 Andrew Pinski <apinski@cavium.com>
39440 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
39441 stmt_info to record_stmt_cost.
39442 (vect_get_known_peeling_cost): Pass stmt_info if known to
39444 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
39445 cpu_vector_cost field into
39446 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
39447 field into vec_int_stmt_cost and vec_fp_stmt_cost.
39448 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
39449 splitting of scalar_stmt_cost and vec_stmt_cost.
39450 (thunderx_vector_cost): Likewise.
39451 (cortexa57_vector_cost): LIkewise.
39452 (exynosm1_vector_cost): Likewise.
39453 (xgene1_vector_cost): Likewise.
39454 (thunderx2t99_vector_cost): Improve after the splitting of the two
39456 (aarch64_builtin_vectorization_cost): Update for the splitting of
39457 scalar_stmt_cost and vec_stmt_cost.
39459 2017-02-01 Torvald Riegel <triegel@redhat.com>
39460 Richard Henderson <rth@redhat.com>
39462 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
39463 conditional on existance of a fast atomic load.
39464 * optabs-query.c (can_atomic_load_p): New function.
39465 * optabs-query.h (can_atomic_load_p): Declare it.
39466 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
39467 no fast atomic load is available for the particular size of access.
39468 (expand_atomic_compare_and_swap): Likewise.
39469 (expand_atomic_load): Likewise.
39470 (expand_atomic_store): Likewise.
39471 (expand_atomic_fetch_op): Likewise.
39472 * testsuite/lib/target-supports.exp
39473 (check_effective_target_sync_int_128): Remove x86 because it provides
39474 no fast atomic load.
39475 (check_effective_target_sync_int_128_runtime): Likewise.
39477 2017-02-01 Richard Biener <rguenther@suse.de>
39479 * graphite.c: Include tree-vectorizer.h for find_loop_location.
39480 (graphite_transform_loops): Provide opt-info for optimized nests.
39481 * tree-parloop.c (parallelize_loops): Provide opt-info for
39482 parallelized loops.
39484 2017-02-01 Richard Biener <rguenther@suse.de>
39486 PR middle-end/79315
39487 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
39488 was not set before.
39490 2017-02-01 Richard Biener <rguenther@suse.de>
39492 PR tree-optimization/71824
39493 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
39494 Verify the loops are valid in the merged SESE region.
39495 (scop_detection::can_represent_loop_1): Check analyzing the
39496 evolution of the number of iterations in the region succeeds.
39498 2017-01-31 Ian Lance Taylor <iant@golang.org>
39500 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
39501 REG_ARGS_SIZE note to 32-bit push insns and call insn.
39503 2017-01-31 David Malcolm <dmalcolm@redhat.com>
39505 PR preprocessor/79210
39506 * input.c (get_substring_ranges_for_loc): Replace line_width
39507 assertion with error-handling.
39509 2017-01-31 Richard Biener <rguenther@suse.de>
39511 PR tree-optimization/77318
39512 * graphite-sese-to-poly.c (extract_affine): Fix assert.
39513 (create_pw_aff_from_tree): Take loop parameter.
39514 (add_condition_to_pbb): Pass loop of the condition to
39515 create_pw_aff_from_tree.
39517 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39519 * config/s390/s390.c (s390_asan_shadow_offset): New function.
39520 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
39522 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
39526 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
39528 (convert_int_to_float128): Likewise.
39529 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
39530 (convert_int_to_float128): Likewise.
39531 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
39532 (UNSPEC_IEEE128_CONVERT): Likewise.
39533 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
39534 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
39535 Use local variables for IBM extended format.
39536 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
39537 (fix_trunc<mode>si2_fprs): Likewise.
39538 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
39539 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
39540 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
39541 to know that we can now have integers of all sizes in vector
39543 (fix<uns>_<mode>di2_hw): Likewise.
39544 (float<uns>_<mode>si2_hw): Likewise.
39545 (fix_<mode>si2_hw): Likewise.
39546 (fixuns_<mode>si2_hw): Likewise.
39547 (float<uns>_<mode>di2_hw): Likewise.
39548 (float_<mode>di2_hw): Likewise.
39549 (float_<mode>si2_hw): Likewise.
39550 (floatuns_<mode>di2_hw): Likewise.
39551 (floatuns_<mode>si2_hw): Likewise.
39552 (xscvqp<su>wz_<mode>): Delete, no longer used.
39553 (xscvqp<su>dz_<mode>): Likewise.
39554 (xscv<su>dqp_<mode>): Likewise.
39555 (ieee128_mfvsrd_64bit): Likewise.
39556 (ieee128_mfvsrd_32bit): Likewise.
39557 (ieee128_mfvsrwz): Likewise.
39558 (ieee128_mtvsrw): Likewise.
39559 (ieee128_mtvsrd_64bit): Likewise.
39560 (ieee128_mtvsrd_32bit): Likewise.
39562 2017-01-31 Martin Liska <mliska@suse.cz>
39565 * ipa-prop.c (ipa_free_all_node_params): Call release method
39566 instead of ~sumbol_summary to not to trigger double times
39569 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
39571 PR tree-optimization/71691
39572 * bitmap.h (class auto_bitmap): New.
39573 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
39574 is_maybe_undefined instead of ssa_undefined_value_p.
39576 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39578 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39579 __S390_ARCH_LEVEL__ to __ARCH__.
39581 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39583 PR tree-optimization/79267
39584 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
39585 if should_remove_lhs_p is true.
39587 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
39590 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
39591 (add_alignment_attribute): New.
39592 (base_type_die): Add alignment attribute.
39593 (subrange_type_die): Likewise.
39594 (modified_type_die): Likewise.
39595 (gen_array_type_die): Likewise.
39596 (gen_descr_array_type_die: Likewise.
39597 (gen_enumeration_type_die): Likewise.
39598 (gen_subprogram_die): Likewise.
39599 (gen_variable_die): Likewise.
39600 (gen_field_die): Likewise.
39601 (gen_ptr_to_mbr_type_die): Likewise.
39602 (gen_struct_or_union_type_die): Likewise.
39603 (gen_subroutine_type_die): Likewise.
39604 (gen_typedef_die): Likewise.
39605 (base_type_cmp): Compare alignment attribute.
39607 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39610 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
39611 (setb_unsigned) New pattern for setb with CCUNS.
39612 * config/rs6000/rs6000.c (expand_block_compare): Use a different
39613 subfc./subfe sequence to avoid overflow problems. Generate a
39614 shorter sequence with cmpld/setb for power9.
39615 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
39616 for generating subfc. instruction.
39617 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
39618 now uses this instruction.
39620 2017-01-30 Ian Lance Taylor <iant@google.com>
39623 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
39624 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
39626 2017-01-30 Martin Sebor <msebor@redhat.com>
39628 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
39629 Move constant to the right of a relational operator.
39630 (get_mpfr_format_length, format_character, format_string): Ditto.
39631 (should_warn_p, maybe_warn): Same.
39633 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
39635 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
39638 * asan.c (get_translation_unit_decl): Remove function.
39639 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
39641 2017-01-30 Martin Liska <mliska@suse.cz>
39643 PR gcov-profile/79259
39644 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
39645 -fprofile-generate.
39647 2017-01-30 Martin Liska <mliska@suse.cz>
39650 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
39651 Initialize variables with NULL value.
39653 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
39656 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
39658 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
39660 2017-01-30 Richard Biener <rguenther@suse.de>
39662 PR tree-optimization/79276
39663 * tree-vrp.c (process_assert_insertions): Properly adjust common
39664 when removing a duplicate.
39666 2017-01-30 Richard Biener <rguenther@suse.de>
39668 PR tree-optimization/79256
39669 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
39670 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
39672 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
39674 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
39677 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
39678 ("*r<noxa>sbg_<mode>_sll_bitmask")
39679 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
39680 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
39681 Use contiguous_bitmask_nowrap_operand.
39683 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39686 * config/rs6000/altivec.h (vec_xl): Revise #define.
39687 (vec_xst): Likewise.
39689 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
39691 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
39693 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
39695 PR rtl-optimization/79194
39696 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
39697 traps before call to bypass_conditional_jumps.
39699 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39701 PR tree-optimization/71374
39702 * lra-constraints.c (check_conflict_input_operands): New.
39703 (match_reload): Use it.
39705 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39708 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
39709 account to calculate conflict_set.
39711 2017-01-27 Bin Cheng <bin.cheng@arm.com>
39713 PR rtl-optimization/78559
39714 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
39715 other_insn in combine.
39717 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
39719 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
39720 uint16_type_node for BT_UINT16.
39722 2017-01-27 David Malcolm <dmalcolm@redhat.com>
39724 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
39725 "RTL Tests" to menu.
39726 (GIMPLE Tests): New node.
39727 (RTL Tests): New node.
39729 2017-01-27 Richard Biener <rguenther@suse.de>
39731 PR tree-optimization/79245
39732 * tree-loop-distribution.c (distribute_loop): Apply cost
39733 modeling also to detected patterns.
39735 2017-01-27 Richard Biener <rguenther@suse.de>
39737 PR tree-optimization/71433
39738 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
39739 (compare_assert_loc): New function.
39740 (process_assert_insertions): Sort and optimize assert locations
39741 to remove duplicates and push down identical assertions on
39742 edges to their destination block.
39744 2017-01-27 Richard Biener <rguenther@suse.de>
39746 PR tree-optimization/79244
39747 * tree-vrp.c (remove_range_assertions): Forcefully propagate
39748 out SSA names even if abnormal.
39750 2017-01-27 Jakub Jelinek <jakub@redhat.com>
39752 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
39753 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
39754 instead of MPFR_RNDN.
39756 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
39759 * arm.c (arm_option_override): Don't call build_target_option_node
39760 until after doing all option overrides.
39761 (arm_valid_target_attribute_tree): Likewise.
39763 2017-01-27 Martin Liska <mliska@suse.cz>
39765 * doc/invoke.texi (-fprofile-arcs): Document profiling support
39766 for {cd}tors and C++ {cd}tors.
39768 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
39770 * config/s390/s390.md ("*setmem_long_and")
39771 ("*setmem_long_and_31z"): Use zero_extend instead of and.
39773 2017-01-26 Martin Sebor <msebor@redhat.com>
39775 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
39778 2017-01-26 Martin Sebor <msebor@redhat.com>
39780 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
39781 HAVE_DFmode before using XFmode or DFmode.
39782 (parse_directive): Avoid using the z length modifier to avoid
39783 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
39785 PR middle-end/78703
39786 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
39787 to accept adjustment as an array.
39788 (get_int_range): New function.
39789 (struct directive): Make width and prec arrays.
39790 (directive::set_width, directive::set_precision): Call get_int_range.
39791 (format_integer, format_floating): Handle width and precision ranges.
39792 (format_string, parse_directive): Same.
39794 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39797 * dwarf2out.c (generate_skeleton_bottom_up): For children with
39798 comdat_type_p set, just clone them, but keep the children in the
39802 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
39803 which have direct callers with -fvar-tracking-assignments enabled
39805 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
39806 inside of type units.
39808 2017-01-26 Martin Sebor <msebor@redhat.com>
39810 PR middle-end/78703
39811 * gimple-ssa-sprintf.c (struct result_range): Add likely and
39813 (struct format_result): Replace number_chars, number_chars_min,
39814 and number_chars_max with a single member of struct result_range.
39816 (format_result::operator+=): Adjust.
39817 (struct fmtresult): Remove bounded. Handle likely and unlikely
39819 (fmtresult::adjust_for_width_or_precision): New function.
39820 (fmtresult:type_max_digits): New function.
39821 (bytes_remaining): Handle likely and unlikely counters.
39822 (min_bytes_remaining): Remove.
39823 (format_percent): Simplify.
39824 (format_integer, format_floating): Set likely and unlikely counters.
39825 (get_string_length, format_character, format_string): Same.
39826 (format_plain, should_warn_p): New function.
39827 (maybe_warn): Call should_warn_p. Update diagnostic messages
39828 and handle those for all directives, including plain strings.
39829 (format_directive): Handle likely and unlikely counters.
39830 Remove unnecessary quoting from diagnostics. Add an informational
39832 (add_bytes): Remove.
39833 (pass_sprintf_length::compute_format_length): Simplify.
39834 (try_substitute_return_value): Handle likely and unlikely counters.
39836 2017-01-26 Carl Love <cel@us.ibm.com>
39838 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
39839 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
39841 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
39844 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
39845 endianess for subregs into account.
39846 * lra-constraints.c (lra_constraints): Do risky transformations
39847 always on the first iteration.
39848 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39849 last_call_used_reg_set.
39850 (process_bb_lives): Define and use last_call_used_reg_set.
39851 * lra.c (lra): Always continue after lra_constraints on the first
39854 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39856 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39858 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39860 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39862 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39863 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39864 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39865 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39866 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39867 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39868 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39869 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39870 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39872 2017-01-26 Marek Polacek <polacek@redhat.com>
39875 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39876 for the third operand.
39878 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39880 PR middle-end/79236
39881 * omp-low.c (struct omp_context): Add simt_stmt field.
39882 (scan_omp_for): Return omp_context *.
39883 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39884 context to the _simt_ SIMD stmt.
39885 (lower_omp_for): For combined SIMD with sibling _simt_
39886 SIMD, make sure to use the same decls in _looptemp_
39887 clauses as in the sibling.
39889 2017-01-26 David Sherwood <david.sherwood@arm.com>
39891 PR middle-end/79212
39892 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39895 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39898 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39899 emit fld b; fld a; if possible.
39901 * brig-builtins.def: Update copyright years.
39902 * config/arm/arm_acle_builtins.def: Update copyright years.
39904 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39907 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39908 constraint instead of o for the stxsd instruction.
39910 2017-01-25 Carl Love <cel@us.ibm.com>
39912 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39913 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39915 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39917 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39919 2017-01-25 Richard Biener <rguenther@suse.de>
39921 PR tree-optimization/69264
39922 * target.def (vector_alignment_reachable): Improve documentation.
39923 * doc/tm.texi: Regenerate.
39924 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39926 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39927 earlier changes with respect to TYPE_USER_ALIGN.
39928 (vector_alignment_reachable_p): Likewise. Improve dumping.
39930 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39933 * config/arm/arm.md (xordi3): Force constant operand into a register
39936 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39938 * doc/invoke.texi (-fstore-merging): Correct default optimization
39939 levels at which it is enabled.
39940 (-O): Move -fstore-merging from list to...
39943 2017-01-25 Richard Biener <rguenther@suse.de>
39946 * omp-expand.c: Include debug.h.
39947 (expand_omp_taskreg): Make sure to generate early debug before
39948 outlining anything from a function.
39949 (expand_omp_target): Likewise.
39950 (grid_expand_target_grid_body): Likewise.
39952 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39955 * asan.c (get_translation_unit_decl): New function.
39956 (asan_add_global): Extract modules file name from globals
39957 TRANSLATION_UNIT_DECL name.
39959 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39962 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39963 for long calls with APCS frame and VFP.
39965 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39967 * cfg.c (original_copy_tables_initialized_p): New function.
39968 * cfg.h (original_copy_tables_initialized_p): New decl.
39969 * cfgrtl.c (relink_block_chain): Guard the call to
39970 free_original_copy_tables with a call to
39971 original_copy_tables_initialized_p.
39972 * cgraph.h (symtab_node::native_rtl_p): New decl.
39973 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39974 (symtab_node::needed_p): Don't assert for early assembly output
39975 for __RTL functions.
39976 (cgraph_node::finalize_function): Set "force_output" for __RTL
39978 (cgraph_node::analyze): Bail out early for __RTL functions.
39979 (analyze_functions): Update assertion to support __RTL functions.
39980 (cgraph_node::expand): Bail out early for __RTL functions.
39981 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39983 * function.h (struct function): Update comment for field
39985 * gimple-expr.c: Include "tree-pass.h".
39986 (gimple_has_body_p): Return false for __RTL functions.
39987 * Makefile.in (OBJS): Add run-rtl-passes.o.
39988 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39990 (gcc::pass_manager::get_clean_slate): New accessor.
39991 * passes.c: Include "insn-addr.h".
39992 (should_skip_pass_p): Add logging. Update logic for running
39993 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39994 property-provider override so it is only done for gimple passes.
39996 (skip_pass): New function.
39997 (execute_one_pass): Call skip_pass when skipping passes.
39998 * read-md.c (md_reader::read_char): Support filtering
39999 the input to a subset of line numbers.
40000 (md_reader::md_reader): Initialize fields
40001 m_first_line and m_last_line.
40002 (md_reader::read_file_fragment): New function.
40003 * read-md.h (md_reader::read_file_fragment): New decl.
40004 (md_reader::m_first_line): New field.
40005 (md_reader::m_last_line): New field.
40006 * read-rtl-function.c (function_reader::create_function): Only
40007 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
40008 curr_properties. Set DECL_INITIAL to a dummy block.
40009 (read_rtl_function_body_from_file_range): New function.
40010 * read-rtl-function.h (read_rtl_function_body_from_file_range):
40012 * run-rtl-passes.c: New file.
40013 * run-rtl-passes.h: New file.
40015 2017-01-24 Jeff Law <law@redhat.com>
40017 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
40020 2017-01-24 Bin Cheng <bin.cheng@arm.com>
40022 PR tree-optimization/79159
40023 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
40024 (record_nonwrapping_iv): Improve boundary using above function if no
40025 value range information.
40027 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
40028 Martin Jambor <mjambor@suse.cz>
40030 * brig-builtins.def: New file.
40031 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
40032 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
40033 (DEF_HSAIL_SAT_BUILTIN): Likewise.
40034 (DEF_HSAIL_INTR_BUILTIN): Likewise.
40035 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
40036 * builtin-types.def (BT_INT8): New.
40037 (BT_INT16): Likewise.
40038 (BT_UINT8): Likewise.
40039 (BT_UINT16): Likewise.
40040 (BT_FN_ULONG): Likewise.
40041 (BT_FN_UINT_INT): Likewise.
40042 (BT_FN_UINT_ULONG): Likewise.
40043 (BT_FN_UINT_LONG): Likewise.
40044 (BT_FN_UINT_PTR): Likewise.
40045 (BT_FN_ULONG_PTR): Likewise.
40046 (BT_FN_INT8_FLOAT): Likewise.
40047 (BT_FN_INT16_FLOAT): Likewise.
40048 (BT_FN_UINT32_FLOAT): Likewise.
40049 (BT_FN_UINT16_FLOAT): Likewise.
40050 (BT_FN_UINT8_FLOAT): Likewise.
40051 (BT_FN_UINT64_FLOAT): Likewise.
40052 (BT_FN_UINT16_UINT32): Likewise.
40053 (BT_FN_UINT32_UINT16): Likewise.
40054 (BT_FN_UINT16_UINT16_UINT16): Likewise.
40055 (BT_FN_INT_PTR_INT): Likewise.
40056 (BT_FN_UINT_PTR_UINT): Likewise.
40057 (BT_FN_LONG_PTR_LONG): Likewise.
40058 (BT_FN_ULONG_PTR_ULONG): Likewise.
40059 (BT_FN_VOID_UINT64_UINT64): Likewise.
40060 (BT_FN_UINT8_UINT8_UINT8): Likewise.
40061 (BT_FN_INT8_INT8_INT8): Likewise.
40062 (BT_FN_INT16_INT16_INT16): Likewise.
40063 (BT_FN_INT_INT_INT): Likewise.
40064 (BT_FN_UINT_FLOAT_UINT): Likewise.
40065 (BT_FN_FLOAT_UINT_UINT): Likewise.
40066 (BT_FN_ULONG_UINT_UINT): Likewise.
40067 (BT_FN_ULONG_UINT_PTR): Likewise.
40068 (BT_FN_ULONG_ULONG_ULONG): Likewise.
40069 (BT_FN_UINT_UINT_UINT): Likewise.
40070 (BT_FN_VOID_UINT_PTR): Likewise.
40071 (BT_FN_UINT_UINT_PTR: Likewise.
40072 (BT_FN_UINT32_UINT64_PTR): Likewise.
40073 (BT_FN_INT_INT_UINT_UINT): Likewise.
40074 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
40075 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
40076 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
40077 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
40078 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
40079 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
40080 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
40081 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
40082 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
40083 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
40084 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
40085 * doc/frontends.texi: List BRIG FE.
40086 * doc/install.texi (Testing): Add BRIG tesring requirements.
40087 * doc/invoke.texi (Overall Options): Mention BRIG.
40088 * doc/standards.texi (Standards): Doucment BRIG HSA version.
40090 2017-01-24 Richard Biener <rguenther@suse.de>
40092 PR translation/79208
40093 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
40095 2017-01-24 Martin Jambor <mjambor@suse.cz>
40098 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
40099 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
40100 and known_contexts.
40102 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
40104 PR middle-end/79123
40105 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
40106 casts from signed to unsigned really don't have a range.
40108 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
40110 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
40111 GMP_RNDx for compatiblity.
40113 2017-01-24 Martin Liska <mliska@suse.cz>
40116 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
40117 that would prevent us to call alloca with -1 as argument.
40119 2017-01-24 Jakub Jelinek <jakub@redhat.com>
40121 * dwarf2out.c (output_compilation_unit_header, output_file_names):
40122 Avoid -Wformat-security warning.
40124 2017-01-23 Andrew Pinski <apinski@cavium.com>
40126 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
40129 2017-01-23 Martin Sebor <msebor@redhat.com>
40131 PR middle-end/78703
40132 * gimple-ssa-sprintf.c (warn_level): New global.
40133 (format_integer): Use it here and throughout the rest of the file.
40134 Use the same switch to compute sign as base.
40135 (maybe_warn): New function.
40136 (format_directive): Factor out warnings into maybe_warn.
40137 Add debugging output. Use warn_level.
40138 (add_bytes): Use warn_level.
40139 (pass_sprintf_length::compute_format_length): Add debugging output.
40140 (try_substitute_return_value): Same.
40141 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
40143 PR middle-end/78703
40144 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
40145 (struct fmtresult, format_integer, format_floating): Adjust.
40146 (fmtresult::fmtresult): Set max correctly in two argument ctor.
40147 (get_string_length, format_string,format_directive): Same.
40148 (pass_sprintf_length::compute_format_length): Same.
40149 (try_substitute_return_value): Simplify slightly.
40151 PR middle-end/78703
40152 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
40153 (fmtresult::operator+=): Outlined.
40154 (struct fmtresult): Add ctors.
40155 (struct conversion_spec): Rename...
40156 (struct directive): ...to this. Add and remove data members.
40157 (directive::set_width, directive::set_precision): New functions.
40158 (format_percent): Use fmtresult ctor.
40159 (get_width_and_precision): Remove.
40160 (format_integer): Make naming changes. Avoid computing width and
40162 (format_floating): Same. Adjust indentation.
40163 (format_character, format_none): New functions.
40164 (format_string): Moved character handling to format_character.
40165 (format_directive): Remove arguments, change return type.
40166 (parse_directive): New function.
40167 (pass_sprintf_length::compute_format_length): Move directive
40168 parsing to parse_directive.
40170 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40172 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
40173 (assign_assembler_name_if_needed): ... this.
40174 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
40175 (assign_assembler_name_if_needed): ... this.
40176 (free_lang_data_in_cgraph): Adjust callers.
40177 * cgraphunit.c (cgraph_node::analyze): Likewise.
40178 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
40181 2017-01-23 Richard Biener <rguenther@suse.de>
40183 PR tree-optimization/79088
40184 PR tree-optimization/79188
40185 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
40186 resetting loop bounds after last path deletion. Reset loop
40187 bounds of the target loop, make code match the comments.
40188 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
40189 Make sure loops need no fixups.
40191 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
40193 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
40194 exponent support with double type for first argument.
40195 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
40196 type returned by __builtin_vec_extract_sig,
40197 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
40198 functions from "vector int" to "vector unsigned int" or from
40199 "vector long long int" to "vector unsigned long long int".
40200 Changed type returned by __builtin_vec_extract_exp,
40201 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
40202 functions from "vector int" to "vector unsigned int" or from
40203 "vector long long int" to "vector unsigned long long int".
40204 Changed return type of __builtin_vec_test_data_class,
40205 __builtin_vec_test_data_class_sp, and
40206 __builtin_vec_test_data_class_dp from "vector int" to
40207 "vector bool int" or from "vector long long int" to "vector bool
40208 long long int" and changed second argument type from "unsigned
40209 int" to "int". Added new overloaded function forms "vector float
40210 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
40211 "vector float __builtin_vec_insert_exp_sp (vector float, vector
40212 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
40213 double, vector unsigned long long int)" and "vector double
40214 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
40215 long int)". Changed return type of
40216 __builtin_scalar_test_data_class and
40217 __builtin_scalar_test_data_class_sp and
40218 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
40219 int" and changed second argument from "unsigned int" to "int".
40220 Changed type returned by __builtin_scalar_test_neg,
40221 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
40222 from "int" to "bool int". Added new overloaded function form
40223 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
40224 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
40225 exponent double-precision with floating point first argument.
40226 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
40227 documentation of scalar_test_data_class, scalar_test_neg,
40228 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
40229 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
40230 vec_test_data_class built-in functions to reflect refinements in
40231 their type signatures.
40233 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
40235 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
40237 (aarch64_elf_asm_destructor): Likewise.
40239 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
40241 PR rtl-optimization/78634
40242 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
40243 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
40244 * ifcvt.c (noce_try_cmove): Add missing cost check.
40246 PR rtl-optimization/71724
40247 * combine.c (if_then_else_cond): Look for situations where it is
40248 beneficial to undo the work of one of the recursive calls.
40250 2017-01-23 Bin Cheng <bin.cheng@arm.com>
40252 PR tree-optimization/70754
40253 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
40254 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
40255 combined stmt before it if not NULL.
40256 (combine_chains): Process refs reversely and compute dominance point
40259 2017-01-23 Martin Liska <mliska@suse.cz>
40261 PR tree-optimization/79196
40262 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
40263 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
40265 (strlen_optimize_stmt): Call the renamed function.
40267 2017-01-23 Michael Matz <matz@suse.de>
40269 PR tree-optimization/78384
40270 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
40272 2017-01-23 Richard Biener <rguenther@suse.de>
40274 PR tree-optimization/79186
40275 * tree-vrp.c (register_new_assert_for): Make sure we've seen
40276 both incoming edges before moving an assert.
40278 2017-01-23 Martin Jambor <mjambor@suse.cz>
40280 * ipa-prop.c (load_from_param_1): Removed.
40281 (load_from_unmodified_param): Bits from load_from_param_1 put back
40283 (load_from_param): Removed.
40284 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
40285 with stmt. Reverted back to use of load_from_unmodified_param.
40287 2017-01-23 Martin Jambor <mjambor@suse.cz>
40290 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
40291 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
40292 field a pointer to garbage collected vector, mark lattices and
40293 ipcp_orig_node with GTY((skip)).
40294 (ipa_get_param_count): Adjust to descriptors being a pointer.
40295 (ipa_get_param): Likewise.
40296 (ipa_get_type): Likewise.
40297 (ipa_get_param_move_cost): Likewise.
40298 (ipa_set_param_used): Likewise.
40299 (ipa_get_controlled_uses): Likewise.
40300 (ipa_set_controlled_uses): Likewise.
40301 (ipa_is_param_used): Likewise.
40302 (ipa_node_params_t): Move into garbage collector. New methods insert
40304 (ipa_node_params_sum): Annotate wth GTY(()).
40305 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
40307 (ipa_load_from_parm_agg): Adjust declaration.
40308 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
40309 * ipa-profile.c (ipa_profile): Likewise.
40310 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
40311 (ipa_populate_param_decls): Make descriptors parameter garbage
40313 (ipa_dump_param): Adjust to descriptors being a pointer.
40314 (ipa_alloc_node_params): Likewise.
40315 (ipa_initialize_node_params): Likewise.
40316 (load_from_param_1): Make descriptors parameter garbage collected.
40317 (load_from_unmodified_param): Likewise.
40318 (load_from_param): Likewise.
40319 (ipa_load_from_parm_agg): Likewise.
40320 (ipa_node_params::~ipa_node_params): Removed.
40321 (ipa_free_all_node_params): Remove call to delete operator.
40322 (ipa_node_params_t::insert): New.
40323 (ipa_node_params_t::remove): Likewise.
40324 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
40325 copy known_csts and known_contexts vectors.
40326 (ipa_read_node_info): Adjust to descriptors being a pointer.
40327 (ipcp_modif_dom_walker): Make m_descriptors field garbage
40329 (ipcp_transform_function): Make descriptors variable garbage
40332 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
40334 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
40335 * config/i386/avx512dqintrin.h: Ditto.
40336 * config/i386/avx512fintrin.h: Ditto.
40337 * config/i386/i386.c: Handle new builtins.
40338 * config/i386/i386-builtin.def: Add new builtins.
40339 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
40340 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
40342 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40343 Martin Liska <mliska@suse.cz>
40345 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
40346 * asan.c (asan_expand_poison_ifn): Support stores and use
40347 appropriate ASAN report function.
40348 * internal-fn.c (expand_ASAN_POISON_USE): New function.
40349 * internal-fn.def (ASAN_POISON_USE): Declare.
40350 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
40351 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
40352 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
40353 ASAN_POISON calls w/o LHS.
40354 * tree-ssa.c (execute_update_addresses_taken): Create clobber
40355 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
40356 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
40357 * gimplify.c (asan_poison_variables): Add attribute
40358 use_after_scope_memory to variables that really needs to live
40360 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
40361 having the attribute.
40363 2017-01-23 Martin Liska <mliska@suse.cz>
40365 * asan.c (create_asan_shadow_var): New function.
40366 (asan_expand_poison_ifn): Likewise.
40367 * asan.h (asan_expand_poison_ifn): New declaration.
40368 * internal-fn.c (expand_ASAN_POISON): Likewise.
40369 * internal-fn.def (ASAN_POISON): New builtin.
40370 * sanopt.c (pass_sanopt::execute): Expand
40371 asan_expand_poison_ifn.
40372 * tree-inline.c (copy_decl_for_dup_finish): Make function
40374 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
40375 * tree-ssa.c (is_asan_mark_p): New function.
40376 (execute_update_addresses_taken): Rewrite local variables
40377 (identified just by use-after-scope as addressable) into SSA.
40379 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40381 * doc/install.texi (Specific): opensource.apple.com uses https
40382 now. Remove trailing slash.
40384 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40386 * README.Portability: Remove note on an Irix compatibility issue.
40388 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
40390 * gcov.c (INCLUDE_ALGORITHM): Define.
40391 (INCLUDE_VECTOR): Define.
40392 No longer include <vector> and <algorithm> directly.
40394 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
40396 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
40398 * doc/invoke.texi (Code Gen Options): Ditto.
40400 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
40403 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
40405 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
40407 rtl-optimization/79125
40408 * cprop.c (local_cprop_pass): Handle cases where we make an
40409 unconditional trap.
40411 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
40415 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
40416 read from, for big endian.
40418 2017-01-20 Jiong Wang <jiong.wang@arm.com>
40420 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
40421 register pauth builtins for LP64 only.
40423 2017-01-20 Marek Polacek <polacek@redhat.com>
40426 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
40429 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40431 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
40433 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
40434 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
40435 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
40437 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40440 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
40441 in big-endian mode when they are not a single duplicated value.
40443 2017-01-20 Richard Biener <rguenther@suse.de>
40445 * BASE-VER: Bump to 7.0.1.
40447 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40449 * omp-low.c (omplow_simd_context): New struct. Use it...
40450 (lower_rec_simd_input_clauses): ...here and...
40451 (lower_rec_input_clauses): ...here to hold common data. Adjust all
40452 references to idx, lane, max_vf, is_simt.
40454 2017-01-20 Graham Markall <graham.markall@embecosm.com>
40456 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
40459 2017-01-20 Martin Jambor <mjambor@suse.cz>
40461 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
40462 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
40464 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
40465 (GTFILES): Rename hsa.c to hsa-common.c.
40466 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
40467 * hsa-dump.c: Likewise.
40468 * hsa-gen.c: Likewise.
40469 * hsa-regalloc.c: Likewise.
40470 * ipa-hsa.c: Likewise.
40471 * omp-expand.c: Likewise.
40472 * omp-low.c: Likewise.
40473 * toplev.c: Likewise.
40475 2017-01-20 Marek Polacek <polacek@redhat.com>
40478 * doc/invoke.texi: Document -Wduplicated-branches.
40479 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
40480 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
40481 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
40482 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
40483 return 0 only when not OEP_LEXICOGRAPHIC.
40484 (fold_build_cleanup_point_expr): Use the expression
40485 location when building CLEANUP_POINT_EXPR.
40486 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
40487 * tree.c (add_expr): Handle error_mark_node.
40489 2017-01-20 Martin Liska <mliska@suse.cz>
40492 * tree-profile.c (init_ic_make_global_vars): Do not call
40494 (gimple_init_gcov_profiler): Likewise.
40496 2017-01-20 Martin Liska <mliska@suse.cz>
40499 * cgraph.h (maybe_create_reference): Remove argument and
40501 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
40503 * ipa-cp.c (create_specialized_node): Likewise.
40504 * symtab.c (symtab_node::maybe_create_reference): Handle
40505 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
40507 2017-01-20 Martin Liska <mliska@suse.cz>
40509 * read-rtl-function.c (function_reader::create_function): Use
40510 build_decl instread of build_decl_stat.
40512 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
40514 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
40515 * config/i386/avx512dqintrin.h: Ditto.
40516 * config/i386/avx512fintrin.h: Ditto.
40517 * config/i386/i386-builtin-types.def: Add new types.
40518 * config/i386/i386.c: Handle new types.
40519 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
40520 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
40521 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
40522 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
40523 (__builtin_ia32_kshiftridi): New.
40524 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
40526 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
40530 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
40531 define to rs6000_init_stack_protect_guard.
40532 (rs6000_init_stack_protect_guard): New function.
40534 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40535 Yunqiang Su <yunqiang.su@imgtec.com>
40537 * config.gcc (supported_defaults): Add madd4.
40538 (with_madd4): Add validation.
40539 (all_defaults): Add madd4.
40540 * config/mips/mips.opt (mmadd4): New option.
40541 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40543 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40545 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
40546 (ISA_HAS_FUSED_MADD4): Likewise.
40547 * doc/invoke.texi (-mmadd4): Document the new option.
40548 * doc/install.texi (--with-madd4): Document the new option.
40550 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40552 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
40553 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
40554 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
40555 (aarch64_init_pauth_hint_builtins): New.
40556 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
40557 (aarch64_expand_builtin): Expand new builtins.
40559 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40561 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
40562 * combine-stack-adj.c (no_unhandled_cfa): Handle
40563 REG_CFA_TOGGLE_RA_MANGLE.
40564 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
40565 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
40566 info for return address signing.
40567 (aarch64_expand_epilogue): Likewise.
40569 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40571 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
40572 * config/aarch64/aarch64-protos.h
40573 (aarch64_return_address_signing_enabled): New declaration.
40574 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
40576 (aarch64_expand_prologue): Sign return address before it's pushed onto
40578 (aarch64_expand_epilogue): Authenticate return address fetched from
40580 (aarch64_override_options): Sanity check for ILP32 and ISA level.
40581 (aarch64_attributes): New function attributes for "sign-return-address".
40582 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
40583 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
40584 ("*do_return"): Generate combined instructions according to key index.
40585 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
40586 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
40588 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
40589 * config/aarch64/aarch64.opt (msign-return-address=): New.
40590 * doc/extend.texi (AArch64 Function Attributes): Documents
40591 "sign-return-address=".
40592 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
40594 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40596 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
40597 overall option summary.
40599 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40601 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
40602 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
40603 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
40604 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
40606 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
40608 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
40609 -mpower9-minmax by default for -mcpu=power9.
40610 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
40611 128-bit floating point.
40613 2017-01-20 Alan Modra <amodra@gmail.com>
40615 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
40616 optimizing for size.
40618 2017-01-20 Alan Modra <amodra@gmail.com>
40621 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
40622 for strcmp and strncmp from corresponding builtin decl.
40624 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40626 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
40627 instead of i386/rtems-64.h.
40628 * config/i386/rtems-64.h: Remove.
40630 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40634 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
40636 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
40638 2017-01-19 Tamar Christina <tamar.christina@arm.com>
40640 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
40641 Change int to HOST_WIDE_INT.
40642 * config/aarch64/aarch64-protos.h
40643 (aarch64_simd_gen_const_vector_dup): Likewise.
40644 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
40646 2017-01-19 David Malcolm <dmalcolm@redhat.com>
40648 * langhooks-def.h (lhd_type_for_size): New decl.
40649 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
40650 * langhooks.c (lhd_type_for_size): New function, taken from
40653 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
40655 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
40656 define_bypass for CR latency.
40657 (power9-cracked-alu): Update bypass latency and remove power9-branch.
40658 (power9-alu2): Add define_bypass for CR latency.
40660 (power9-mul): Update insn latency.
40661 (power9-mul-compare): Update insn latency, bypass latency and remove
40664 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40666 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
40668 * config/aarch64/aarch64.md
40669 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
40670 aarch64_nopcrelative_literal_loads.
40671 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
40673 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
40675 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
40676 TARGET_LOONGSON_3A.
40677 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
40679 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
40682 * config.gcc (supported_defaults): Add lxc1-sxc1.
40683 (with_lxc1_sxc1): Add validation.
40684 (all_defaults): Add lxc1-sxc1.
40685 * config/mips/mips.opt (mlxc1-sxc1): New option.
40686 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40688 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40689 __mips_no_lxc1_sxc1.
40690 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
40691 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
40692 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
40694 2017-01-19 Richard Biener <rguenther@suse.de>
40696 PR tree-optimization/72488
40697 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
40698 sure to restore SSA info.
40699 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
40701 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
40703 PR rtl-optimization/79121
40704 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
40705 of the inner type when shifting an extended value.
40707 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40710 * symtab.c (symtab_node::equal_address_to): Fix comparing of
40711 interposable aliases.
40713 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
40716 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
40717 Use the evmergelohi instruction.
40718 (mov_si<mode>_e500_subreg4_2_le): Likewise.
40719 (mov_sitf_e500_subreg8_2_be): Likewise.
40720 (mov_sitf_e500_subreg12_2_le): Likewise.
40721 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
40722 (mov_si<mode>_e500_subreg4_2_be): Likewise.
40723 (mov_sitf_e500_subreg8_2_le): Likewise.
40724 (mov_sitf_e500_subreg12_2_be): Likewise.
40726 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40728 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
40729 attribute from vecsimple to vecperm.
40730 (altivec_vbpermq2): Likewise.
40732 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40735 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
40737 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
40738 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
40739 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
40740 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
40741 case where N arg is SIZE_MAX.
40742 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
40743 (cmpstrsi): Add pattern.
40745 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
40747 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
40748 __builtin_vec_revb builtins.
40749 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
40750 built-in functions to support generation of the ISA 3.0 XXBR<x>
40751 vector byte reverse instructions.
40752 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
40753 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
40754 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
40755 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
40756 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
40757 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
40758 (P9V_BUILTIN_VEC_REVB): Likewise.
40759 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
40760 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
40761 (p9_xxbrq_v16qi): Likewise.
40762 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
40763 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
40764 (p9_xxbrh_v8hi): Likewise.
40765 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
40766 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
40767 vec_revb built-in functions.
40769 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
40771 PR rtl-optimization/78952
40772 * config/i386/i386.md (any_extract): New code iterator.
40773 (*insvqi_2): Use any_extract for source operand.
40774 (*insvqi_3): Use any_shiftrt for source operand.
40776 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
40778 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
40780 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
40782 2017-01-18 Matthias Klose <doko@ubuntu.com>
40784 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
40786 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40788 * config/rs6000/altivec.h (vec_bperm): Change #define.
40789 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
40790 (altivec_vbpermq2): New define_insn.
40791 (altivec_vbpermd): Likewise.
40792 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
40793 function interface.
40794 (VBPERMD): Likewise.
40795 (VBPERM): New polymorphic function interface.
40796 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
40797 Add entries for P9V_BUILTIN_VEC_VBPERM.
40798 * doc/extend.texi: Add interfaces for vec_bperm.
40800 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40802 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
40803 first letter of error messages.
40804 (s390_resolve_overloaded_builtin): Likewise.
40805 * config/s390/s390.c (s390_expand_builtin): Likewise.
40806 (s390_invalid_arg_for_unprototyped_fn): Likewise.
40807 (s390_valid_target_attribute_inner_p): Likewise.
40808 * config/s390/s390.md ("tabort"): Likewise.
40810 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
40812 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
40813 (ISA_AVOID_DIV_HILO): New macro.
40814 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
40815 (ISA_HAS_DDIV): Likewise.
40817 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40819 * doc/invoke.texi (fabi-version): Correct number of occurrences.
40821 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40823 * doc/invoke.texi (fabi-version): Spelling fix.
40825 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40828 * doc/invoke.texi (fabi-version): Mention mangling fix for
40831 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40834 * doc/invoke.texi (fabi-version): Document discriminator mangling.
40836 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
40839 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
40840 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
40842 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
40844 (stack_protect_test): Ditto.
40845 * config/rs6000/rs6000.opt (mstack-protector-guard=,
40846 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
40848 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40849 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40850 -mstack-protector-guard-offset=.
40851 (RS/6000 and PowerPC Options): Ditto.
40853 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40855 * config/i386/i386.h (MASK_CLASS_P): New define.
40856 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40857 there are no registers from different register sets also when
40858 mask registers are used. Update function comment.
40859 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40860 to (*k/*r) and (*k/*km) alternatives.
40862 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40864 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40865 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40866 (EH_RETURN_HANDLER_RTX): New define.
40867 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40868 Force frame pointer in EH return functions.
40869 (aarch64_expand_epilogue): Add barrier for eh_return.
40870 (aarch64_final_eh_return_addr): Remove.
40871 (aarch64_eh_return_handler_rtx): New function.
40872 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40874 (aarch64_eh_return_handler_rtx): New prototype.
40876 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40878 * config/rs6000/altivec.h (vec_rlmi): New #define.
40879 (vec_vrlnm): Likewise.
40880 (vec_rlnm): Likewise.
40881 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40882 (UNSPEC_VRLNM): Likewise.
40883 (VIlong): New mode iterator.
40884 (altivec_vrl<VI_char>mi): New define_insn.
40885 (altivec_vrl<VI_char>nm): Likewise.
40886 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40888 (VRLDNM): Likewise.
40889 (RLNM): New polymorphic function entry.
40890 (VRLWMI): New monomorphic function entry.
40891 (VRLDMI): Likewise.
40892 (RLMI): New polymorphic function entry.
40893 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40894 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40895 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40898 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40901 * dwarf2out.c (field_byte_offset): Restore the
40902 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40903 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40904 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40907 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40910 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40912 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40915 * dwarf2out.c (add_data_member_location_attribute): For constant
40916 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40917 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40918 DW_AT_byte_size attributes.
40920 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40922 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40923 after forcing to constant memory when the code model is medium.
40925 2017-01-17 Julia Koval <julia.koval@intel.com>
40928 * config/i386/avx512fintrin.h
40929 (_mm512_i32gather_ps): Change __addr type to void const*.
40930 (_mm512_mask_i32gather_ps): Ditto.
40931 (_mm512_i32gather_pd): Ditto.
40932 (_mm512_mask_i32gather_pd): Ditto.
40933 (_mm512_i64gather_ps): Ditto.
40934 (_mm512_mask_i64gather_ps): Ditto.
40935 (_mm512_i64gather_pd): Ditto.
40936 (_mm512_mask_i64gather_pd): Ditto.
40937 (_mm512_i32gather_epi32): Ditto.
40938 (_mm512_mask_i32gather_epi32): Ditto.
40939 (_mm512_i32gather_epi64): Ditto.
40940 (_mm512_mask_i32gather_epi64): Ditto.
40941 (_mm512_i64gather_epi32): Ditto.
40942 (_mm512_mask_i64gather_epi32): Ditto.
40943 (_mm512_i64gather_epi64): Ditto.
40944 (_mm512_mask_i64gather_epi64): Ditto.
40945 (_mm512_i32scatter_ps): Change __addr type to void*.
40946 (_mm512_mask_i32scatter_ps): Ditto.
40947 (_mm512_i32scatter_pd): Ditto.
40948 (_mm512_mask_i32scatter_pd): Ditto.
40949 (_mm512_i64scatter_ps): Ditto.
40950 (_mm512_mask_i64scatter_ps): Ditto.
40951 (_mm512_i64scatter_pd): Ditto.
40952 (_mm512_mask_i64scatter_pd): Ditto.
40953 (_mm512_i32scatter_epi32): Ditto.
40954 (_mm512_mask_i32scatter_epi32): Ditto.
40955 (_mm512_i32scatter_epi64): Ditto.
40956 (_mm512_mask_i32scatter_epi64): Ditto.
40957 (_mm512_i64scatter_epi32): Ditto.
40958 (_mm512_mask_i64scatter_epi32): Ditto.
40959 (_mm512_i64scatter_epi64): Ditto.
40960 (_mm512_mask_i64scatter_epi64): Ditto.
40961 * config/i386/avx512pfintrin.h
40962 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40963 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40964 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40965 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40966 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40967 (_mm512_prefetch_i32scatter_ps): Ditto.
40968 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40969 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40970 (_mm512_prefetch_i64scatter_pd): Ditto.
40971 (_mm512_prefetch_i64scatter_ps): Ditto.
40972 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40973 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40974 * config/i386/avx512vlintrin.h
40975 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40976 (_mm_mmask_i32gather_ps): Ditto.
40977 (_mm256_mmask_i32gather_pd): Ditto.
40978 (_mm_mmask_i32gather_pd): Ditto.
40979 (_mm256_mmask_i64gather_ps): Ditto.
40980 (_mm_mmask_i64gather_ps): Ditto.
40981 (_mm256_mmask_i64gather_pd): Ditto.
40982 (_mm_mmask_i64gather_pd): Ditto.
40983 (_mm256_mmask_i32gather_epi32): Ditto.
40984 (_mm_mmask_i32gather_epi32): Ditto.
40985 (_mm256_mmask_i32gather_epi64): Ditto.
40986 (_mm_mmask_i32gather_epi64): Ditto.
40987 (_mm256_mmask_i64gather_epi32): Ditto.
40988 (_mm_mmask_i64gather_epi32): Ditto.
40989 (_mm256_mmask_i64gather_epi64): Ditto.
40990 (_mm_mmask_i64gather_epi64): Ditto.
40991 (_mm256_i32scatter_ps): Change __addr type to void*.
40992 (_mm256_mask_i32scatter_ps): Ditto.
40993 (_mm_i32scatter_ps): Ditto.
40994 (_mm_mask_i32scatter_ps): Ditto.
40995 (_mm256_i32scatter_pd): Ditto.
40996 (_mm256_mask_i32scatter_pd): Ditto.
40997 (_mm_i32scatter_pd): Ditto.
40998 (_mm_mask_i32scatter_pd): Ditto.
40999 (_mm256_i64scatter_ps): Ditto.
41000 (_mm256_mask_i64scatter_ps): Ditto.
41001 (_mm_i64scatter_ps): Ditto.
41002 (_mm_mask_i64scatter_ps): Ditto.
41003 (_mm256_i64scatter_pd): Ditto.
41004 (_mm256_mask_i64scatter_pd): Ditto.
41005 (_mm_i64scatter_pd): Ditto.
41006 (_mm_mask_i64scatter_pd): Ditto.
41007 (_mm256_i32scatter_epi32): Ditto.
41008 (_mm256_mask_i32scatter_epi32): Ditto.
41009 (_mm_i32scatter_epi32): Ditto.
41010 (_mm_mask_i32scatter_epi32): Ditto.
41011 (_mm256_i32scatter_epi64): Ditto.
41012 (_mm256_mask_i32scatter_epi64): Ditto.
41013 (_mm_i32scatter_epi64): Ditto.
41014 (_mm_mask_i32scatter_epi64): Ditto.
41015 (_mm256_i64scatter_epi32): Ditto.
41016 (_mm256_mask_i64scatter_epi32): Ditto.
41017 (_mm_i64scatter_epi32): Ditto.
41018 (_mm_mask_i64scatter_epi32): Ditto.
41019 (_mm256_i64scatter_epi64): Ditto.
41020 (_mm256_mask_i64scatter_epi64): Ditto.
41021 (_mm_i64scatter_epi64): Ditto.
41022 (_mm_mask_i64scatter_epi64): Ditto.
41023 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
41024 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
41025 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
41026 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
41027 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
41028 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
41029 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
41030 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
41031 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
41032 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
41033 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
41034 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
41035 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
41036 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
41037 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
41038 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
41039 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
41040 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
41041 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
41042 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
41043 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
41044 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
41045 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
41046 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
41047 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
41048 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
41049 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
41050 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
41051 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
41052 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
41053 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
41054 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
41055 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
41056 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
41057 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
41058 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
41059 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
41060 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
41061 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
41062 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
41063 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
41064 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
41065 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
41066 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
41067 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
41068 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
41069 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
41070 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
41071 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
41072 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
41073 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
41074 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
41075 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
41076 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
41077 definitions accordingly.
41079 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
41080 Kuan-Lin Chen <kuanlinchentw@gmail.com>
41083 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
41086 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
41089 * ira-conflicts.c (ira_build_conflicts): Update total conflict
41090 hard regs for inner regno.
41092 2017-01-17 Martin Liska <mliska@suse.cz>
41095 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
41096 assumption and add comment.
41098 2017-01-17 Nathan Sidwell <nathan@acm.org>
41100 * ipa-visibility.c (localize_node): New function, broken out of ...
41101 (function_and_variable_visibility): ... here. Call it.
41103 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
41105 PR middle-end/77445
41106 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
41107 correctly set frequency of oudgoing edge.
41108 (duplicate_thread_path): Fix profile updating.
41110 2017-01-17 Jakub Jelinek <jakub@redhat.com>
41113 * configure.ac: Add GCC_BASE_VER.
41114 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
41115 version from BASE-VER file.
41116 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
41117 (gcc.o): Depend on $(BASEVER).
41118 * common.opt (dumpfullversion): New option.
41119 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
41120 * doc/invoke.texi: Document -dumpfullversion.
41121 * doc/install.texi: Document --with-gcc-major-version-only.
41122 * configure: Regenerated.
41124 2017-01-17 Richard Biener <rguenther@suse.de>
41126 PR tree-optimization/71433
41127 * tree-vrp.c (register_new_assert_for): Merge same asserts
41128 on all incoming edges.
41129 (process_assert_insertions_for): Handle insertions at the
41132 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
41134 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
41135 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
41137 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
41140 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
41143 2017-01-17 Alan Modra <amodra@gmail.com>
41146 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
41147 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
41148 symbolic stack limit when pic.
41150 2017-01-16 Martin Sebor <msebor@redhat.com>
41152 PR tree-optimization/78608
41153 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
41155 2017-01-16 Jeff Law <law@redhat.com>
41158 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
41159 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
41160 for several include directories that may be relative to sysroot.
41161 * config/i386/x-mingw32 (gplus_includedir): Define.
41162 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
41163 (native_system_includedir): Likewise.
41164 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
41165 override if TARGET_SYSTEM_ROOT is defined.
41166 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
41168 PR tree-optimization/79090
41169 PR tree-optimization/33562
41170 PR tree-optimization/61912
41171 PR tree-optimization/77485
41172 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
41173 and computed trims into the dump file.
41175 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
41177 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
41179 2017-01-16 Jakub Jelinek <jakub@redhat.com>
41182 * gimplify.c (gimplify_init_constructor): If want_value and
41183 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
41187 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
41188 sequence. Formatting fixes.
41189 (doloop_optimize): Formatting fixes.
41192 * gcc.c (debug_level_greater_than_spec_func): New function.
41193 (static_spec_functions): Add debug-level-gt spec function.
41194 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
41196 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
41197 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
41198 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
41199 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
41200 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
41201 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
41203 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
41205 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
41206 QImode fixups to general and mask registers only.
41208 2017-01-16 Carl Love <cel@us.ibm.com>
41210 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
41211 for built-in functions
41212 vector signed char vec_nabs (vector signed char)
41213 vector signed short vec_nabs (vector signed short)
41214 vector signed int vec_nabs (vector signed int)
41215 vector signed long long vec_nabs (vector signed long long)
41216 vector float vec_nabs (vector float)
41217 vector double vec_nabs (vector double)
41218 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
41220 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
41221 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
41222 * doc/extend.texi: Update the documentation file for the new built-in
41225 2017-01-16 Martin Sebor <msebor@redhat.com>
41227 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
41230 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41232 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
41233 UNSPEC_VSX__XXSPLTD to require special splat handling.
41235 2017-01-16 David Malcolm <dmalcolm@redhat.com>
41238 * system.h: Poison strndup.
41240 2017-01-16 Alan Modra <amodra@gmail.com>
41243 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
41246 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
41248 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
41250 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
41252 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
41253 call recog here. Assert that INSN_CODE (insn) is non-negative.
41255 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41258 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
41260 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
41261 in the currently scheduled RTL fragment.
41263 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41265 PR rtl-optimization/78751
41266 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
41269 2017-01-14 Jeff Law <law@redhat.com>
41271 PR tree-optimization/79090
41272 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
41273 variable length stores.
41274 (compute_trims): Delete dead assignment to *trim_tail.
41275 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
41278 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
41280 PR rtl-optimization/78626
41281 PR rtl-optimization/78727
41282 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
41283 of a block, and split such blocks after everything else is finished.
41285 2017-01-14 Alan Modra <amodra@gmail.com>
41288 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
41289 target legitimate_combined_insn.
41290 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
41291 (rs6000_legitimate_combined_insn): New function.
41292 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
41294 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
41295 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
41296 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
41298 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
41300 * doc/frontends.texi (G++ and GCC): Remove references to Java.
41302 2017-01-13 Jeff Law <law@redhat.com>
41304 PR tree-optimization/33562
41305 PR tree-optimization/61912
41306 PR tree-optimization/77485
41307 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
41309 (delete_dead_assignment): Likewise.
41310 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
41311 statement to delete_dead_call and delete_dead_assignment.
41313 2017-01-13 David Malcolm <dmalcolm@redhat.com>
41316 * substring-locations.c (format_warning_va): Strengthen case 1 so
41317 that both endpoints of the substring must be within the format
41318 range for just the substring to be printed.
41320 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
41322 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
41323 * config/i386/i386.c (ix86_target_string): Add missing options
41324 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
41325 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
41326 flags_other and ix86_target_other to flags2_other. Display unknown
41328 (ix86_valid_target_attribute_inner_p): Add missing options and
41329 reorder options by implied ISAs, as in ix86_target_string.
41331 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41333 * hash-table.h (hash_table::too_empty_p): New function.
41334 (hash_table::expand): Use it.
41335 (hash_table::traverse): Likewise.
41336 (hash_table::empty_slot): Use sizeof (value_type) instead of
41337 sizeof (PTR) to convert bytes to elements. Shrink the table
41338 if the current size is excessive for the current number of
41341 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41343 * ira-costs.c (record_reg_classes): Break from the inner loop
41344 early once alt_fail is known to be true. Update outer loop
41345 handling accordingly.
41347 2017-01-13 Jeff Law <law@redhat.com>
41349 * tree-ssa-dse.c (decrement_count): New function.
41350 (increment_start_addr, maybe_trim_memstar_call): Likewise.
41351 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
41352 when we know the partially dead statement is a mem* function.
41354 PR tree-optimization/61912
41355 PR tree-optimization/77485
41356 * tree-ssa-dse.c: Include expr.h.
41357 (maybe_trim_constructor_store): New function.
41358 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
41360 PR tree-optimization/33562
41361 PR tree-optimization/61912
41362 PR tree-optimization/77485
41363 * doc/invoke.texi: Document new dse-max-object-size param.
41364 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
41365 * tree-ssa-dse.c: Include params.h.
41366 (dse_store_status): New enum.
41367 (initialize_ao_ref_for_dse): New, partially extracted from
41369 (valid_ao_ref_for_dse, normalize_ref): New.
41370 (setup_live_bytes_from_ref, compute_trims): Likewise.
41371 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
41372 (maybe_trim_partially_dead_store): Likewise.
41373 (maybe_trim_complex_store): Likewise.
41374 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
41375 Track what bytes live from the original store. Return tri-state
41376 for dead, partially dead or live.
41377 (dse_dom_walker): Add constructor, destructor and new private members.
41378 (delete_dead_call, delete_dead_assignment): New extracted from
41380 (dse_optimize_stmt): Make a member of dse_dom_walker.
41381 Use initialize_ao_ref_for_dse.
41383 PR tree-optimization/33562
41384 PR tree-optimization/61912
41385 PR tree-optimization/77485
41386 * sbitmap.h (bitmap_count_bits): Prototype.
41387 (bitmap_clear_range, bitmap_set_range): Likewise.
41388 * sbitmap.c (bitmap_clear_range): New function.
41389 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
41391 2017-01-13 Martin Liska <mliska@suse.cz>
41394 * function.c (set_cfun): Add new argument force.
41395 * function.h (set_cfun): Likewise.
41396 * ipa-inline-transform.c (inline_call): Use the function when
41397 strict alising from is dropped for function we inline to.
41399 2017-01-13 Richard Biener <rguenther@suse.de>
41401 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
41402 for dumping GIMPLE INTEGER_CSTs.
41404 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
41406 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
41407 to 201112L since C++17.
41409 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
41412 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
41413 if -fsanitize=kernel-address is present.
41415 2017-01-13 Richard Biener <rguenther@suse.de>
41417 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
41418 as _Literal ( type ) number in case usual suffixes do not
41419 preserve all information.
41421 2017-01-13 Richard Biener <rguenther@suse.de>
41423 PR tree-optimization/77283
41424 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
41425 and ssa-iterators.h.
41426 (is_feasible_trace): Implement a cost model based on joiner
41429 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
41432 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
41433 char or short to __float128/_Float128 directly.
41435 2017-01-12 Martin Sebor <msebor@redhat.com>
41437 to -Wformat-overflow.
41438 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
41439 (min_bytes_remaining): Same.
41440 (get_string_length): Same.
41441 (format_string): Same.
41442 (format_directive): Same.
41444 (pass_sprintf_length::handle_gimple_call): Same.
41446 2017-01-12 Jakub Jelinek <jakub@redhat.com>
41448 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
41449 info.nowrite calls with no lhs that can't throw. Return bool
41450 whether gsi_remove has been called or not.
41451 (pass_sprintf_length::handle_gimple_call): Return bool whether
41452 try_substitute_return_value called gsi_remove. Formatting fix.
41453 (pass_sprintf_length::execute): Don't use gsi_remove if
41454 handle_gimple_call returned true.
41457 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
41458 be removed due to side-effects, don't remove following barrier nor
41459 turn the successor edge into fallthru edge.
41461 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41464 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
41465 element-reversing loads and stores as not swappable.
41467 2017-01-12 Nathan Sidwell <nathan@acm.org>
41468 Nicolai Stange <nicstange@gmail.com>
41470 * combine.c (try_combine): Don't ignore result of overlap checking
41471 loop. Combine overlap & asm check into single loop.
41473 2017-01-12 Richard Biener <rguenther@suse.de>
41475 * tree-pretty-print.c (dump_generic_node): Provide -gimple
41476 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
41478 2017-01-12 Richard Biener <rguenther@suse.de>
41480 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
41481 and TS_TARGET_OPTION directly derive from TS_BASE.
41482 * tree-core.h (tree_optimization_option): Derive from tree_base.
41483 (tree_target_option): Likewise.
41485 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
41487 * config/i386/i386.c (memory_address_length): Increase len
41488 only when rip_relative_addr_p returns false.
41490 2017-01-11 Julia Koval <julia.koval@intel.com>
41492 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
41493 (OPTION_MASK_ISA_SGX_SET): New.
41494 (ix86_handle_option): Handle OPT_msgx.
41495 * config.gcc: Added sgxintrin.h.
41496 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
41497 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
41498 * config/i386/i386.c (ix86_target_string): Add -msgx.
41500 (ix86_option_override_internal): Handle new options.
41501 (ix86_valid_target_attribute_inner_p): Add sgx.
41502 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
41503 * config/i386/i386.opt: Add msgx.
41504 * config/i386/sgxintrin.h: New file.
41505 * config/i386/x86intrin.h: Add sgxintrin.h.
41507 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41510 * fold-const.c (maybe_nonzero_address): Return 1 for function
41512 (tree_single_nonzero_warnv_p): Don't handle function local objects
41516 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
41519 2017-01-11 David Malcolm <dmalcolm@redhat.com>
41522 * opts.c: Include "spellcheck.h"
41523 (struct string_fragment): New struct.
41524 (struct edit_distance_traits<const string_fragment &>): New
41526 (get_closest_sanitizer_option): New function.
41527 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
41529 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41531 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
41533 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
41534 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
41535 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
41536 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
41537 for initial die_offset if dwarf_split_debug_info.
41538 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
41539 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
41541 (output_skeleton_debug_sections): Formatting fix. Use
41542 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
41543 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
41545 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
41547 * config/arm/cortex-a53.md: Add bypasses for
41548 cortex_a53_r2f_cvt.
41549 (cortex_a53_r2f): Only use for transfers.
41550 (cortex_a53_f2r): Likewise.
41551 (cortex_a53_r2f_cvt): Add reservation for conversions.
41552 (cortex_a53_f2r_cvt): Likewise.
41554 2017-01-11 Tamar Christina <tamar.christina@arm.com>
41556 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
41557 to all inlined functions, change static to extern.
41559 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
41562 * config/arm/arm.c (legitimize_pic_address): Handle reference to
41564 (arm_assemble_integer): Likewise.
41566 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
41568 * config.gcc: Use new awk script to check CPU, FPU and architecture
41569 parameters for --with-... options.
41570 * config/arm/parsecpu.awk: New file
41571 * config/arm/arm-cpus.in: New file.
41572 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
41574 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
41576 * config/arm/t-arm: Update dependency rules.
41577 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
41578 of processing .def files.
41579 * config/arm/genopt.sh: Deleted.
41580 * config/arm/gentune.sh: Deleted.
41581 * config/arm/arm-cores.def: Deleted.
41582 * config/arm/arm-arches.def: Deleted.
41583 * config/arm/arm-fpus.def: Deleted.
41584 * config/arm/arm-tune.md: Regenerated.
41585 * config/arm/arm-tables.opt: Regenerated.
41586 * config/arm/arm-cpu.h: New generated file.
41587 * config/arm/arm-cpu-data.h: New generated file.
41588 * config/arm/arm-cpu-cdata.h: New generated file.
41590 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
41593 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
41595 (input_varpool_node): Unpack dynamically_initialized bit.
41597 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
41599 PR rtl-optimization/79032
41600 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
41601 the alignment of the adjusted memory reference against that of MODE,
41602 instead of the alignment of the original memory reference.
41604 2017-01-11 Martin Jambor <mjambor@suse.cz>
41606 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
41608 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
41609 decorated functions.
41611 2017-01-11 Richard Biener <rguenther@suse.de>
41613 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
41614 set range/nonnull info for PHI results. Do not set it on
41615 stmts marked for removal.
41617 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
41619 * expr.c (store_field): In the bitfield case, fetch the return value
41620 from the registers before applying a single big-endian adjustment.
41621 Always do a final load for a BLKmode value not larger than a word.
41623 2017-01-10 David Malcolm <dmalcolm@redhat.com>
41626 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41627 that we correctly handle column numbers greater than
41628 LINE_MAP_MAX_COLUMN_NUMBER.
41630 2017-01-10 Martin Sebor <msebor@redhat.com>
41632 PR middle-end/78245
41633 * gimple-ssa-sprintf.c (get_destination_size): Call
41634 {init,fini}object_sizes.
41635 * tree-object-size.c (addr_object_size): Adjust.
41636 (pass_through_call): Adjust.
41637 (pass_object_sizes::execute): Adjust.
41638 * tree-object-size.h (fini_object_sizes): Declare.
41640 2017-01-10 Martin Sebor <msebor@redhat.com>
41642 PR tree-optimization/78775
41643 * builtins.c (get_size_range): Move...
41644 * calls.c: ...to here.
41645 (alloc_max_size): Accept zero argument.
41646 (operand_signed_p): Remove.
41647 (maybe_warn_alloc_args_overflow): Call get_size_range.
41648 * calls.h (get_size_range): Declare.
41650 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
41652 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
41653 from TI's devices.csv file as of September 2016.
41654 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
41656 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
41658 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
41659 * doc/invoke.texi: Likewise.
41660 * doc/md.texi: Likewise.
41661 * doc/objc.texi: Likewise.
41663 2017-01-10 Joshua Conner <joshconner@google.com>
41665 * config/arm/fuchsia-elf.h: New file.
41666 * config/fuchsia.h: New file.
41667 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
41668 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
41670 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
41672 2016-01-10 Richard Biener <rguenther@suse.de>
41674 PR tree-optimization/79034
41675 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
41676 Propagate out degenerate PHIs in the joiner.
41678 2017-01-10 Martin Liska <mliska@suse.cz>
41680 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
41681 (sort_congruence_classes_by_decl_uid): Likewise.
41682 (sort_congruence_class_groups_by_decl_uid): Likewise.
41683 (sem_item_optimizer::merge_classes): Sort class, groups in these
41684 classes and members in the groups by DECL_UID of declarations.
41685 This would make merge operations stable.
41687 2017-01-10 Martin Liska <mliska@suse.cz>
41689 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
41690 usage of m_classes_vec.
41691 (sem_item_optimizer::~sem_item_optimizer): Likewise.
41692 (sem_item_optimizer::get_group_by_hash): Likewise.
41693 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
41694 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
41695 (sem_item_optimizer::verify_classes): Likewise.
41696 (sem_item_optimizer::process_cong_reduction): Likewise.
41697 (sem_item_optimizer::dump_cong_classes): Likewise.
41698 (sem_item_optimizer::merge_classes): Likewise.
41699 * ipa-icf.h (congruence_class_hash): Rename from
41700 congruence_class_group_hash. Remove declaration of m_classes_vec.
41702 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
41704 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
41705 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
41706 * config.gcc: Add avx512vpopcntdqintrin.h.
41707 * config/i386/avx512vpopcntdqintrin.h: New.
41708 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
41709 * config/i386/i386-builtin-types.def: Add new types.
41710 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
41711 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
41712 __builtin_ia32_vpopcountq_v8di_mask): New.
41713 * config/i386/i386-c.c (ix86_target_macros_internal): Define
41714 __AVX512VPOPCNTDQ__.
41715 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
41716 (PTA_AVX512VPOPCNTDQ): Define.
41717 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
41718 TARGET_AVX512VPOPCNTDQ_P): Define.
41719 * config/i386/i386.opt: Add mavx512vpopcntdq.
41720 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
41721 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
41723 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41725 PR middle-end/77484
41726 * predict.def (PRED_CALL): Set to 67.
41728 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
41730 * expr.c (store_field): In the bitfield case, if the value comes from
41731 a function call and is of an aggregate type returned in registers, do
41732 not modify the field mode; extract the value in all cases if the mode
41733 is BLKmode and the size is not larger than a word.
41735 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
41738 * config/i386/cpuid.h: Fix undefined behavior.
41740 2017-01-04 Jeff Law <law@redhat.com>
41742 PR tree-optimization/79007
41743 PR tree-optimization/67955
41744 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
41745 conservative for pt.null when flag_non_call_exceptions is on.
41747 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41749 PR translation/79019
41750 PR translation/79020
41751 * params.def (PARAM_INLINE_MIN_SPEEDUP,
41752 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
41753 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
41755 * config/avr/avr.opt (maccumulate-args): Likewise.
41756 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
41757 * common.opt (freport-bug): Likewise.
41758 * cif-code.def (CIF_FINAL_ERROR): Likewise.
41759 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
41760 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
41761 translatable string.
41762 * config/i386/i386.c (function_value_32): Likewise.
41763 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
41764 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
41766 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
41767 * common/config/msp430/msp430-common.c (msp430_handle_option):
41769 * symtab.c (symtab_node::verify_base): Likewise.
41770 * opts.c (set_debug_level): Likewise.
41771 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
41772 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
41773 missing whitespace to translatable strings.
41774 * config/avr/avr.md (bswapsi2): Fix typo in comment.
41775 * config/sh/superh.h: Likewise.
41776 * config/i386/xopintrin.h: Likewise.
41777 * config/i386/znver1.md: Likewise.
41778 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
41779 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
41780 * double-int.h (struct double_int): Likewise.
41781 * double-int.c (div_and_round_double): Likewise.
41782 * wide-int.cc: Likewise.
41783 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
41784 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
41785 * cfgcleanup.c (crossjumps_occured): Renamed to ...
41786 (crossjumps_occurred): ... this.
41787 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
41790 PR tree-optimization/78899
41791 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
41792 returning bool return struct loop *, NULL for failure and the new
41794 (versionable_outer_loop_p): Don't version outer loop if it has
41795 dont_vectorized bit set.
41796 (tree_if_conversion): When versioning outer loop, ensure
41797 tree_if_conversion is performed also on the inner loop of the
41798 non-vectorizable outer loop copy.
41799 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
41800 LOOP_VECTORIZED in inner loop of the scalar outer loop and
41801 prevent vectorization of it.
41802 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
41803 the outer loop vectorization of the non-scalar version is attempted
41804 before vectorization of the inner loop in scalar version. If
41805 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
41806 vectorization of its inner loop.
41807 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
41808 has 2 inner loops, rename also on edges from bb whose single pred
41809 is outer_loop->header. Fix typo in function comment.
41811 2017-01-09 Martin Sebor <msebor@redhat.com>
41814 * asan.c (asan_emit_stack_protection): Increase local buffer size
41815 to avoid snprintf truncation warning.
41817 2017-01-09 Andrew Pinski <apinski@cavium.com>
41819 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
41820 to reference thunderx2t99 for the tuning structure
41821 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
41823 (thunderx2t99_extra_costs): This.
41824 * config/aarch64/aarch64-tune.md: Regenerate.
41825 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
41826 (vulcan_addrcost_table): This.
41827 (vulcan_regmove_cost): Rename to ...
41828 (thunderx2t99_regmove_cost): This.
41829 (vulcan_vector_cost): Rename to ...
41830 (thunderx2t99_vector_cost): this.
41831 (vulcan_branch_cost): Rename to ...
41832 (thunderx2t99_branch_cost): This.
41833 (vulcan_tunings): Rename to ...
41834 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
41835 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
41837 2017-01-09 Martin Jambor <mjambor@suse.cz>
41841 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
41842 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
41843 (propagate_vr_accross_jump_function): Use the above function for all
41844 value range computations for pass-through jump functions and type
41845 converasion from explicit value range values.
41846 (ipcp_propagate_stage): Do not attempt to deduce types of formal
41847 parameters from TYPE_ARG_TYPES.
41848 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41849 (ipa_write_node_info): Stream type of the actual argument.
41850 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41852 2017-01-09 Martin Liska <mliska@suse.cz>
41855 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41856 (lookup_compiler): Do not show error message with have_E.
41858 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41860 PR tree-optimization/78938
41861 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41862 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41863 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41864 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41867 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41869 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41872 2017-01-09 Richard Biener <rguenther@suse.de>
41874 PR tree-optimization/78997
41875 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41876 name condition properly.
41878 2017-01-09 Richard Biener <rguenther@suse.de>
41881 * dwarf2out.c (is_cxx): New overload with context.
41882 (is_naming_typedef_decl): Use it.
41884 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41886 * invoke.texi (Option Summary): Correct spacing in option lists
41887 and add line breaks to fix over-long lines.
41889 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41891 PR middle-end/17660
41893 * extend.texi (Common Variable Attributes): Add xref to GCC
41894 Internals manual to explain mode attribute keywords.
41896 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41899 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41900 and Preprocessor Options.
41901 (Options for Linking): Document -pthread here....
41902 (RS/6000 and PowerPC Options): ...not here.
41903 (Solaris 2 Options): ...or here.
41904 * doc/cppopts.texi: Document -pthread.
41906 2017-01-08 Martin Sebor <msebor@redhat.com>
41908 PR middle-end/77708
41909 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41910 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41911 New member functions.
41912 (format_directive): Used them.
41914 (pass_sprintf_length::handle_gimple_call): Same.
41915 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41916 to avoid truncation for any argument.
41917 (extract_affine_mul): Same.
41918 * tree.c (get_file_function_name): Same.
41920 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41922 PR middle-end/77484
41923 * predict.def (PRED_INDIR_CALL): Set to 86.
41925 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41927 PR preprocessor/54124
41928 * doc/cppopts.texi: Reformat -d subtable to list the full name
41929 of the options. Add cross-reference to the docs for the general
41930 compiler -d options.
41931 * doc/invoke.texi (Developer Options): Add cross-reference to the
41932 preprocessor-specific -d option documentation.
41934 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41936 PR preprocessor/13498
41937 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41938 redudant material, and reflect new command-line options.
41939 (System Headers): Likewise.
41941 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41943 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41944 -isystem, and -idirafter. Copy-edit.
41945 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41946 default for -ftrack-macro-expansion. Delete obsolete and
41947 badly-formatted implementation details about -fdebug-cpp output.
41948 * doc/cppwarnopts.texi: Copy-edit.
41950 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41953 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41954 that the transition from a max line width >= 1<<10 to narrower
41955 lines works correctly.
41957 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41959 * doc/options.texi (PerFunction): New.
41960 * opt-functions.awk (switch_flags): Map both Optimization and
41961 PerFunction to CL_OPTIMIZATION.
41962 * opth-gen.awk: Test for PerFunction flag along with
41964 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41965 it only when the latter is present. Skip those that don't in
41966 the hash function generator.
41967 * common.opt (fvar-tracking): Mark as PerFunction instead of
41969 (fvar-tracking-assignments): Likewise.
41970 (fvar-tracking-assignments-toggle): Likewise.
41971 (fvar-tracking-uninit): Likewise.
41973 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41975 PR translation/79018
41976 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41979 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41982 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41983 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41984 TARGET_LONG_JUMP_TABLE_OFFSETS.
41985 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41986 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41987 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41988 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41989 * config/m68k/m68k.md (tablejump expander): Likewise.
41990 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41991 TARGET_LONG_JUMP_TABLE_OFFSETS.
41992 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41993 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41995 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41996 David Holsgrove <david.holsgrove@xilinx.com>
41998 * common/config/microblaze/microblaze-common.c
41999 (TARGET_EXCEPT_UNWIND_INFO): Remove.
42000 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
42002 * config/microblaze/microblaze.c (microblaze_must_save_register)
42003 (microblaze_expand_epilogue, microblaze_return_addr): Handle
42005 (microblaze_eh_return): New function.
42006 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
42007 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
42008 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
42009 * config/microblaze/microblaze.md (eh_return): New pattern.
42011 2017-01-06 Jakub Jelinek <jakub@redhat.com>
42013 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
42014 GCC_DIAGNOSTIC_STRINGIFY): Define.
42016 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
42018 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
42020 * config/arm/arm.md (<mcrr>): New.
42022 * config/arm/arm.c (arm_arch5te): New.
42023 (arm_option_override): Set arm_arch5te.
42024 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
42026 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
42027 (arm_mcrr_qualifiers): ... this. New.
42028 (MRRC_QUALIFIERS): Define to...
42029 (arm_mrrc_qualifiers): ... this. New.
42030 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
42032 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
42033 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
42034 (MRRCI, mrrc, MRRC): New.
42035 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
42036 VUNSPEC_MRRC2): New.
42038 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
42040 * config/arm/arm.md (<mcr>): New.
42042 * config/arm/arm.c (arm_coproc_builtin_available): Add
42043 support for mcr, mrc, mcr2 and mrc2.
42044 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
42045 (arm_mcr_qualifiers): ... this. New.
42046 (MRC_QUALIFIERS): Define to ...
42047 (arm_mrc_qualifiers): ... this. New.
42048 (MCR_QUALIFIERS): Define to ...
42049 (arm_mcr_qualifiers): ... this. New.
42050 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
42052 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
42053 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
42054 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
42055 VUNSPEC_MRC2): New.
42057 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
42059 * config/arm/arm.md (*ldc): New.
42063 * config/arm/arm.c (arm_coproc_builtin_available): Add
42064 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
42065 (arm_coproc_ldc_stc_legitimate_address): New.
42066 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
42067 'qualifier_const_pointer'.
42068 (LDC_QUALIFIERS): Define to...
42069 (arm_ldc_qualifiers): ... this. New.
42070 (STC_QUALIFIERS): Define to...
42071 (arm_stc_qualifiers): ... this. New.
42072 * config/arm/arm-protos.h
42073 (arm_coproc_ldc_stc_legitimate_address): New.
42074 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
42075 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
42076 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
42077 stc2, stcl, stc2l): New.
42078 * config/arm/constraints.md (Uz): New.
42079 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
42080 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
42081 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
42082 VUNSPEC_STC2L): New.
42084 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
42086 * config/arm/arm.md (<cdp>): New.
42087 * config/arm/arm.c (neon_const_bounds): Rename this ...
42088 (arm_const_bounds): ... this.
42089 (arm_coproc_builtin_available): New.
42090 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
42091 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
42092 (CDP_QUALIFIERS): Define to...
42093 (arm_cdp_qualifiers): ... this. New.
42095 (arm_expand_builtin_args): Add case for 6 arguments.
42096 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
42097 (arm_const_bounds): ... this.
42098 (arm_coproc_builtin_available): New.
42099 * config/arm/arm_acle.h (__arm_cdp): New.
42101 * config/arm/arm_acle_builtins.def (cdp): New.
42103 * config/arm/iterators.md (CDPI,CDP,cdp): New.
42104 * config/arm/neon.md: Rename all 'neon_const_bounds' to
42105 'arm_const_bounds'.
42106 * config/arm/types.md (coproc): New.
42107 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
42108 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
42109 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
42110 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
42112 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
42114 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
42115 (UBINOP_QUALIFIERS): New.
42117 (acle_builtin_data): New. Change comment.
42118 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
42119 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
42120 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
42121 arm_acle_builtins.def.
42122 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
42123 (arm_init_acle_builtins): New.
42124 (CRC32_BUILTIN): Remove.
42125 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
42126 crc32cb, crc32ch and crc32cw.
42127 (arm_init_crc32_builtins): Remove.
42128 (arm_init_builtins): Use arm_init_acle_builtins rather
42129 than arm_init_crc32_builtins.
42130 (arm_expand_acle_builtin): New.
42131 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
42132 * config/arm/arm_acle_builtins.def: New.
42134 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
42136 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
42137 (arm_builtin_datum): ... this.
42138 (arm_init_neon_builtin): Rename to ...
42139 (arm_init_builtin): ... this. Add a new parameters PREFIX
42140 and USE_SIG_IN_NAME.
42141 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
42142 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
42143 'arm_builtin_datum'.
42144 (arm_init_vfp_builtins): Likewise.
42145 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
42146 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
42147 (arm_expand_neon_args): Rename to ...
42148 (arm_expand_builtin_args): ... this. Rename builtin_arg
42149 enum values and differentiate between ARG_BUILTIN_MEMORY
42150 and ARG_BUILTIN_NEON_MEMORY.
42151 (arm_expand_neon_builtin_1): Rename to ...
42152 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
42153 values, arm_expand_builtin_args and add bool parameter NEON.
42154 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
42155 (arm_expand_vfp_builtin): Likewise.
42156 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
42158 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42160 PR middle-end/77484
42161 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
42162 * predict.c (tree_estimate_probability_bb): Reverse direction of
42163 polymorphic call predictor.
42165 2017-01-06 David Malcolm <dmalcolm@redhat.com>
42167 * passes.c (execute_one_pass): Split out pass-skipping logic into...
42168 (determine_pass_name_match): ...this new function and...
42169 (should_skip_pass_p): ...this new function.
42171 2017-01-06 Nathan Sidwell <nathan@acm.org>
42173 * ipa-visibility.c (function_and_variable_visibility): Reformat
42174 comments and long lines. Remove extrneous if.
42175 * symtab.c (symtab_node::make_decl_local): Fix code format.
42176 (symtab_node::set_section_for_node): Fix comment typo.
42178 2017-01-06 Martin Liska <mliska@suse.cz>
42181 * lra-constraints.c: Rename invariant to lra_invariant.
42182 * predict.c (set_even_probabilities): Initialize e to NULL.
42184 2017-01-05 Martin Sebor <msebor@redhat.com>
42186 PR tree-optimization/78910
42187 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
42188 (format_integer): Correct off-by-one error in the handling
42189 of precision with negative numbers in signed conversions..
42191 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
42193 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
42195 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42197 PR tree-optimization/71016
42198 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
42199 factor_out_conditional_conversion. Formatting fix.
42200 (factor_out_conditional_conversion): Add cond_stmt argument.
42201 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
42202 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
42205 2017-01-05 David Malcolm <dmalcolm@redhat.com>
42207 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
42208 read-rtl-function.o, and selftest-rtl.o.
42209 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
42210 (selftest::aarch64_test_loading_full_dump): New function.
42211 (selftest::aarch64_run_selftests): New function.
42212 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
42213 selftest::aarch64_run_selftests.
42214 * config/i386/i386.c
42215 (selftest::ix86_test_loading_dump_fragment_1): New function.
42216 (selftest::ix86_test_loading_call_insn): New function.
42217 (selftest::ix86_test_loading_full_dump): New function.
42218 (selftest::ix86_test_loading_unspec): New function.
42219 (selftest::ix86_run_selftests): Call the new functions.
42220 * emit-rtl.c (maybe_set_max_label_num): New function.
42221 * emit-rtl.h (maybe_set_max_label_num): New decl.
42222 * function.c (instantiate_decls): Guard call to
42223 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
42224 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
42226 * gensupport.c (gen_reader::gen_reader): Pass "false"
42227 for new "compact" param of rtx_reader.
42228 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
42229 rather than an empty string for NULL strings.
42230 * read-md.c: Potentially include config.h rather than bconfig.h.
42231 Wrap include of errors.h with #ifdef GENERATOR_FILE.
42232 (have_error): New global, copied from errors.c.
42233 (md_reader::read_name): Rename to...
42234 (md_reader::read_name_1): ...this, adding "out_loc" param,
42235 and converting "missing name or number" to returning false, rather
42237 (md_reader::read_name): Reimplement in terms of read_name_1.
42238 (md_reader::read_name_or_nil): New function.
42239 (md_reader::read_string): Handle "(nil)" by returning NULL.
42240 (md_reader::md_reader): Add new param "compact".
42241 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
42242 (md_reader::read_file): New method.
42243 * read-md.h (md_reader::md_reader): Add new param "compact".
42244 (md_reader::read_file): New method.
42245 (md_reader::is_compact): New accessor.
42246 (md_reader::read_name): Convert return type from void to file_location.
42247 (md_reader::read_name_or_nil): New decl.
42248 (md_reader::read_name_1): New decl.
42249 (md_reader::m_compact): New field.
42250 (noop_reader::noop_reader): Pass "false" for new "compact" param
42252 (rtx_reader::rtx_reader): Add new "compact" param.
42253 (rtx_reader::read_rtx_operand): Make virtual and convert return
42254 type from void to rtx.
42255 (rtx_reader::read_until): New decl.
42256 (rtx_reader::handle_any_trailing_information): New virtual function.
42257 (rtx_reader::postprocess): New virtual function.
42258 (rtx_reader::finalize_string): New virtual function.
42259 (rtx_reader::m_in_call_function_usage): New field.
42260 (rtx_reader::m_reuse_rtx_by_id): New field.
42261 * read-rtl-function.c: New file.
42262 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
42263 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
42264 (selftest::verify_three_block_rtl_cfg): New decl.
42265 * read-rtl-function.h: New file.
42266 * read-rtl.c: Potentially include config.h rather than bconfig.h.
42267 For host, include function.h, memmodel.h, and emit-rtl.h.
42268 (one_time_initialization): New function.
42269 (struct compact_insn_name): New struct.
42270 (compact_insn_names): New array.
42271 (find_code): Handle insn codes in compact dumps.
42272 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
42273 (bind_subst_iter_and_attr): Likewise.
42274 (add_condition_to_string): Likewise.
42275 (add_condition_to_rtx): Likewise.
42276 (apply_attribute_uses): Likewise.
42277 (add_current_iterators): Likewise.
42278 (apply_iterators): Likewise.
42279 (initialize_iterators): Guard usage of apply_subst_iterator with
42280 #ifdef GENERATOR_FILE.
42281 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
42282 (md_reader::read_mapping): Likewise.
42283 (add_define_attr_for_define_subst): Likewise.
42284 (add_define_subst_attr): Likewise.
42285 (read_subst_mapping): Likewise.
42286 (check_code_iterator): Likewise.
42287 (rtx_reader::read_rtx): Likewise. Move one-time initialization
42289 (one_time_initialization): New function.
42290 (rtx_reader::read_until): New method.
42291 (read_flags): New function.
42292 (parse_reg_note_name): New function.
42293 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
42294 Handle reuse_rtx ids.
42295 Wrap iterator lookup within #ifdef GENERATOR_FILE.
42296 Add parsing support for RTL dumps, mirroring the special-cases in
42297 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
42298 values, and calling handle_any_trailing_information.
42299 (rtx_reader::read_rtx_operand): Convert return type from void
42300 to rtx, returning return_rtx. Handle case 'e'. Call
42301 finalize_string on XSTR and XTMPL fields.
42302 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
42303 "(nil)" values were omitted. Call the postprocess vfunc on the
42305 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
42306 class ctor. Initialize m_in_call_function_usage. Call
42307 one_time_initialization.
42308 * rtl-tests.c (selftest::test_uncond_jump): Call
42309 set_new_first_and_last_insn.
42310 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
42311 * selftest-rtl.c: New file.
42312 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
42313 (selftest::get_insn_by_uid): New decl.
42314 * selftest-run-tests.c (selftest::run_tests): Call
42315 read_rtl_function_c_tests.
42316 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
42317 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
42320 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
42322 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
42323 operands in a special way. Assert that pos+len <= mode precision.
42325 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42327 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
42328 3 argument Alias with unlimited for the negative form.
42329 (fno-vect-cost-model): Removed.
42331 2017-01-05 Martin Liska <mliska@suse.cz>
42333 * hsa-gen.c (gen_hsa_divmod): New function.
42334 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
42336 2017-01-05 Martin Liska <mliska@suse.cz>
42339 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
42342 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42344 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
42345 small constant length operands.
42347 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42349 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
42350 between loop iterations.
42352 2017-01-05 Martin Liska <mliska@suse.cz>
42355 * gimplify.c (gimplify_decl_expr): Compare to
42356 asan_poisoned_variables instread of checking flags.
42357 (gimplify_target_expr): Likewise.
42358 (gimplify_expr): Likewise.
42359 (gimplify_function_tree): Conditionally initialize
42360 asan_poisoned_variables.
42362 2017-01-04 Jeff Law <law@redhat.com>
42364 PR tree-optimizatin/78812
42365 * rtl.h (contains_mem_rtx_p): Prototype.
42366 * ifcvt.c (containts_mem_rtx_p): Move from here to...
42367 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
42368 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
42369 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
42370 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
42372 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42374 * input.c (assert_char_at_range): Default-initialize actual_range.
42376 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42378 * df-scan.c (df_ref_create_structure): Make regno unsigned,
42379 to match the caller.
42381 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42383 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
42384 insns after final jump in test to emit dummy move.
42386 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42388 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
42389 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
42391 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42393 * multiple_target.c (create_dispatcher_calls): Init e_next.
42394 * tree-ssa-loop-split.c (split_loop): Init border.
42395 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
42398 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
42403 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
42404 (altivec_register_operand): Do not return true if the operand
42405 contains a SUBREG mixing SImode and SFmode.
42406 (vsx_register_operand): Likewise.
42407 (vsx_reg_sfsubreg_ok): New predicate.
42408 (vfloat_operand): Do not return true if the operand contains a
42409 SUBREG mixing SImode and SFmode.
42410 (vint_operand): Likewise.
42411 (vlogical_operand): Likewise.
42412 (gpc_reg_operand): Likewise.
42413 (int_reg_operand): Likewise.
42414 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
42415 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
42416 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
42418 (rs6000_emit_move_si_sf_subreg): New helper function.
42419 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
42420 fixup SUBREGs involving SImode and SFmode.
42421 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
42422 numbers for the new peephole2 optimization.
42423 (peephole2 for SFmode unions): New peephole2 to optimize cases in
42424 the GLIBC math library that do AND/IOR/XOR operations on single
42425 precision floating point.
42426 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
42427 target macros to say whether we need to avoid SUBREGs mixing
42429 (TARGET_ALLOW_SF_SUBREG): Likewise.
42430 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
42431 (UNSPEC_SI_FROM_SF): Likewise.
42432 (iorxor): Change spacing.
42433 (and_ior_xor): New iterator for AND, IOR, and XOR.
42434 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
42435 (movdi_from_sf_zero_ext): Likewise.
42436 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
42437 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
42438 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
42439 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
42440 (fms<mode>4): Likewise.
42441 (fnma<mode>4): Likewise.
42442 (fnms<mode>4): Likewise.
42443 (nfma<mode>4): Likewise.
42444 (nfms<mode>4): Likewise.
42446 2017-01-04 Marek Polacek <polacek@redhat.com>
42449 * doc/invoke.texi: Document -Wpointer-compare.
42451 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42453 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
42456 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
42457 descriptions for -gdwarf-5 and emit them as uleb128 instead of
42460 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
42463 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
42464 documentation of the powerpc_popcntb_ok attribute.
42465 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
42466 code to issue warning messages if a requested CPU configuration is
42467 not supported by the binary (assembler and loader) toolchain.
42468 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
42469 made to define a built-in function that has been disabled.
42470 (paired_init_builtins): Add assertion to prevent ICE if attempt is
42471 made to define a built-in function that has been disabled.
42472 (altivec_init_builtins): Add comment explaining why definition
42473 of the DST built-in functions is not preceded by an assertion
42474 check. Add assertions to prevent ICE if attempts are made to
42475 define an altivec predicate or an abs* built-in function that has
42477 (htm_init_builtins): Add comment explaining why definition of the
42478 htm built-in functions is not preceded by an assertion check.
42480 2017-01-04 Jeff Law <law@redhat.com>
42482 PR tree-optimizatin/67955
42483 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
42484 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
42485 the points-to solution does not include pt_null. Use DECL_PT_UID
42488 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
42490 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
42491 Use gen_int_mode instead of gen_lopwart for const_int operands.
42493 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42495 PR tree-optimization/71563
42496 * match.pd: Simplify X << Y into X if Y is known to be 0 or
42497 out of range value - has low bits known to be zero.
42499 2017-01-04 Alan Modra <amodra@gmail.com>
42501 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
42502 * configure: Regenerate.
42503 * config.in: Regenerate.
42505 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42508 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
42509 a substring of the message, but strcmp with the whole message. Ifdef
42510 ENABLE_NLS, translate the message first using dgettext.
42512 2017-01-03 Jeff Law <law@redhat.com>
42514 PR tree-optimizatin/78856
42515 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
42516 (mark_threaded_blocks): Remove code to truncate thread paths that
42517 cross multiple loop headers. Instead invalidate the cached loop
42518 iteration information and handle case of a thread path walking
42519 into an irreducible region.
42521 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
42524 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
42525 assertions. Add support for doing the signbit if the IEEE 128-bit
42526 floating point value is in a GPR.
42527 * config/rs6000/rs6000.md (Fsignbit): Delete.
42528 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
42529 Update the length attribute if the value is in a GPR.
42530 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
42531 the sign or zero extension instruction, since the value is always 0/1.
42532 (signbit<mode>2_dm2): Delete using <Fsignbit>.
42535 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
42536 extracting SImode to a GPR register so that we can generate a
42537 store, limit the vector to be in a traditional Altivec register
42538 for the vextuwrx instruction.
42540 2017-01-03 Ian Lance Taylor <iant@google.com>
42542 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
42544 2017-01-03 Martin Sebor <msebor@redhat.com>
42546 PR tree-optimization/78696
42547 * gimple-ssa-sprintf.c (format_floating): Correct handling of
42548 precision. Use MPFR for %f for greater fidelity. Correct handling
42550 (pass_sprintf_length::compute_format_length): Set width and precision
42551 specified by asrerisk to void_node for vararg functions.
42552 (try_substitute_return_value): Adjust dump output.
42554 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
42556 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
42558 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
42560 * doc/invoke.texi (SPARC options): Document -mlra as the default.
42561 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
42562 -mlra/-mno-lra was passed to the compiler.
42564 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
42566 PR rtl-optimization/65618
42567 * emit-rtl.c (try_split): Move initialization of "before" and
42568 "after" to just before the call to emit_insn_after_setloc.
42570 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
42572 * doc/md.texi (Standard Names): Remove reference to Java frontend.
42574 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
42576 * dwarf2out.c (gen_enumeration_type_die): When
42577 -gno-strict-dwarf, add a DW_AT_encoding attribute.
42579 2017-01-03 Jakub Jelinek <jakub@redhat.com>
42581 PR tree-optimization/78965
42582 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
42583 Change first argument from const call_info & to call_info &. For %n
42584 set info.nowrite to false.
42586 PR middle-end/78901
42587 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
42588 possibly throwing calls.
42590 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
42591 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
42592 and fns handling, rather than in a separate case SSA_NAME.
42594 2017-01-02 Jeff Law <law@redhat.com>
42596 * config/darwin-driver.c (darwin_driver_init): Const-correctness
42597 fixes for first_period and second_period variables.
42599 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
42602 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
42603 (*insvqi_1): New insn pattern.
42604 (*insvqi_1_mem_rex64): Ditto.
42605 (*insvqi_2): Ditto.
42606 (*insvqi_3): Rename from *insvqi.
42608 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
42610 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
42612 * doc/cfg.texi (Edges): Remove reference to Java.
42613 (Maintaining the CFG): Ditto.
42615 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42617 PR middle-end/77674
42618 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
42619 transparent aliases.
42621 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42623 PR middle-end/77484
42624 * predict.def (PRED_CALL): Update hitrate.
42625 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
42626 * predict.c (tree_estimate_probability_bb): Split CALL predictor
42627 into direct/indirect/polymorphic variants.
42629 2017-01-01 Jakub Jelinek <jakub@redhat.com>
42631 Update copyright years.
42633 * gcc.c (process_command): Update copyright notice dates.
42634 * gcov-dump.c (print_version): Ditto.
42635 * gcov.c (print_version): Ditto.
42636 * gcov-tool.c (print_version): Ditto.
42637 * gengtype.c (create_file): Ditto.
42638 * doc/cpp.texi: Bump @copying's copyright year.
42639 * doc/cppinternals.texi: Ditto.
42640 * doc/gcc.texi: Ditto.
42641 * doc/gccint.texi: Ditto.
42642 * doc/gcov.texi: Ditto.
42643 * doc/install.texi: Ditto.
42644 * doc/invoke.texi: Ditto.
42646 Copyright (C) 2017 Free Software Foundation, Inc.
42648 Copying and distribution of this file, with or without modification,
42649 are permitted in any medium without royalty provided the copyright
42650 notice and this notice are preserved.