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 * expr.c (expand_expr_real_2): When handling conversions involving
6 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
7 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
10 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
11 Alan Hayward <alan.hayward@arm.com>
12 David Sherwood <david.sherwood@arm.com>
14 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
16 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17 Alan Hayward <alan.hayward@arm.com>
18 David Sherwood <david.sherwood@arm.com>
20 * combine.c (can_change_dest_mode): Handle polynomial
22 * expmed.c (store_bit_field_1): Likewise.
23 * expr.c (store_constructor): Likewise.
24 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
25 and polynomial REGMODE_NATURAL_SIZE.
26 (gen_lowpart_common): Likewise.
27 * reginfo.c (record_subregs_of_mode): Likewise.
28 * rtlanal.c (read_modify_subreg_p): Likewise.
30 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31 Alan Hayward <alan.hayward@arm.com>
32 David Sherwood <david.sherwood@arm.com>
34 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
37 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
38 Alan Hayward <alan.hayward@arm.com>
39 David Sherwood <david.sherwood@arm.com>
41 * match.pd: Cope with polynomial numbers of vector elements.
43 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
44 Alan Hayward <alan.hayward@arm.com>
45 David Sherwood <david.sherwood@arm.com>
47 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
48 in a POINTER_PLUS_EXPR.
50 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
51 Alan Hayward <alan.hayward@arm.com>
52 David Sherwood <david.sherwood@arm.com>
54 * omp-simd-clone.c (simd_clone_subparts): New function.
55 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
56 (ipa_simd_modify_function_body): Likewise.
58 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
59 Alan Hayward <alan.hayward@arm.com>
60 David Sherwood <david.sherwood@arm.com>
62 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
63 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
64 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
65 (expand_vector_condition, vector_element): Likewise.
66 (subparts_gt): New function.
67 (get_compute_type): Use subparts_gt.
68 (count_type_subparts): Delete.
69 (expand_vector_operations_1): Use subparts_gt instead of
72 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
73 Alan Hayward <alan.hayward@arm.com>
74 David Sherwood <david.sherwood@arm.com>
76 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
77 (vect_compile_time_alias): ...this new function. Do the calculation
78 on poly_ints rather than trees.
79 (vect_prune_runtime_alias_test_list): Update call accordingly.
81 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
82 Alan Hayward <alan.hayward@arm.com>
83 David Sherwood <david.sherwood@arm.com>
85 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
87 (vect_schedule_slp_instance): 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 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
94 constant and extern definitions for variable-length vectors.
95 (vect_get_constant_vectors): Note that the number of units
96 is known to be constant.
98 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
99 Alan Hayward <alan.hayward@arm.com>
100 David Sherwood <david.sherwood@arm.com>
102 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
103 of units as polynomial. Choose between WIDE and NARROW based
106 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
107 Alan Hayward <alan.hayward@arm.com>
108 David Sherwood <david.sherwood@arm.com>
110 * tree-vect-stmts.c (simd_clone_subparts): New function.
111 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
113 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
114 Alan Hayward <alan.hayward@arm.com>
115 David Sherwood <david.sherwood@arm.com>
117 * tree-vect-stmts.c (vectorizable_call): Treat the number of
118 vectors as polynomial. Use build_index_vector for
121 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
122 Alan Hayward <alan.hayward@arm.com>
123 David Sherwood <david.sherwood@arm.com>
125 * tree-vect-stmts.c (get_load_store_type): Treat the number of
126 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
127 for variable-length vectors.
128 (vectorizable_mask_load_store): Treat the number of units as
129 polynomial, asserting that it is constant if the condition has
130 already been enforced.
131 (vectorizable_store, vectorizable_load): Likewise.
133 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
134 Alan Hayward <alan.hayward@arm.com>
135 David Sherwood <david.sherwood@arm.com>
137 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
138 of units as polynomial. Punt if we can't tell at compile time
139 which vector contains the final result.
141 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
142 Alan Hayward <alan.hayward@arm.com>
143 David Sherwood <david.sherwood@arm.com>
145 * tree-vect-loop.c (vectorizable_induction): Treat the number
146 of units as polynomial. Punt on SLP inductions. Use an integer
147 VEC_SERIES_EXPR for variable-length integer reductions. Use a
148 cast of such a series for variable-length floating-point
151 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
152 Alan Hayward <alan.hayward@arm.com>
153 David Sherwood <david.sherwood@arm.com>
155 * tree.h (build_index_vector): Declare.
156 * tree.c (build_index_vector): New function.
157 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
158 of units as polynomial, forcibly converting it to a constant if
159 vectorizable_reduction has already enforced the condition.
160 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
161 to create a {1,2,3,...} vector.
162 (vectorizable_reduction): Treat the number of units as polynomial.
163 Choose vectype_in based on the largest scalar element size rather
164 than the smallest number of units. Enforce the restrictions
167 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
168 Alan Hayward <alan.hayward@arm.com>
169 David Sherwood <david.sherwood@arm.com>
171 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
172 number of units as polynomial.
174 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
175 Alan Hayward <alan.hayward@arm.com>
176 David Sherwood <david.sherwood@arm.com>
178 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
179 * target.def (autovectorize_vector_sizes): Return the vector sizes
180 by pointer, using vector_sizes rather than a bitmask.
181 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
182 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
183 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
185 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
186 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
187 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
188 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
189 * omp-general.c (omp_max_vf): Likewise.
190 * omp-low.c (omp_clause_aligned_alignment): Likewise.
191 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
192 * tree-vect-loop.c (vect_analyze_loop): Likewise.
193 * tree-vect-slp.c (vect_slp_bb): Likewise.
194 * doc/tm.texi: Regenerate.
195 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
197 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
198 the vector size as a poly_uint64 rather than an unsigned int.
199 (current_vector_size): Change from an unsigned int to a poly_uint64.
200 (get_vectype_for_scalar_type): Update accordingly.
201 * tree.h (build_truth_vector_type): Take the size and number of
202 units as a poly_uint64 rather than an unsigned int.
203 (build_vector_type): Add a temporary overload that takes
204 the number of units as a poly_uint64 rather than an unsigned int.
205 * tree.c (make_vector_type): Likewise.
206 (build_truth_vector_type): Take the number of units as a poly_uint64
207 rather than an unsigned int.
209 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
210 Alan Hayward <alan.hayward@arm.com>
211 David Sherwood <david.sherwood@arm.com>
213 * target.def (get_mask_mode): Take the number of units and length
214 as poly_uint64s rather than unsigned ints.
215 * targhooks.h (default_get_mask_mode): Update accordingly.
216 * targhooks.c (default_get_mask_mode): Likewise.
217 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
218 * doc/tm.texi: Regenerate.
220 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
221 Alan Hayward <alan.hayward@arm.com>
222 David Sherwood <david.sherwood@arm.com>
224 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
225 * omp-general.c (omp_max_vf): Likewise.
226 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
227 (expand_omp_simd): Handle polynomial safelen.
228 * omp-low.c (omplow_simd_context): Add a default constructor.
229 (omplow_simd_context::max_vf): Change from int to poly_uint64.
230 (lower_rec_simd_input_clauses): Update accordingly.
231 (lower_rec_input_clauses): Likewise.
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 * tree-vectorizer.h (vect_nunits_for_cost): New function.
238 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
239 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
240 (vect_analyze_slp_cost): Likewise.
241 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
242 (vect_model_load_cost): Likewise.
244 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
245 Alan Hayward <alan.hayward@arm.com>
246 David Sherwood <david.sherwood@arm.com>
248 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
249 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
250 from an unsigned int * to a poly_uint64_pod *.
251 (calculate_unrolling_factor): New function.
252 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
254 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
255 Alan Hayward <alan.hayward@arm.com>
256 David Sherwood <david.sherwood@arm.com>
258 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
259 from an unsigned int to a poly_uint64.
260 (_loop_vec_info::slp_unrolling_factor): Likewise.
261 (_loop_vec_info::vectorization_factor): Change from an int
263 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
264 (vect_get_num_vectors): New function.
265 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
266 (vect_get_num_copies): Use vect_get_num_vectors.
267 (vect_analyze_data_ref_dependences): Change max_vf from an int *
268 to an unsigned int *.
269 (vect_analyze_data_refs): Change min_vf from an int * to a
271 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
272 than an unsigned HOST_WIDE_INT.
273 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
274 (vect_analyze_data_ref_dependence): Change max_vf from an int *
275 to an unsigned int *.
276 (vect_analyze_data_ref_dependences): Likewise.
277 (vect_compute_data_ref_alignment): Handle polynomial vf.
278 (vect_enhance_data_refs_alignment): Likewise.
279 (vect_prune_runtime_alias_test_list): Likewise.
280 (vect_shift_permute_load_chain): Likewise.
281 (vect_supportable_dr_alignment): Likewise.
282 (dependence_distance_ge_vf): Take the vectorization factor as a
283 poly_uint64 rather than an unsigned HOST_WIDE_INT.
284 (vect_analyze_data_refs): Change min_vf from an int * to a
286 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
287 vfm1 as a poly_uint64 rather than an int. Make the same change
288 for the returned bound_scalar.
289 (vect_gen_vector_loop_niters): Handle polynomial vf.
290 (vect_do_peeling): Likewise. Update call to
291 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
292 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
294 * tree-vect-loop.c (vect_determine_vectorization_factor)
295 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
296 (vect_get_known_peeling_cost): Likewise.
297 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
298 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
299 (vect_transform_loop): Likewise. Use the lowest possible VF when
300 updating the upper bounds of the loop.
301 (vect_min_worthwhile_factor): Make static. Return an unsigned int
303 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
304 polynomial unroll factors.
305 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
306 (vect_make_slp_decision): Likewise.
307 (vect_supported_load_permutation_p): Likewise, and polynomial
309 (vect_analyze_slp_cost): Handle polynomial vf.
310 (vect_slp_analyze_node_operations): Likewise.
311 (vect_slp_analyze_bb_1): Likewise.
312 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
313 than an unsigned HOST_WIDE_INT.
314 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
315 (vectorizable_load): Handle polynomial vf.
316 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
318 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
320 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
321 Alan Hayward <alan.hayward@arm.com>
322 David Sherwood <david.sherwood@arm.com>
324 * match.pd: Handle bit operations involving three constants
325 and try to fold one pair.
327 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
329 * tree-vect-loop-manip.c: Include gimple-fold.h.
330 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
331 niters_maybe_zero parameters. Handle other cases besides a step of 1.
332 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
333 Add a path that uses a step of VF instead of 1, but disable it
335 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
336 and niters_no_overflow parameters. Update calls to
337 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
338 Create a new SSA name if the latter choses to use a ste other
339 than zero, and return it via niters_vector_mult_vf_var.
340 * tree-vect-loop.c (vect_transform_loop): Update calls to
341 vect_do_peeling, vect_gen_vector_loop_niters and
342 slpeel_make_loop_iterate_ntimes.
343 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
344 (vect_gen_vector_loop_niters): Update declarations after above changes.
346 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
348 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
349 128-bit round to integer instructions.
350 (ceil<mode>2): Likewise.
351 (btrunc<mode>2): Likewise.
352 (round<mode>2): Likewise.
354 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
356 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
357 unaligned VSX load/store on P8/P9.
358 (expand_block_clear): Allow the use of unaligned VSX
361 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
363 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
365 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
366 swap associated with both a load and a store.
368 2018-01-02 Andrew Waterman <andrew@sifive.com>
370 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
371 * config/riscv/riscv.md (clear_cache): Use it.
373 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
375 * web.c: Remove out-of-date comment.
377 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
379 * expr.c (fixup_args_size_notes): Check that any existing
380 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
381 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
382 (emit_single_push_insn): ...here.
384 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
386 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
387 (const_vector_encoded_nelts): New function.
388 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
389 (const_vector_int_elt, const_vector_elt): Declare.
390 * emit-rtl.c (const_vector_int_elt_1): New function.
391 (const_vector_elt): Likewise.
392 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
395 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
397 * expr.c: Include rtx-vector-builder.h.
398 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
399 directly on the tree encoding.
400 (const_vector_from_tree): Likewise.
401 * optabs.c: Include rtx-vector-builder.h.
402 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
403 sequence of "u" values.
404 * vec-perm-indices.c: Include rtx-vector-builder.h.
405 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
406 directly on the vec_perm_indices encoding.
408 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
410 * doc/rtl.texi (const_vector): Describe new encoding scheme.
411 * Makefile.in (OBJS): Add rtx-vector-builder.o.
412 * rtx-vector-builder.h: New file.
413 * rtx-vector-builder.c: Likewise.
414 * rtl.h (rtx_def::u2): Add a const_vector field.
415 (CONST_VECTOR_NPATTERNS): New macro.
416 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
417 (CONST_VECTOR_DUPLICATE_P): Likewise.
418 (CONST_VECTOR_STEPPED_P): Likewise.
419 (CONST_VECTOR_ENCODED_ELT): Likewise.
420 (const_vec_duplicate_p): Check for a duplicated vector encoding.
421 (unwrap_const_vec_duplicate): Likewise.
422 (const_vec_series_p): Check for a non-duplicated vector encoding.
423 Say that the function only returns true for integer vectors.
424 * emit-rtl.c: Include rtx-vector-builder.h.
425 (gen_const_vec_duplicate_1): Delete.
426 (gen_const_vector): Call gen_const_vec_duplicate instead of
427 gen_const_vec_duplicate_1.
428 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
429 (gen_const_vec_duplicate): Use rtx_vector_builder.
430 (gen_const_vec_series): Likewise.
431 (gen_rtx_CONST_VECTOR): Likewise.
432 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
433 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
434 Build a new vector rather than modifying a CONST_VECTOR in-place.
435 (handle_special_swappables): Update call accordingly.
436 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
437 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
438 Build a new vector rather than modifying a CONST_VECTOR in-place.
439 (handle_special_swappables): Update call accordingly.
441 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
443 * simplify-rtx.c (simplify_const_binary_operation): Use
444 CONST_VECTOR_ELT instead of XVECEXP.
446 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
448 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
449 the selector elements to be different from the data elements
450 if the selector is a VECTOR_CST.
451 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
452 ssizetype for the selector.
454 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
456 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
457 before testing each element individually.
458 * tree-vect-generic.c (lower_vec_perm): Likewise.
460 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
462 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
463 * selftest-run-tests.c (selftest::run_tests): Call it.
464 * vector-builder.h (vector_builder::operator ==): New function.
465 (vector_builder::operator !=): Likewise.
466 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
467 (vec_perm_indices::all_from_input_p): New function.
468 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
469 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
470 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
471 instead of reading the VECTOR_CST directly. Detect whether both
472 vector inputs are the same before constructing the vec_perm_indices,
473 and update the number of inputs argument accordingly. Use the
474 utility functions added above. Only construct sel2 if we need to.
476 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
478 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
479 the broadcast of the low byte.
480 (expand_mult_highpart): Use an explicit encoding for the permutes.
481 * optabs-query.c (can_mult_highpart_p): Likewise.
482 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
483 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
484 (vectorizable_bswap): Likewise.
485 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
486 explicit encoding for the power-of-2 permutes.
487 (vect_permute_store_chain): Likewise.
488 (vect_grouped_load_supported): Likewise.
489 (vect_permute_load_chain): Likewise.
491 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
493 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
494 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
495 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
496 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
497 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
498 (vect_gen_perm_mask_any): Likewise.
500 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
502 * int-vector-builder.h: New file.
503 * vec-perm-indices.h: Include int-vector-builder.h.
504 (vec_perm_indices): Redefine as an int_vector_builder.
505 (auto_vec_perm_indices): Delete.
506 (vec_perm_builder): Redefine as a stand-alone class.
507 (vec_perm_indices::vec_perm_indices): New function.
508 (vec_perm_indices::clamp): Likewise.
509 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
510 (vec_perm_indices::new_vector): New function.
511 (vec_perm_indices::new_expanded_vector): Update for new
512 vec_perm_indices class.
513 (vec_perm_indices::rotate_inputs): New function.
514 (vec_perm_indices::all_in_range_p): Operate directly on the
515 encoded form, without computing elided elements.
516 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
517 encoding. Update for new vec_perm_indices class.
518 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
519 the given vec_perm_builder.
520 (expand_vec_perm_var): Update vec_perm_builder constructor.
521 (expand_mult_highpart): Use vec_perm_builder instead of
522 auto_vec_perm_indices.
523 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
524 vec_perm_indices instead of auto_vec_perm_indices. Use a single
525 or double series encoding as appropriate.
526 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
527 vec_perm_indices instead of auto_vec_perm_indices.
528 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
529 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
530 (vect_permute_store_chain): Likewise.
531 (vect_grouped_load_supported): Likewise.
532 (vect_permute_load_chain): Likewise.
533 (vect_shift_permute_load_chain): Likewise.
534 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
535 (vect_transform_slp_perm_load): Likewise.
536 (vect_schedule_slp_instance): Likewise.
537 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
538 (vectorizable_mask_load_store): Likewise.
539 (vectorizable_bswap): Likewise.
540 (vectorizable_store): Likewise.
541 (vectorizable_load): Likewise.
542 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
543 vec_perm_indices instead of auto_vec_perm_indices. Use
544 tree_to_vec_perm_builder to read the vector from a tree.
545 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
546 vec_perm_builder instead of a vec_perm_indices.
547 (have_whole_vector_shift): Use vec_perm_builder and
548 vec_perm_indices instead of auto_vec_perm_indices. Leave the
549 truncation to calc_vec_perm_mask_for_shift.
550 (vect_create_epilog_for_reduction): Likewise.
551 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
552 from auto_vec_perm_indices to vec_perm_indices.
553 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
554 instead of changing individual elements.
555 (aarch64_vectorize_vec_perm_const): Use new_vector to install
556 the vector in d.perm.
557 * config/arm/arm.c (expand_vec_perm_d::perm): Change
558 from auto_vec_perm_indices to vec_perm_indices.
559 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
560 instead of changing individual elements.
561 (arm_vectorize_vec_perm_const): Use new_vector to install
562 the vector in d.perm.
563 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
564 Update vec_perm_builder constructor.
565 (rs6000_expand_interleave): Likewise.
566 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
567 (rs6000_expand_interleave): Likewise.
569 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
571 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
572 to qimode could truncate the indices.
573 * optabs.c (expand_vec_perm_var): Likewise.
575 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
577 * Makefile.in (OBJS): Add vec-perm-indices.o.
578 * vec-perm-indices.h: New file.
579 * vec-perm-indices.c: Likewise.
580 * target.h (vec_perm_indices): Replace with a forward class
582 (auto_vec_perm_indices): Move to vec-perm-indices.h.
583 * optabs.h: Include vec-perm-indices.h.
584 (expand_vec_perm): Delete.
585 (selector_fits_mode_p, expand_vec_perm_var): Declare.
586 (expand_vec_perm_const): Declare.
587 * target.def (vec_perm_const_ok): Replace with...
588 (vec_perm_const): ...this new hook.
589 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
590 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
591 * doc/tm.texi: Regenerate.
592 * optabs.def (vec_perm_const): Delete.
593 * doc/md.texi (vec_perm_const): Likewise.
594 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
595 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
596 expand_vec_perm for constant permutation vectors. Assert that
597 the mode of variable permutation vectors is the integer equivalent
598 of the mode that is being permuted.
599 * optabs-query.h (selector_fits_mode_p): Declare.
600 * optabs-query.c: Include vec-perm-indices.h.
601 (selector_fits_mode_p): New function.
602 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
603 is defined, instead of checking whether the vec_perm_const_optab
604 exists. Use targetm.vectorize.vec_perm_const instead of
605 targetm.vectorize.vec_perm_const_ok. Check whether the indices
606 fit in the vector mode before using a variable permute.
607 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
608 vec_perm_indices instead of an rtx.
609 (expand_vec_perm): Replace with...
610 (expand_vec_perm_const): ...this new function. Take the selector
611 as a vec_perm_indices rather than an rtx. Also take the mode of
612 the selector. Update call to shift_amt_for_vec_perm_mask.
613 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
614 Use vec_perm_indices::new_expanded_vector to expand the original
615 selector into bytes. Check whether the indices fit in the vector
616 mode before using a variable permute.
617 (expand_vec_perm_var): Make global.
618 (expand_mult_highpart): Use expand_vec_perm_const.
619 * fold-const.c: Includes vec-perm-indices.h.
620 * tree-ssa-forwprop.c: Likewise.
621 * tree-vect-data-refs.c: Likewise.
622 * tree-vect-generic.c: Likewise.
623 * tree-vect-loop.c: Likewise.
624 * tree-vect-slp.c: Likewise.
625 * tree-vect-stmts.c: Likewise.
626 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
628 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
629 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
630 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
631 (aarch64_vectorize_vec_perm_const): ...this new function.
632 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
633 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
634 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
635 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
636 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
637 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
638 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
640 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
641 check for NEON modes.
642 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
643 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
644 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
645 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
647 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
648 the old VEC_PERM_CONST conditions.
649 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
650 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
651 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
652 (ia64_vectorize_vec_perm_const_ok): Merge into...
653 (ia64_vectorize_vec_perm_const): ...this new function.
654 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
655 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
656 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
657 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
658 * config/mips/mips.c (mips_expand_vec_perm_const)
659 (mips_vectorize_vec_perm_const_ok): Merge into...
660 (mips_vectorize_vec_perm_const): ...this new function.
661 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
662 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
663 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
664 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
665 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
666 (rs6000_expand_vec_perm_const): Delete.
667 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
669 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
670 (altivec_expand_vec_perm_const_le): Take each operand individually.
671 Operate on constant selectors rather than rtxes.
672 (altivec_expand_vec_perm_const): Likewise. Update call to
673 altivec_expand_vec_perm_const_le.
674 (rs6000_expand_vec_perm_const): Delete.
675 (rs6000_vectorize_vec_perm_const_ok): Delete.
676 (rs6000_vectorize_vec_perm_const): New function.
677 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
678 an element count and rtx array.
679 (rs6000_expand_extract_even): Update call accordingly.
680 (rs6000_expand_interleave): Likewise.
681 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
682 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
683 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
684 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
685 (rs6000_expand_vec_perm_const): Delete.
686 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
687 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
688 (altivec_expand_vec_perm_const_le): Take each operand individually.
689 Operate on constant selectors rather than rtxes.
690 (altivec_expand_vec_perm_const): Likewise. Update call to
691 altivec_expand_vec_perm_const_le.
692 (rs6000_expand_vec_perm_const): Delete.
693 (rs6000_vectorize_vec_perm_const_ok): Delete.
694 (rs6000_vectorize_vec_perm_const): New function. Remove stray
695 reference to the SPE evmerge intructions.
696 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
697 an element count and rtx array.
698 (rs6000_expand_extract_even): Update call accordingly.
699 (rs6000_expand_interleave): Likewise.
700 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
701 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
703 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
705 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
707 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
708 vector mode and that that mode matches the mode of the data
710 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
711 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
712 directly using expand_vec_perm_1 when forcing selectors into
714 (expand_vec_perm_var): New function, split out from expand_vec_perm.
716 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
718 * optabs-query.h (can_vec_perm_p): Delete.
719 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
720 * optabs-query.c (can_vec_perm_p): Split into...
721 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
722 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
723 particular selector is valid.
724 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
725 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
726 (vect_grouped_load_supported): Likewise.
727 (vect_shift_permute_load_chain): Likewise.
728 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
729 (vect_transform_slp_perm_load): Likewise.
730 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
731 (vectorizable_bswap): Likewise.
732 (vect_gen_perm_mask_checked): Likewise.
733 * fold-const.c (fold_ternary_loc): Likewise. Don't take
734 implementations of variable permutation vectors into account
735 when deciding which selector to use.
736 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
737 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
738 with a false third argument.
739 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
740 to test whether the constant selector is valid and can_vec_perm_var_p
741 to test whether a variable selector is valid.
743 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
745 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
746 * optabs-query.c (can_vec_perm_p): Likewise.
747 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
748 instead of vec_perm_indices.
749 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
750 (vect_gen_perm_mask_checked): Likewise,
751 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
752 (vect_gen_perm_mask_checked): Likewise,
754 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
756 * optabs-query.h (qimode_for_vec_perm): Declare.
757 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
758 (qimode_for_vec_perm): ...this new function.
759 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
761 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
763 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
764 does not have a conditional at the top.
766 2018-01-02 Richard Biener <rguenther@suse.de>
768 * ipa-inline.c (big_speedup_p): Fix expression.
770 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
773 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
776 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
780 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
781 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
782 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
783 cond_taken_branch_cost 3->4.
785 2018-01-01 Jakub Jelinek <jakub@redhat.com>
787 PR tree-optimization/83581
788 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
789 TODO_cleanup_cfg if any changes have been made.
792 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
793 convert_modes if target mode has the right side, but different mode
797 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
798 last argument when extracting from CONCAT. If either from_real or
799 from_imag is NULL, use expansion through memory. If result is not
800 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
801 the parts directly to inner mode, if even that fails, use expansion
805 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
806 check for bswap in mode rather than HImode and use that in expand_unop
809 2017-12-31 Jakub Jelinek <jakub@redhat.com>
812 * config/i386/i386.c (ix86_attribute_table): Remove excess
813 initializer for "shared" attribute.
815 2017-12-30 Tom de Vries <tom@codesourcery.com>
818 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
820 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
821 offload_funcs entries that no longer have a corresponding cgraph_node.
822 Mark the remaining ones as DECL_PRESERVE_P.
823 (output_lto): Call prune_offload_funcs.
825 2017-12-30 Jakub Jelinek <jakub@redhat.com>
827 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
828 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
829 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
830 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
831 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
832 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
833 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
834 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
835 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
836 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
837 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
838 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
839 vaesenc_<mode>, vpclmulqdq_<mode>,
840 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
842 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
844 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
845 _Float<N> and _Float<N>X variants for rounding built-in
847 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
848 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
849 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
850 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
851 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
852 * builtins.c (mathfn_built_in_2): Likewise.
853 * internal-fn.def (CEIL): Likewise.
855 (NEARBYINT): Likewise.
859 * convert.c (convert_to_integer_1): Likewise.
860 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
861 (integer_valued_real_call_p): Likewise.
862 * fold-const-call.c (fold_const_call_ss): Likewise.
863 * gencfn-macros.c (print_case_cfn): Change CFN and operator
864 printers to take a const char * suffix instead of a bool.
865 (print_define_operator_list): Likewise.
866 (fltall_suffixes): New list of suffixes, that include the
867 traditional suffixes as well as all of the _Float<N> and
869 (main): For _Float<N> and _Float<N>X functions, emit both
870 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
871 has the _Float<N> and _Float<N>X case names or operators. The
872 <name>_ALL variant has both the traditional and the
873 _Float<N>/_Float<N>X case names or operators.
874 * match.pd (COPYSIGN optimizations): Provide optimizations for
875 _Float<N> and _Float<N>X types where possible.
876 (MIN/MAX optimizations): Likewise.
877 (sqrt optimizations): Likewise.
878 (rounding optimizations): Likewise.
880 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
882 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
883 gen_int_mode rather than GEN_INT.
884 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
886 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
888 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
889 instead of CONSTANT_P.
890 (gen_vec_series): Likewise.
891 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
893 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
895 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
898 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
900 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
901 (aarch64_ld1x2<VDC:mode>): Likewise.
902 (aarch64_simd_ld1<mode>_x2): Likewise.
903 (aarch64_simd_ld1<mode>_x2): Likewise.
904 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
905 (vld1_s8_x2): Likewise.
906 (vld1_u16_x2): Likewise.
907 (vld1_s16_x2): Likewise.
908 (vld1_u32_x2): Likewise.
909 (vld1_s32_x2): Likewise.
910 (vld1_u64_x2): Likewise.
911 (vld1_s64_x2): Likewise.
912 (vld1_f16_x2): Likewise.
913 (vld1_f32_x2): Likewise.
914 (vld1_f64_x2): Likewise.
915 (vld1_p8_x2): Likewise.
916 (vld1_p16_x2): Likewise.
917 (vld1_p64_x2): Likewise.
918 (vld1q_u8_x2): Likewise.
919 (vld1q_s8_x2): Likewise.
920 (vld1q_u16_x2): Likewise.
921 (vld1q_s16_x2): Likewise.
922 (vld1q_u32_x2): Likewise.
923 (vld1q_s32_x2): Likewise.
924 (vld1q_u64_x2): Likewise.
925 (vld1q_s64_x2): Likewise.
926 (vld1q_f16_x2): Likewise.
927 (vld1q_f32_x2): Likewise.
928 (vld1q_f64_x2): Likewise.
929 (vld1q_p8_x2): Likewise.
930 (vld1q_p16_x2): Likewise.
931 (vld1q_p64_x2): Likewise.
933 2017-12-27 Martin Liska <mliska@suse.cz>
935 PR tree-optimization/83552
936 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
937 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
939 2017-12-27 Tom de Vries <tom@codesourcery.com>
941 * config/nvptx/nvptx.c (nvptx_option_override): Disable
942 -gstatement-frontiers.
944 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
946 PR rtl-optimization/83513
947 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
948 before priority comparison.
950 2017-12-25 Jakub Jelinek <jakub@redhat.com>
953 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
954 ix86_isa_flags2 to ix86_isa_flags.
955 * config/i386/i386-c.c (ix86_target_macros_internal): Test
956 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
957 isa_flags rather than isa_flags2.
958 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
959 and -mavx512bitalg from isa2_opts to isa_opts.
960 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
961 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
962 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
964 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
965 OPTION_MASK_ISA_AVX512F ored with another option similarly to
966 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
967 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
968 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
969 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
970 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
971 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
972 builtins from bdesc_args2 to bdesc_args section.
973 (__builtin_ia32_compressstoreuqi512_mask,
974 __builtin_ia32_compressstoreuhi512_mask,
975 __builtin_ia32_compressstoreuqi256_mask,
976 __builtin_ia32_expandloadqi512_mask,
977 __builtin_ia32_expandloadqi512_maskz,
978 __builtin_ia32_expandloadhi512_mask,
979 __builtin_ia32_expandloadhi512_maskz,
980 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
981 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
982 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
983 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
984 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
985 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
986 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
987 __builtin_ia32_vpshldv_v32hi_maskz,
988 __builtin_ia32_vpopcountb_v64qi_mask,
989 __builtin_ia32_vpopcountw_v32hi_mask,
990 __builtin_ia32_vpshufbitqmb512_mask,
991 __builtin_ia32_vpshufbitqmb256_mask): Add
992 " | OPTION_MASK_ISA_AVX512BW".
993 (__builtin_ia32_expandloadqi256_mask,
994 __builtin_ia32_expandloadqi256_maskz,
995 __builtin_ia32_vpopcountb_v32qi_mask): Add
996 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
997 (__builtin_ia32_expandloadhi256_mask,
998 __builtin_ia32_expandloadhi256_maskz,
999 __builtin_ia32_expandloadqi128_mask,
1000 __builtin_ia32_expandloadqi128_maskz,
1001 __builtin_ia32_expandloadhi128_mask,
1002 __builtin_ia32_expandloadhi128_maskz,
1003 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
1004 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
1005 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
1006 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
1007 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
1008 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
1009 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
1010 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
1011 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
1012 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
1013 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
1014 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
1015 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
1016 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
1017 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
1018 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
1019 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
1020 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
1021 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
1022 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
1023 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
1024 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
1025 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
1026 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
1027 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
1028 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
1029 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
1030 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
1031 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
1032 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
1033 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
1034 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
1035 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
1036 __builtin_ia32_vpopcountw_v8hi_mask): Add
1037 " | OPTION_MASK_ISA_AVX512VL".
1038 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1039 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
1040 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
1041 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
1042 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
1043 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
1044 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
1045 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
1046 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
1047 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
1048 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
1049 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
1050 _mm_mask_bitshuffle_epi64_mask): Likewise.
1051 * common/config/i386/i386-common.c
1052 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
1053 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
1054 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1055 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
1056 OPTION_MASK_ISA_AVX512BITALG_UNSET.
1057 (OPTION_MASK_ISA2_AVX512F_UNSET,
1058 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
1059 (ix86_handle_option): For -mno-general-regs-only, clear from
1060 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
1061 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
1062 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
1063 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
1064 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
1065 for moving from ix86_isa_flags2 to ix86_isa_flags.
1067 2017-12-23 Jakub Jelinek <jakub@redhat.com>
1070 * fold-const.c (struct contains_label_data): New type.
1071 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
1072 inside of a SWITCH_BODY seen during the walk.
1073 (contains_label_p): Use walk_tree instead of
1074 walk_tree_without_duplicates, prepare data for contains_label_1 and
1077 2017-12-22 Jakub Jelinek <jakub@redhat.com>
1080 * tree-iterator.c (alloc_stmt_list): Start with cleared
1081 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
1085 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
1086 enable avx512f explicitly in #pragma GCC target.
1087 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
1088 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
1089 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
1090 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
1091 __builtin_ia32_vpdpbusds_v8si_mask,
1092 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
1093 __builtin_ia32_vpdpbusds_v4si_mask,
1094 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
1095 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
1096 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
1097 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
1098 __builtin_ia32_vpdpwssds_v8si_mask,
1099 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
1100 __builtin_ia32_vpdpwssds_v4si_mask,
1101 __builtin_ia32_vpdpwssds_v4si_maskz): Use
1102 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
1103 just OPTION_MASK_ISA_AVX512VNNI.
1105 2017-12-22 Martin Jambor <mjambor@suse.cz>
1108 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
1111 2017-12-22 Julia Koval <julia.koval@intel.com>
1112 Sebastian Peryt <sebastian.peryt@intel.com>
1114 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
1115 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
1116 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
1117 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
1118 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
1119 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
1120 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
1121 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
1122 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
1123 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
1124 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
1125 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
1126 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
1127 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
1128 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
1129 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
1130 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
1131 _mm256_maskz_popcnt_epi64): New intrinsics.
1132 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
1133 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
1134 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
1135 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
1136 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
1137 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
1138 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
1139 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
1140 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
1141 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
1142 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
1143 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
1144 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
1145 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
1146 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
1147 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
1148 __builtin_ia32_vpshufbitqmb256_mask,
1149 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
1150 * config/i386/i386-c.c (__AVX512BITALG__): New.
1151 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
1152 (ix86_valid_target_attribute_inner_p): Ditto.
1153 (ix86_expand_args_builtin): Handle new types.
1154 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
1155 * config/i386/i386.opt: Add -mavx512bitalg.
1156 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
1157 avx512bitalgintrin.h.
1158 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
1159 (vpopcount<mode><mask_name>): Add more types.
1160 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
1161 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
1163 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1165 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
1166 Or in OPTION_MASK_ISA_AVX512F_SET.
1167 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1168 OPTION_MASK_ISA_AVX512VNNI_UNSET.
1169 (ix86_handle_option): Adjust for
1170 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
1171 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
1173 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
1174 isa_flag instead of isa_flag2.
1175 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
1176 isa_opts2 to isa_opts.
1177 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
1180 2017-12-22 Mike Stump <mikestump@comcast.net>
1181 Eric Botcazou <ebotcazou@adacore.com>
1183 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
1185 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1188 * config/i386/i386.c (ix86_function_arg_boundary): Return
1189 PARM_BOUNDARY for TYPE_EMPTY_P types.
1192 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
1193 if navail is >= dir.len.
1195 2017-12-21 Steve Ellcey <sellcey@cavium.com>
1197 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
1200 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1202 PR rtl-optimization/80747
1203 PR rtl-optimization/83512
1204 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
1205 succ edge from ENTRY, copy partition from e->dest to the newly
1207 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
1208 ENTRY, use BB_PARTITION of its successor block as current_partition.
1209 Don't copy partition when splitting succ edge from ENTRY.
1211 PR tree-optimization/83523
1212 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
1213 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
1214 (convert_mult_to_fma): Likewise.
1216 PR tree-optimization/83521
1217 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
1218 gimple_build_assign without code on result of
1219 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
1220 a VIEW_CONVERT_EXPR.
1222 2017-12-21 Andrew Pinski <apinski@cavium.com>
1223 Steve Ellcey <sellcey@cavium.com>
1225 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
1226 multi-arch for ilp32.
1228 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
1231 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
1232 constraints to operand 2.
1233 (*ashl<mode>3_mask_1): Ditto.
1234 (*<shift_insn><mode>3_mask): Ditto.
1235 (*<shift_insn><mode>3_mask_1): Ditto.
1236 (*<rotate_insn><mode>3_mask): Ditto.
1237 (*<rotate_insn><mode>3_mask_1): Ditto.
1239 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
1241 * reorg.c (make_return_insns): Reemit each insn with its own location.
1243 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1245 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
1246 generation for cases where splatting a value is not useful.
1247 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1248 across a vec_duplicate and a paradoxical subreg forming a vector
1249 mode to a vec_concat.
1251 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1253 * combine.c (simplify_set): Do not transform subregs to zero_extends
1254 if the destination is not a scalar int mode.
1256 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
1259 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
1260 shared zero if the input has overflowed.
1262 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1264 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
1265 support for Cortex-A55 and Cortex-A75.
1267 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1269 * common/config/arm/arm-common.c (compare_opt_names): Add function
1270 comment. Use strcmp instead of manual loop.
1272 2017-12-21 Martin Liska <mliska@suse.cz>
1274 PR gcov-profile/83509
1275 * gcov-dump.c (dump_gcov_file): Do not read info about
1276 support_unexecuted_blocks for gcda files.
1278 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1280 PR rtl-optimization/82973
1281 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
1282 (valid_for_const_vector_p): ... this.
1283 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
1284 (valid_for_const_vector_p): ... this. Adjust function comment.
1285 (gen_vec_duplicate): Adjust caller.
1286 * optabs.c (expand_vector_broadcast): Likewise.
1287 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
1288 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
1290 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
1291 instead of GET_CODE == CONST_FIXED.
1292 (simplify_subreg): Use CONST_FIXED_P macro instead of
1293 GET_CODE == CONST_FIXED.
1296 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
1297 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
1298 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
1299 (ix86_option_override_internal): Adjust for
1300 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
1301 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
1302 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
1303 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
1304 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
1305 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
1306 def_builtin2 for CET builtins.
1307 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
1308 up formatting in IX86_BUILTIN_RDPID code.
1309 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
1310 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
1311 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
1312 ix86_isa_flags2 to ix86_isa_flags.
1313 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
1315 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
1316 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
1317 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
1319 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
1320 Or in OPTION_MASK_ISA_AVX512F_SET.
1321 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1322 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
1323 (ix86_handle_option): Adjust for
1324 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
1325 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
1327 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1328 Alan Hayward <alan.hayward@arm.com>
1329 David Sherwood <david.sherwood@arm.com>
1331 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
1332 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1333 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
1334 Track polynomial offsets.
1336 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1337 Alan Hayward <alan.hayward@arm.com>
1338 David Sherwood <david.sherwood@arm.com>
1340 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1341 Treat drb->init as a poly_int. Fail if its misalignment wrt
1342 vector_alignment isn't known.
1344 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1345 Alan Hayward <alan.hayward@arm.com>
1346 David Sherwood <david.sherwood@arm.com>
1348 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
1350 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
1351 (vect_loop_versioning): Take the loop versioning threshold as a
1353 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
1354 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
1355 versioning_threshold.
1356 (vect_analyze_loop_2): Compute the loop versioning threshold
1357 whenever loop versioning is needed, and store it in the new
1358 field rather than combining it with the cost model threshold.
1359 (vect_transform_loop): Update call to vect_loop_versioning.
1360 Try to combine the loop versioning and cost thresholds here.
1362 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1363 Alan Hayward <alan.hayward@arm.com>
1364 David Sherwood <david.sherwood@arm.com>
1366 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
1367 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
1368 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
1369 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
1370 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
1371 (group_compare_offset): Update accordingly.
1372 (split_small_address_groups_p): Likewise.
1373 (record_use): Take addr_offset as a poly_uint64 rather than
1374 an unsigned HOST_WIDE_INT.
1375 (strip_offset): Return the offset as a poly_uint64 rather than
1376 an unsigned HOST_WIDE_INT.
1377 (record_group_use, split_address_groups): Track polynomial offsets.
1378 (add_iv_candidate_for_use): Likewise.
1379 (addr_offset_valid_p): Take the offset as a poly_int64 rather
1380 than a HOST_WIDE_INT.
1381 (strip_offset_1): Return the offset as a poly_int64 rather than
1384 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1385 Alan Hayward <alan.hayward@arm.com>
1386 David Sherwood <david.sherwood@arm.com>
1388 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
1389 rather than a HOST_WIDE_INT.
1390 * tree.c (get_binfo_at_offset): Likewise.
1392 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1393 Alan Hayward <alan.hayward@arm.com>
1394 David Sherwood <david.sherwood@arm.com>
1396 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
1397 rather than a HOST_WIDE_INT.
1398 * tree-sra.c (build_ref_for_offset): Likewise.
1400 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1401 Alan Hayward <alan.hayward@arm.com>
1402 David Sherwood <david.sherwood@arm.com>
1404 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
1406 * tree.c (mem_ref_offset): Likewise.
1407 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
1408 * builtins.c (get_object_alignment_2): Likewise.
1409 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
1410 * gimple-fold.c (get_base_constructor): Likewise.
1411 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
1412 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1414 * ipa-polymorphic-call.c
1415 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1416 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
1417 (get_ancestor_addr_info): Likewise.
1418 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
1419 * match.pd: Likewise.
1420 * tree-data-ref.c (dr_analyze_innermost): Likewise.
1421 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1422 * tree-eh.c (tree_could_trap_p): Likewise.
1423 * tree-object-size.c (addr_object_size): Likewise.
1424 * tree-ssa-address.c (copy_ref_info): Likewise.
1425 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
1426 (indirect_refs_may_alias_p): Likewise.
1427 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1428 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
1429 (non_rewritable_mem_ref_base): Likewise.
1430 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1431 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1432 * varasm.c (decode_addr_const): Likewise.
1434 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1435 Alan Hayward <alan.hayward@arm.com>
1436 David Sherwood <david.sherwood@arm.com>
1438 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
1439 offsets for MEM_REFs.
1441 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1442 Alan Hayward <alan.hayward@arm.com>
1443 David Sherwood <david.sherwood@arm.com>
1445 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
1446 as a poly_uint64 rather than an unsigned int.
1447 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
1449 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1450 Alan Hayward <alan.hayward@arm.com>
1451 David Sherwood <david.sherwood@arm.com>
1453 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
1455 (decode_addr_const): Update accordingly.
1457 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1458 Alan Hayward <alan.hayward@arm.com>
1459 David Sherwood <david.sherwood@arm.com>
1461 * tree.h (bit_field_size, bit_field_offset): New functions.
1462 * hsa-gen.c (gen_hsa_addr): Use them.
1463 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
1464 (simplify_vector_constructor): Likewise.
1465 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1466 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
1467 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
1468 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
1471 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1472 Alan Hayward <alan.hayward@arm.com>
1473 David Sherwood <david.sherwood@arm.com>
1475 * expr.h (emit_group_load, emit_group_load_into_temps)
1476 (emit_group_store): Take the size as a poly_int64 rather than an int.
1477 * expr.c (emit_group_load_1, emit_group_load): Likewise.
1478 (emit_group_load_into_temp, emit_group_store): Likewise.
1480 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1481 Alan Hayward <alan.hayward@arm.com>
1482 David Sherwood <david.sherwood@arm.com>
1484 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
1485 an unsigned int to a poly_uint64.
1486 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
1487 sizes as poly_uint64s rather than unsigned ints.
1488 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
1491 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1492 Alan Hayward <alan.hayward@arm.com>
1493 David Sherwood <david.sherwood@arm.com>
1495 * lra-constraints.c (emit_inc): Change inc_amount from an int
1498 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1499 Alan Hayward <alan.hayward@arm.com>
1500 David Sherwood <david.sherwood@arm.com>
1502 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
1504 (add_stack_var, stack_var_cmp, partition_stack_vars)
1505 (dump_stack_var_partition): Update accordingly.
1506 (alloc_stack_frame_space): Take the size as a poly_int64 rather
1507 than a HOST_WIDE_INT.
1508 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
1509 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
1510 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
1511 rather than a HOST_WIDE_INT.
1513 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1514 Alan Hayward <alan.hayward@arm.com>
1515 David Sherwood <david.sherwood@arm.com>
1517 * target.def (return_pops_args): Treat both the input and output
1518 sizes as poly_int64s rather than HOST_WIDE_INTS.
1519 * targhooks.h (default_return_pops_args): Update accordingly.
1520 * targhooks.c (default_return_pops_args): Likewise.
1521 * doc/tm.texi: Regenerate.
1522 * emit-rtl.h (incoming_args): Change pops_args, size and
1523 outgoing_args_size from int to poly_int64_pod.
1524 * function.h (expr_status): Change x_pending_stack_adjust and
1525 x_stack_pointer_delta from int to poly_int64.
1526 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
1527 (ARGS_SIZE_RTX): Update accordingly.
1528 * calls.c (highest_outgoing_arg_in_use): Change from int to
1530 (stack_usage_watermark, stored_args_watermark): New variables.
1531 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
1532 (emit_call_1): Change the stack_size and rounded_stack_size
1533 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
1535 (save_fixed_argument_area): Check stack_usage_watermark.
1536 (initialize_argument_information): Change old_pending_adj from
1537 a HOST_WIDE_INT * to a poly_int64_pod *.
1538 (compute_argument_block_size): Return the size as a poly_int64
1540 (finalize_must_preallocate): Track polynomial argument sizes.
1541 (compute_argument_addresses): Likewise.
1542 (internal_arg_pointer_based_exp): Track polynomial offsets.
1543 (mem_overlaps_already_clobbered_arg_p): Rename to...
1544 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
1545 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1546 Check stored_args_used_watermark.
1547 (load_register_parameters): Update accordingly.
1548 (check_sibcall_argument_overlap_1): Likewise.
1549 (combine_pending_stack_adjustment_and_call): Take the unadjusted
1550 args size as a poly_int64 rather than an int. Return a bool
1551 indicating whether the optimization was possible and return
1552 the new adjustment by reference.
1553 (check_sibcall_argument_overlap): Track polynomail argument sizes.
1554 Update stored_args_watermark.
1555 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
1556 (expand_call): Likewise. Maintain stack_usage_watermark and
1557 stored_args_watermark. Update calls to
1558 combine_pending_stack_adjustment_and_call.
1559 (emit_library_call_value_1): Handle polynomial argument sizes.
1560 Call stack_region_maybe_used_p and mark_stack_region_used.
1561 Maintain stack_usage_watermark.
1562 (store_one_arg): Likewise. Update call to
1563 mem_overlaps_already_clobbered_arg_p.
1564 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
1566 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
1567 * config/microblaze/microblaze.c (microblaze_function_prologue):
1569 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
1570 TARGET_RETURN_POPS_ARGS interface.
1571 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
1573 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
1574 * config/i386/i386.c (ix86_return_pops_args): Update for new
1575 TARGET_RETURN_POPS_ARGS interface.
1576 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
1577 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
1578 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
1579 TARGET_RETURN_POPS_ARGS interface.
1580 * config/vax/vax.c (vax_return_pops_args): Likewise.
1581 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
1582 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
1583 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
1584 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1585 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1586 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
1587 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1588 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1589 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
1590 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
1591 and x_stack_pointer_delta from int to poly_int64.
1592 * dojump.c (do_pending_stack_adjust): Update accordingly.
1593 * explow.c (allocate_dynamic_stack_space): Handle polynomial
1594 stack_pointer_deltas.
1595 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
1596 (pad_to_arg_alignment): Track polynomial offsets.
1597 (assign_parm_find_stack_rtl): Likewise.
1598 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
1599 * toplev.c (output_stack_usage): Update reference to
1600 current_function_pushed_stack_size.
1602 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1603 Alan Hayward <alan.hayward@arm.com>
1604 David Sherwood <david.sherwood@arm.com>
1606 * function.c (in_arg_offset, var_offset, dynamic_offset)
1607 (out_arg_offset, cfa_offset): Change from int to poly_int64.
1608 (instantiate_new_reg): Return the new offset as a poly_int64_pod
1609 rather than a HOST_WIDE_INT.
1610 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
1611 (instantiate_virtual_regs_in_insn): Likewise.
1613 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1614 Alan Hayward <alan.hayward@arm.com>
1615 David Sherwood <david.sherwood@arm.com>
1617 * rtl.h (get_args_size, add_args_size_note): New functions.
1618 (find_args_size_adjust): Return a poly_int64 rather than a
1620 (fixup_args_size_notes): Likewise. Make the same change to the
1621 end_args_size parameter.
1622 * rtlanal.c (get_args_size, add_args_size_note): New functions.
1623 * builtins.c (expand_builtin_trap): Use add_args_size_note.
1624 * calls.c (emit_call_1): Likewise.
1625 * explow.c (adjust_stack_1): Likewise.
1626 * cfgcleanup.c (old_insns_match_p): Update use of
1627 find_args_size_adjust.
1628 * combine.c (distribute_notes): Track polynomial arg sizes.
1629 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
1630 end_true_args_size, beg_delay_args_size and end_delay_args_size
1631 from HOST_WIDE_INT to poly_int64.
1632 (add_cfi_args_size): Take the args_size as a poly_int64 rather
1633 than a HOST_WIDE_INT.
1634 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
1635 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
1636 polynomial arg sizes.
1637 * emit-rtl.c (try_split): Use get_args_size.
1638 * recog.c (peep2_attempt): Likewise.
1639 * reload1.c (reload_as_needed): Likewise.
1640 * expr.c (find_args_size_adjust): Return the adjustment as a
1641 poly_int64 rather than a HOST_WIDE_INT.
1642 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
1643 to a poly_int64 and change the return type in the same way.
1644 (emit_single_push_insn): Track polynomial arg sizes.
1646 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1647 Alan Hayward <alan.hayward@arm.com>
1648 David Sherwood <david.sherwood@arm.com>
1650 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
1651 from HOST_WIDE_INT to poly_int64.
1652 * expr.c (push_block, emit_push_insn): Likewise.
1654 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1655 Alan Hayward <alan.hayward@arm.com>
1656 David Sherwood <david.sherwood@arm.com>
1658 * function.h (frame_space): Change start and length from HOST_WIDE_INT
1660 (get_frame_size): Return the size as a poly_int64 rather than a
1662 (frame_offset_overflow): Take the offset as a poly_int64 rather
1663 than a HOST_WIDE_INT.
1664 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1665 (assign_stack_temp): Likewise for the size.
1666 * function.c (get_frame_size): Return a poly_int64 rather than
1668 (frame_offset_overflow): Take the offset as a poly_int64 rather
1669 than a HOST_WIDE_INT.
1670 (try_fit_stack_local): Take the start, length and size as poly_int64s
1671 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
1672 rather than a HOST_WIDE_INT.
1673 (add_frame_space): Take the start and end as poly_int64s rather than
1675 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1676 (assign_stack_temp): Likewise for the size.
1677 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
1679 (find_temp_slot_from_address): Handle polynomial offsets.
1680 (combine_temp_slots): Likewise.
1681 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
1683 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
1684 poly_int64 rather than a HOST_WIDE_INT.
1685 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
1686 than a HOST_WIDE_INT.
1687 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
1688 polynomial frame offsets.
1689 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
1690 as a poly_int64 rather than an int.
1691 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
1692 * config/v850/v850-protos.h (compute_frame_size): Likewise.
1693 * config/v850/v850.c (compute_frame_size): Likewise.
1694 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
1695 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
1696 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
1697 * config/pa/pa.c (pa_compute_frame_size): Likewise.
1698 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
1699 rather than a HOST_WIDE_INT.
1700 * explow.c (get_dynamic_stack_base): Likewise.
1701 * final.c (final_start_function): Use the constant lower bound
1702 of the frame size for -Wframe-larger-than.
1703 * ira.c (do_reload): Adjust for new get_frame_size return type.
1704 * lra.c (lra): Likewise.
1705 * reload1.c (reload): Likewise.
1706 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
1707 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
1708 * rtlanal.c (get_initial_register_offset): Return the offset as
1709 a poly_int64 rather than a HOST_WIDE_INT.
1711 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1712 Alan Hayward <alan.hayward@arm.com>
1713 David Sherwood <david.sherwood@arm.com>
1715 * reload1.c (elim_table): Change initial_offset, offset and
1716 previous_offset from HOST_WIDE_INT to poly_int64_pod.
1717 (offsets_at): Change the target array's element type from
1718 HOST_WIDE_INT to poly_int64_pod.
1719 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
1720 (elimination_costs_in_insn, update_eliminable_offsets)
1721 (verify_initial_elim_offsets, set_offsets_for_label)
1722 (init_eliminable_invariants): Update after above changes.
1724 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1725 Alan Hayward <alan.hayward@arm.com>
1726 David Sherwood <david.sherwood@arm.com>
1728 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
1729 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
1730 (decomposition): Change start and end from HOST_WIDE_INT
1732 (decompose, immune_p): Update accordingly.
1733 (find_inc_amount): Return a poly_int64 rather than an int.
1734 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
1737 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1738 Alan Hayward <alan.hayward@arm.com>
1739 David Sherwood <david.sherwood@arm.com>
1741 * tree.h (get_inner_reference): Return the bitsize and bitpos
1742 as poly_int64_pods rather than HOST_WIDE_INT.
1743 * fold-const.h (ptr_difference_const): Return the pointer difference
1744 as a poly_int64_pod rather than a HOST_WIDE_INT.
1745 * expr.c (get_inner_reference): Return the bitsize and bitpos
1746 as poly_int64_pods rather than HOST_WIDE_INT.
1747 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
1749 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
1750 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
1751 (optimize_bit_field_compare): Update call to get_inner_reference.
1752 (decode_field_reference): Likewise.
1753 (fold_unary_loc): Track polynomial offsets and sizes.
1754 (split_address_to_core_and_offset): Return the bitpos as a
1755 poly_int64_pod rather than a HOST_WIDE_INT.
1756 (ptr_difference_const): Likewise for the pointer difference.
1757 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
1758 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
1759 * dbxout.c (dbxout_expand_expr): Likewise.
1760 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
1761 (loc_list_from_tree_1, fortran_common): Likewise.
1762 * gimple-laddress.c (pass_laddress::execute): Likewise.
1763 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
1764 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
1765 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1766 * tree-affine.c (tree_to_aff_combination): Likewise.
1767 (get_inner_reference_aff): Likewise.
1768 * tree-data-ref.c (split_constant_offset_1): Likewise.
1769 (dr_analyze_innermost): Likewise.
1770 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
1771 * tree-sra.c (ipa_sra_check_caller): Likewise.
1772 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1773 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
1774 (instrument_bool_enum_load, instrument_object_size): Likewise.
1775 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
1776 to get_inner_reference.
1777 * hsa-gen.c (gen_hsa_addr): Likewise.
1778 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
1779 * tsan.c (instrument_expr): Likewise.
1780 * match.pd: Update call to ptr_difference_const.
1782 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1783 Alan Hayward <alan.hayward@arm.com>
1784 David Sherwood <david.sherwood@arm.com>
1786 * fold-const.c (fold_comparison): Track sizes and offsets as
1787 poly_int64s rather than HOST_WIDE_INTs when folding address
1790 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1791 Alan Hayward <alan.hayward@arm.com>
1792 David Sherwood <david.sherwood@arm.com>
1794 * expr.h (get_bit_range): Return the bitstart and bitend as
1795 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1796 as a poly_int64 rather than a HOST_WIDE_INT.
1797 * expr.c (get_bit_range): Likewise.
1798 (expand_assignment): Update call accordingly.
1799 * fold-const.c (optimize_bit_field_compare): Likewise.
1801 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1802 Alan Hayward <alan.hayward@arm.com>
1803 David Sherwood <david.sherwood@arm.com>
1805 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1806 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1807 as an optional final parameter.
1808 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1809 (aarch64_legitimate_address_p): Likewise.
1810 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1811 instead of an rtx code.
1812 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1813 (aarch64_legitimate_address_hook_p): Likewise.
1814 (aarch64_print_ldpstp_address): Likewise.
1815 (aarch64_print_operand_address): Likewise.
1816 (aarch64_address_cost): Likewise.
1817 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1818 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1819 (aarch64_mem_pair_lanes_operand): Likewise.
1821 2017-12-20 Richard Biener <rguenther@suse.de>
1823 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1824 update_stmt_if_modified.
1826 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1828 PR tree-optimization/83491
1829 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1830 before walking uses. Improve coding style and comments.
1832 2017-12-20 Tom de Vries <tom@codesourcery.com>
1834 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1836 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1839 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1841 (pass_ipa_free_fn_summary): Add small_p private data member,
1842 initialize to false in the ctor.
1843 (pass_ipa_free_fn_summary::clone,
1844 pass_ipa_free_fn_summary::set_pass_param,
1845 pass_ipa_free_fn_summary::gate): New methods.
1846 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1847 | TODO_dump_symtab if small_p.
1848 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1849 entry and add another instance of the pass with false parm after
1851 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1852 ipa_free_fn_summary here.
1854 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1856 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1858 2017-12-20 Martin Sebor <msebor@redhat.com>
1861 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1862 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1863 (expand_builtin_strncmp): Same.
1865 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1868 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1869 allowed debug stmts before labels.
1870 (expand_gimple_basic_block): Likewise.
1871 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1872 * gimple-iterator.h (gsi_after_labels): Likewise.
1873 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1874 rename reused variable, and simplify using gsi_move_before.
1875 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1876 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1877 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1878 (gimple_verify_flow_info, gimple_block_label): Likewise.
1879 (make_blocks): Move debug markers after adjacent labels.
1880 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1881 allowed debug insns outside blocks.
1882 * df-scan.c (df_insn_delete): Likewise.
1883 * lra-constraints.c (update_ebb_live_info): Likewise.
1884 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1885 (vt_initialize, delete_vta_debug_insns): Likewise.
1886 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1888 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1889 Alan Hayward <alan.hayward@arm.com>
1890 David Sherwood <david.sherwood@arm.com>
1892 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1893 (round_up_to_byte_boundary): Likewise.
1894 * expr.h (get_bit_range): Add temporary shim.
1895 * gimple-ssa-store-merging.c (store_operand_info): Change the
1896 bitsize, bitpos, bitregion_start and bitregion_end fields from
1897 unsigned HOST_WIDE_INT to poly_uint64.
1898 (merged_store_group): Likewise load_align_base.
1899 (compatible_load_p, compatible_load_p): Update accordingly.
1900 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1901 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1902 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1903 bitregion_start and bitregion_end as poly_uint64s rather than
1904 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1905 (handled_load): Take the bitsize, bitpos,
1906 bitregion_start and bitregion_end as poly_uint64s rather than
1907 unsigned HOST_WIDE_INTs.
1908 (pass_store_merging::process_store): Update call to
1909 mem_valid_for_store_merging.
1911 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1912 Alan Hayward <alan.hayward@arm.com>
1913 David Sherwood <david.sherwood@arm.com>
1915 * builtins.c (get_object_alignment_2): Track polynomial offsets
1916 and sizes. Update the alignment handling.
1918 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1919 Alan Hayward <alan.hayward@arm.com>
1920 David Sherwood <david.sherwood@arm.com>
1922 * tree.h (get_inner_reference): Add a version that returns the
1923 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1924 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1925 the case in which bitpos is not associated with the first byte.
1927 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1928 Alan Hayward <alan.hayward@arm.com>
1929 David Sherwood <david.sherwood@arm.com>
1931 * tree-affine.h (get_inner_reference_aff): Return the size as a
1933 * tree-affine.c (get_inner_reference_aff): Likewise.
1934 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1935 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1937 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1938 Alan Hayward <alan.hayward@arm.com>
1939 David Sherwood <david.sherwood@arm.com>
1941 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1942 HOST_WIDE_INT rather than a poly_int64.
1944 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1945 Alan Hayward <alan.hayward@arm.com>
1946 David Sherwood <david.sherwood@arm.com>
1948 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1949 HOST_WIDE_INT to poly_int64_pod.
1950 (perform_symbolic_merge): Update accordingly.
1951 (bswap_replace): Likewise.
1953 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1954 Alan Hayward <alan.hayward@arm.com>
1955 David Sherwood <david.sherwood@arm.com>
1957 * tree-affine.h (aff_tree::offset): Change from widest_int
1959 (wide_int_ext_for_comb): Delete.
1960 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1961 constants as poly_widest_int rather than widest_int.
1962 (aff_combination_constant_multiple_p): Return the multiplier
1963 as a poly_widest_int.
1964 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1966 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1967 version static and add an overload for poly_widest_int.
1968 (aff_combination_const, aff_combination_add_cst)
1969 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1970 the constants as poly_widest_int rather than widest_int.
1971 (tree_to_aff_combination): Generalize INTEGER_CST case to
1973 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1974 (aff_combination_add_product, aff_combination_mult): Handle
1976 (aff_combination_constant_multiple_p): Return the multiplier
1977 as a poly_widest_int.
1978 * tree-predcom.c (determine_offset): Return the offset as a
1980 (split_data_refs_to_components, suitable_component_p): Update
1982 (valid_initializer_p): Update call to
1983 aff_combination_constant_multiple_p.
1984 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1985 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1986 as a poly_int64 rather than a HOST_WIDE_INT.
1987 (get_address_cost): Handle polynomial offsets.
1988 (iv_elimination_compare_lt): Likewise.
1989 (rewrite_use_nonlinear_expr): Likewise.
1991 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1992 Alan Hayward <alan.hayward@arm.com>
1993 David Sherwood <david.sherwood@arm.com>
1995 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1996 as a poly_int64_pod rather than a HOST_WIDE_INT.
1997 (get_addr_base_and_unit_offset): Likewise.
1998 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1999 (get_addr_base_and_unit_offset): Likewise.
2000 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
2001 to poly_int64 in example.
2002 * fold-const.c (fold_binary_loc): Update call to
2003 get_addr_base_and_unit_offset.
2004 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
2005 (maybe_canonicalize_mem_ref_addr): Likewise.
2006 (gimple_fold_stmt_to_constant_1): Likewise.
2007 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2009 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
2010 * match.pd: Likewise.
2011 * omp-low.c (lower_omp_target): Likewise.
2012 * tree-sra.c (build_ref_for_offset): Likewise.
2013 (build_debug_ref_for_model): Likewise.
2014 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
2015 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
2016 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
2017 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
2018 (constant_pointer_difference): Likewise.
2019 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
2020 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
2021 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
2022 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
2023 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
2024 (set_ssa_val_to): Likewise.
2025 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
2026 (maybe_diag_stxncpy_trunc): Likewise.
2027 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
2028 * tree.c (build_simple_mem_ref_loc): Likewise.
2029 (array_at_struct_end_p): Likewise.
2031 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2032 Alan Hayward <alan.hayward@arm.com>
2033 David Sherwood <david.sherwood@arm.com>
2035 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
2036 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
2037 (get_ref_base_and_extent_hwi): Declare.
2038 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
2039 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
2040 (get_ref_base_and_extent_hwi): New function.
2041 * cfgexpand.c (expand_debug_expr): Update call to
2042 get_ref_base_and_extent.
2043 * dwarf2out.c (add_var_loc_to_decl): Likewise.
2044 * gimple-fold.c (get_base_constructor): Return the offset as a
2045 poly_int64_pod rather than a HOST_WIDE_INT.
2046 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
2047 * ipa-polymorphic-call.c
2048 (ipa_polymorphic_call_context::set_by_invariant)
2049 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
2050 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
2051 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
2052 rather than get_ref_base_and_extent.
2053 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
2054 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
2055 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
2057 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
2058 call to get_ref_base_and_extent.
2059 * tree-sra.c (create_access, get_access_for_expr): Likewise.
2060 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
2061 (stmt_kills_ref_p): Likewise.
2062 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2063 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
2065 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
2066 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
2067 when calling native_encode_expr.
2068 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
2069 call to get_ref_base_and_extent.
2070 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
2071 get_ref_base_and_extent.
2072 * var-tracking.c (track_expr_p): Likewise.
2074 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2075 Alan Hayward <alan.hayward@arm.com>
2076 David Sherwood <david.sherwood@arm.com>
2078 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
2079 HOST_WIDE_INT to poly_int64_pod.
2080 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
2081 polynomail parameter offsets.
2083 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2084 Alan Hayward <alan.hayward@arm.com>
2085 David Sherwood <david.sherwood@arm.com>
2087 * gengtype.c (main): Handle poly_int64_pod.
2088 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
2089 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
2090 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
2091 from HOST_WIDE_INT to poly_int64_pod.
2092 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
2093 (copy_cfa): New function.
2094 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
2095 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
2096 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
2097 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
2098 (initial_return_save): Treat offsets as poly_ints.
2099 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
2100 offset is nonconstant.
2101 (reg_save): Take the offset as a poly_int64. Fall back to
2102 DW_CFA_expression for nonconstant offsets.
2103 (queue_reg_save): Take the offset as a poly_int64.
2104 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
2106 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2107 Alan Hayward <alan.hayward@arm.com>
2108 David Sherwood <david.sherwood@arm.com>
2110 * rtl.h (operand_subword, operand_subword_force): Take the offset
2111 as a poly_uint64 an unsigned int.
2112 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
2114 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2115 Alan Hayward <alan.hayward@arm.com>
2116 David Sherwood <david.sherwood@arm.com>
2118 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
2119 'p' format code. Use INT_LIST rather than SUBREG as the example of
2120 a code with an XINT and an XEXP. Remove the implication that
2121 accessing an rtx field using XINT is expected to work.
2122 * rtl.def (SUBREG): Change format from "ei" to "ep".
2123 * rtl.h (rtunion::rt_subreg): New field.
2124 (XCSUBREG): New macro.
2125 (SUBREG_BYTE): Use it.
2126 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
2127 Update constructor accordingly.
2128 (subreg_shape::operator ==): Update accordingly.
2129 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
2130 than an unsigned int.
2131 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
2132 a poly_uint64 rather than an unsigned int.
2133 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
2134 than an unsigned int.
2135 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
2136 (subreg_size_highpart_offset): Return a poly_uint64 rather than
2137 an unsigned int. Take the sizes as poly_uint64s.
2138 (subreg_offset_from_lsb): Return a poly_uint64 rather than
2139 an unsigned int. Take the shift as a poly_uint64 rather than
2141 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
2142 as a poly_uint64 rather than an unsigned int.
2143 (simplify_subreg_regno): Likewise.
2144 (byte_lowpart_offset): Return the memory offset as a poly_int64
2146 (subreg_memory_offset): Likewise. Take the subreg offset as a
2147 poly_uint64 rather than an unsigned int.
2148 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
2149 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
2150 poly_uint64 rather than an unsigned int.
2151 * rtl.c (rtx_format): Describe 'p' in comment.
2152 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
2153 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
2154 offset as a poly_uint64 rather than an unsigned int.
2155 (byte_lowpart_offset): Return the memory offset as a poly_int64
2157 (subreg_memory_offset): Likewise. Take the subreg offset as a
2158 poly_uint64 rather than an unsigned int.
2159 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
2160 mode sizes as poly_uint64s rather than unsigned ints. Return a
2161 poly_uint64 rather than an unsigned int.
2162 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
2163 (copy_insn_1): Handle 'p'.
2164 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
2165 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
2166 an unsigned int. Return the shift in the same way.
2167 (subreg_lsb): Return the shift as a poly_uint64 rather than an
2169 (subreg_size_offset_from_lsb): Take the sizes and shift as
2170 poly_uint64s rather than unsigned ints. Return the offset as
2172 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
2173 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
2175 * rtlhash.c (add_rtx): Handle 'p'.
2176 * genemit.c (gen_exp): Likewise.
2177 * gengenrtl.c (type_from_format, gendef): Likewise.
2178 * gensupport.c (subst_pattern_match, get_alternatives_number)
2179 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
2180 (subst_dup): Likewise.
2181 * gengtype.c (adjust_field_rtx_def): Likewise.
2182 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
2183 (match_pattern_2): Likewise.
2184 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
2185 (rtx_test::subreg_field): New function.
2186 (operator ==, safe_to_hoist_p, transition_parameter_type)
2187 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
2188 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
2189 * genpeep.c (match_rtx): Likewise.
2190 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
2191 (rtx_writer::print_rtx_operand): Handle 'p'.
2192 (print_value): Handle SUBREG.
2193 * read-rtl.c (apply_int_iterator): Likewise.
2194 (rtx_reader::read_rtx_operand): Handle 'p'.
2195 * alias.c (rtx_equal_for_memref_p): Likewise.
2196 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
2197 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
2199 * calls.c (expand_call): Likewise.
2200 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
2201 (make_extraction, gen_lowpart_for_combine): Likewise.
2202 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
2204 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
2205 rather than an unsigned int. Treat subreg offsets as poly_ints.
2206 (exp_equiv_p): Handle 'p'.
2207 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
2208 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
2209 * dse.c (find_shift_sequence): Likewise.
2210 * dwarf2out.c (rtl_for_decl_location): Likewise.
2211 * expmed.c (extract_low_bits): Likewise.
2212 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
2213 (expand_expr_real_2): Likewise.
2214 * final.c (alter_subreg): Likewise.
2215 (leaf_renumber_regs_insn): Handle 'p'.
2216 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
2217 Treat subreg offsets as poly_ints.
2218 * fwprop.c (forward_propagate_and_simplify): Likewise.
2219 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
2220 * ira.c (get_subreg_tracking_sizes): Likewise.
2221 * ira-conflicts.c (go_through_subreg): Likewise.
2222 * ira-lives.c (process_single_reg_class_operands): Likewise.
2223 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
2224 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
2225 as a poly_uint64 rather than an unsigned int.
2226 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
2227 subreg offsets as poly_ints.
2228 * lra-constraints.c (operands_match_p): Handle 'p'.
2229 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
2230 * lra-spills.c (assign_mem_slot): Likewise.
2231 * postreload.c (move2add_valid_value_p): Likewise.
2232 * recog.c (general_operand, indirect_operand): Likewise.
2233 * regcprop.c (copy_value, maybe_mode_change): Likewise.
2234 (copyprop_hardreg_forward_1): Likewise.
2235 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
2236 (record_subregs_of_mode): Likewise.
2237 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
2238 * reload.c (operands_match_p): Handle 'p'.
2239 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
2240 * reload1.c (alter_reg, choose_reload_regs): Likewise.
2241 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
2242 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
2243 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
2244 (simplify_const_poly_int_tests<N>::run): Likewise.
2245 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
2246 a poly_uint64 rather than an unsigned int.
2247 * valtrack.c (debug_lowpart_subreg): Likewise.
2248 * var-tracking.c (var_lowpart): Likewise.
2249 (loc_cmp): Handle 'p'.
2251 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2252 Alan Hayward <alan.hayward@arm.com>
2253 David Sherwood <david.sherwood@arm.com>
2255 * ira.c (get_subreg_tracking_sizes): New function.
2256 (init_live_subregs): Take an integer size rather than a register.
2257 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
2258 to init_live_subregs.
2260 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2261 Alan Hayward <alan.hayward@arm.com>
2262 David Sherwood <david.sherwood@arm.com>
2264 * expr.c (store_constructor_field): Change bitsize from a
2265 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
2266 HOST_WIDE_INT to a poly_int64.
2267 (store_constructor): Change size from a HOST_WIDE_INT to
2269 (store_field): Likewise bitsize and bitpos.
2271 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2272 Alan Hayward <alan.hayward@arm.com>
2273 David Sherwood <david.sherwood@arm.com>
2275 * expmed.h (store_bit_field): Change bitregion_start and
2276 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
2277 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
2278 (store_bit_field_1, store_integral_bit_field, store_bit_field)
2279 (store_fixed_bit_field, store_split_bit_field): Likewise.
2280 * expr.c (store_constructor_field, store_field): Likewise.
2281 (optimize_bitfield_assignment_op): Likewise. Make the same change
2282 to bitsize and bitpos.
2283 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
2284 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
2285 same change in the constructor arguments.
2286 (get_best_mode): Change bitregion_start and bitregion_end from
2287 unsigned HOST_WIDE_INT to poly_uint64.
2288 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2289 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
2291 (bit_field_mode_iterator::next_mode): Update for new types
2292 of m_bitregion_start and m_bitregion_end.
2293 (get_best_mode): Change bitregion_start and bitregion_end from
2294 unsigned HOST_WIDE_INT to poly_uint64.
2296 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2297 Alan Hayward <alan.hayward@arm.com>
2298 David Sherwood <david.sherwood@arm.com>
2300 * rtl.h (simplify_gen_subreg): Add a temporary overload that
2301 accepts poly_uint64 offsets.
2302 * expmed.h (extract_bit_field): Take bitsize and bitnum as
2303 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2304 * expmed.c (lowpart_bit_field_p): Likewise.
2305 (extract_bit_field_as_subreg): New function, split out from...
2306 (extract_bit_field_1): ...here. Take bitsize and bitnum as
2307 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
2308 extractions, check that BITSIZE matches the size of the extracted
2309 value and that BITNUM is an exact multiple of that size.
2310 If all else fails, try forcing the value into memory if
2311 BITNUM is variable, and adjusting the address so that the
2312 offset is constant. Split the part that can only handle constant
2313 bitsize and bitnum out into...
2314 (extract_integral_bit_field): ...this new function.
2315 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
2316 rather than unsigned HOST_WIDE_INTs.
2318 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2319 Alan Hayward <alan.hayward@arm.com>
2320 David Sherwood <david.sherwood@arm.com>
2322 * expmed.h (store_bit_field): Take bitsize and bitnum as
2323 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2324 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
2325 that returns the byte size.
2326 (store_bit_field_1): Take bitsize and bitnum as
2327 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
2328 to simple_mem_bitfield_p. Split the part that can only handle
2329 constant bitsize and bitnum out into...
2330 (store_integral_bit_field): ...this new function.
2331 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
2332 than unsigned HOST_WIDE_INTs.
2333 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
2335 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2336 Alan Hayward <alan.hayward@arm.com>
2337 David Sherwood <david.sherwood@arm.com>
2339 * lra-int.h (lra_reg): Change offset from int to poly_int64.
2340 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
2342 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2343 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
2344 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
2345 offset as a poly_int64 rather than an int.
2346 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
2347 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2348 * lra-constraints.c (equiv_address_substitution): Track offsets
2350 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
2351 (curr_insn_transform): Handle the new form of sp_offset.
2352 * lra-eliminations.c (lra_elim_table): Change previous_offset
2353 and offset from HOST_WIDE_INT to poly_int64.
2354 (print_elim_table, update_reg_eliminate): Update accordingly.
2355 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
2356 (get_elimination): Update accordingly.
2357 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
2358 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2359 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
2360 poly_int64 offsets generally.
2361 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
2362 (mark_not_eliminable, init_elimination): Update accordingly.
2363 (remove_reg_equal_offset_note): Return a bool and pass the new
2364 offset back by pointer as a poly_int64.
2365 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
2366 rather than a HOST_WIDE_INT.
2367 (do_remat): Track offsets poly_int64s.
2368 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
2370 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2371 Alan Hayward <alan.hayward@arm.com>
2372 David Sherwood <david.sherwood@arm.com>
2374 * rtl.h (mem_attrs): Add a default constructor. Change size and
2375 offset from HOST_WIDE_INT to poly_int64.
2376 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
2377 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
2378 (widen_memory_access): Take the sizes and offsets as poly_int64s
2379 rather than HOST_WIDE_INTs.
2380 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
2381 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
2383 (adjust_offset_for_component_ref): Change the offset from a
2384 HOST_WIDE_INT to a poly_int64.
2385 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
2386 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
2387 * dce.c (find_call_stack_args): Likewise.
2388 * dse.c (record_store): Likewise.
2389 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
2390 * print-rtl.c (rtx_writer::print_rtx): Likewise.
2391 * read-rtl-function.c (test_loading_mem): Likewise.
2392 * rtlanal.c (may_trap_p_1): Likewise.
2393 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
2394 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
2395 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
2396 (mem_attrs::mem_attrs): New function.
2397 (set_mem_attributes_minus_bitpos): Change bitpos from a
2398 HOST_WIDE_INT to poly_int64.
2399 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
2400 (clear_mem_offset, clear_mem_size, change_address)
2401 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
2402 initialize mem_attrs.
2403 (set_mem_offset, set_mem_size, adjust_address_1)
2404 (adjust_automodify_address_1, offset_address, widen_memory_access):
2405 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
2407 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2408 Alan Hayward <alan.hayward@arm.com>
2409 David Sherwood <david.sherwood@arm.com>
2411 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
2412 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
2413 rather than 0 to represent an unknown size. Assert that the size
2414 is known when the mode isn't BLKmode.
2415 (may_trap_p_1): Use -1 for unknown sizes.
2416 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
2418 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2419 Alan Hayward <alan.hayward@arm.com>
2420 David Sherwood <david.sherwood@arm.com>
2422 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
2423 to poly_int64. Update commentary for positions_needed.large.
2424 (read_info_type): Change offset and width from HOST_WIDE_INT
2426 (set_usage_bits): Likewise.
2427 (canon_address): Return the offset as a poly_int64 rather than
2428 a HOST_WIDE_INT. Use strip_offset_and_add.
2429 (set_all_positions_unneeded, any_positions_needed_p): Use
2430 positions_needed.large to track stores with non-constant widths.
2431 (all_positions_needed_p): Likewise. Take the offset and width
2432 as poly_int64s rather than ints. Assert that rhs is nonnull.
2433 (record_store): Cope with non-constant offsets and widths.
2434 Nullify the rhs of an earlier store if we can't tell which bytes
2436 (find_shift_sequence): Take the access_size and shift as poly_int64s
2438 (get_stored_val): Take the read_offset and read_width as poly_int64s
2439 rather than HOST_WIDE_INTs.
2440 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
2441 non-constant offsets and widths.
2443 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2444 Alan Hayward <alan.hayward@arm.com>
2445 David Sherwood <david.sherwood@arm.com>
2447 * inchash.h (inchash::hash::add_poly_int): New function.
2448 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
2449 Use poly_int64 rather than HOST_WIDE_INT.
2450 (ao_ref::max_size_known_p): New function.
2451 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
2452 rather than HOST_WIDE_INT.
2453 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
2454 to temporaries until its interface is adjusted to match.
2455 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
2456 (aliasing_component_refs_p, decl_refs_may_alias_p)
2457 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
2458 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
2459 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
2461 * alias.c (ao_ref_from_mem): Likewise.
2462 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2463 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
2464 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
2465 (maybe_trim_complex_store, maybe_trim_constructor_store)
2466 (live_bytes_read, dse_classify_store): Likewise.
2467 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
2468 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
2469 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
2470 (vn_reference_lookup_3): Likewise.
2471 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
2473 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2474 Alan Hayward <alan.hayward@arm.com>
2475 David Sherwood <david.sherwood@arm.com>
2477 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
2478 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
2479 instead of ranges_overlap_p.
2481 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2482 Alan Hayward <alan.hayward@arm.com>
2483 David Sherwood <david.sherwood@arm.com>
2485 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
2486 sizes as poly_int64s rather than HOST_WIDE_INTs.
2488 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2489 Alan Hayward <alan.hayward@arm.com>
2490 David Sherwood <david.sherwood@arm.com>
2492 * gimple-fold.h (fold_ctor_reference): Take the offset and size
2493 as poly_uint64 rather than unsigned HOST_WIDE_INT.
2494 * gimple-fold.c (fold_ctor_reference): Likewise.
2496 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2497 Alan Hayward <alan.hayward@arm.com>
2498 David Sherwood <david.sherwood@arm.com>
2500 * target.def (dwarf_poly_indeterminate_value): New hook.
2501 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
2502 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
2503 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
2504 * doc/tm.texi: Regenerate.
2505 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
2506 offset as a poly_int64.
2507 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
2508 offset as a poly_int64.
2509 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
2510 Take the offset as a poly_int64.
2511 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
2512 (frame_pointer_fb_offset): Change to a poly_int64.
2513 (int_loc_descriptor): Take the offset as a poly_int64. Use
2514 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
2515 (based_loc_descr): Take the offset as a poly_int64.
2516 Use strip_offset_and_add to handle (plus X (const)).
2517 Use new_reg_loc_descr instead of an open-coded version of the
2518 previous implementation.
2519 (mem_loc_descriptor): Handle CONST_POLY_INT.
2520 (compute_frame_pointer_to_fb_displacement): Take the offset as a
2521 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
2523 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2524 Alan Hayward <alan.hayward@arm.com>
2525 David Sherwood <david.sherwood@arm.com>
2527 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
2528 (gen_rtx_REG_offset): Take the offset as a poly_int64.
2529 * inchash.h (inchash::hash::add_poly_hwi): New function.
2530 * gengtype.c (main): Register poly_int64.
2531 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
2532 offset as a poly_int.
2533 (reg_attr_hasher::equal): Use must_eq to compare offsets.
2534 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
2535 offset as a poly_int64.
2536 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
2537 * print-rtl.c (print_poly_int): New function.
2538 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
2540 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
2542 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
2543 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
2544 (vt_get_decl_and_offset): Return the offset as a poly_int64.
2545 Enforce track_offset_p for parts of a PARALLEL.
2546 (vt_add_function_parameter): Use const_offset for the final
2547 offset to track. Use get_tracked_reg_offset for the parts
2550 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2551 Alan Hayward <alan.hayward@arm.com>
2552 David Sherwood <david.sherwood@arm.com>
2554 * target.def (truly_noop_truncation): Take poly_uint64s instead of
2555 unsigned ints. Change default to hook_bool_puint64_puint64_true.
2556 * doc/tm.texi: Regenerate.
2557 * hooks.h (hook_bool_uint_uint_true): Delete.
2558 (hook_bool_puint64_puint64_true): Declare.
2559 * hooks.c (hook_bool_uint_uint_true): Delete.
2560 (hook_bool_puint64_puint64_true): New function.
2561 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
2562 instead of unsigned ints.
2563 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
2564 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
2566 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2567 Alan Hayward <alan.hayward@arm.com>
2568 David Sherwood <david.sherwood@arm.com>
2570 * optabs.h (expand_operand): Add an int_value field.
2571 (create_expand_operand): Add an int_value parameter and use it
2572 to initialize the new expand_operand field.
2573 (create_integer_operand): Replace with a declaration of a function
2574 that accepts poly_int64s. Move the implementation to...
2575 * optabs.c (create_integer_operand): ...here.
2576 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
2577 the mode preserves the value of int_value, instead of calling
2578 const_int_operand on the rtx. Use gen_int_mode to generate
2581 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2582 Alan Hayward <alan.hayward@arm.com>
2583 David Sherwood <david.sherwood@arm.com>
2585 * dumpfile.h (dump_dec): Declare.
2586 * dumpfile.c (dump_dec): New function.
2587 * pretty-print.h (pp_wide_integer): Turn into a function and
2588 declare a poly_int version.
2589 * pretty-print.c (pp_wide_integer): New function for poly_ints.
2591 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2592 Alan Hayward <alan.hayward@arm.com>
2593 David Sherwood <david.sherwood@arm.com>
2595 * doc/generic.texi (POLY_INT_CST): Document.
2596 * tree.def (POLY_INT_CST): New tree code.
2597 * treestruct.def (TS_POLY_INT_CST): New tree layout.
2598 * tree-core.h (tree_poly_int_cst): New struct.
2599 (tree_node): Add a poly_int_cst field.
2600 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
2601 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
2602 instead of a wide_int_ref.
2603 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
2605 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2606 instead of an unsigned HOST_WIDE_INT.
2607 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
2608 (ptrdiff_tree_p): Declare.
2609 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
2610 extern inline implementations if the target doesn't use POLY_INT_CST.
2611 (poly_int_tree_p): New function.
2612 (wi::unextended_tree): New class.
2613 (wi::int_traits <unextended_tree>): New override.
2614 (wi::extended_tree): Add a default constructor.
2615 (wi::extended_tree::get_tree): New function.
2616 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
2617 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
2618 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
2619 (wi::tree_to_poly_wide_ref): New typedefs.
2620 (wi::ints_for): Provide overloads for extended_tree and
2622 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
2623 (wi::to_wide): New functions.
2624 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
2625 * tree.c (poly_int_cst_hasher): New struct.
2626 (poly_int_cst_hash_table): New variable.
2627 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
2628 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
2630 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
2631 (init_ttree): Initialize poly_int_cst_hash_table.
2632 (build_int_cst, build_int_cst_type, build_invariant_address): Take
2633 a poly_int64 instead of a HOST_WIDE_INT.
2634 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2635 instead of an unsigned HOST_WIDE_INT.
2636 (wide_int_to_tree): Rename to...
2637 (wide_int_to_tree_1): ...this.
2638 (build_new_poly_int_cst, build_poly_int_cst): New functions.
2639 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
2640 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
2641 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
2642 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
2643 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
2645 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
2646 (streamer_read_tree_body): Likewise.
2647 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
2648 (streamer_write_tree_body): Likewise.
2649 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
2650 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
2651 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
2652 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
2653 * gimple-expr.h (is_gimple_constant): Likewise.
2654 * gimplify.c (maybe_with_size_expr): Likewise.
2655 * print-tree.c (print_node): Likewise.
2656 * tree-data-ref.c (data_ref_compare_tree): Likewise.
2657 * tree-pretty-print.c (dump_generic_node): Likewise.
2658 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2659 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
2660 * tree-vrp.c (compare_values_warnv): Likewise.
2661 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
2662 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
2663 (force_expr_to_var_cost): Likewise.
2664 * tree-ssa-loop.c (for_each_index): Likewise.
2665 * fold-const.h (build_invariant_address, size_int_kind): Take a
2666 poly_int64 instead of a HOST_WIDE_INT.
2667 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
2668 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
2670 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
2671 (int_const_binop_2): New function, split out from...
2672 (int_const_binop_1): ...here. Handle POLY_INT_CST.
2673 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
2674 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
2675 * gimple-ssa-strength-reduction.c (slsr_process_add)
2676 (slsr_process_mul): Check for INTEGER_CSTs before using them
2678 * stor-layout.c (bits_from_bytes): New function.
2679 (bit_from_pos): Use it.
2680 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
2681 by BITS_PER_UNIT to get the TYPE_SIZE.
2682 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
2683 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
2685 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2686 Alan Hayward <alan.hayward@arm.com>
2687 David Sherwood <david.sherwood@arm.com>
2689 * doc/rtl.texi (const_poly_int): Document. Also document the
2690 rtl sharing behavior.
2691 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
2692 * rtl.h (const_poly_int_def): New struct.
2693 (rtx_def::u): Add a cpi field.
2694 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
2695 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
2696 (wi::rtx_to_poly_wide_ref): New typedef
2697 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
2698 (poly_int_rtx_p): New functions.
2699 (trunc_int_for_mode): Declare a poly_int64 version.
2700 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2701 (immed_wide_int_const): Take a poly_wide_int_ref rather than
2703 (strip_offset): Declare.
2704 (strip_offset_and_add): New function.
2705 * rtl.def (CONST_POLY_INT): New rtx code.
2706 * rtl.c (rtx_size): Handle CONST_POLY_INT.
2707 (shared_const_p): Use poly_int_rtx_p.
2708 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
2710 (gen_int_shift_amount): Likewise.
2711 * emit-rtl.c (const_poly_int_hasher): New class.
2712 (const_poly_int_htab): New variable.
2713 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
2714 (const_poly_int_hasher::hash): New function.
2715 (const_poly_int_hasher::equal): Likewise.
2716 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
2717 (immed_wide_int_const): Rename to...
2718 (immed_wide_int_const_1): ...this and make static.
2719 (immed_wide_int_const): New function, taking a poly_wide_int_ref
2720 instead of a wide_int_ref.
2721 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
2722 (gen_lowpart_common): Handle CONST_POLY_INT.
2723 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
2724 * cselib.c (cselib_hash_rtx): Likewise.
2725 * dwarf2out.c (const_ok_for_output_1): Likewise.
2726 * expr.c (convert_modes): Likewise.
2727 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
2728 * rtlhash.c (add_rtx): Likewise.
2729 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
2730 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2731 Handle existing CONST_POLY_INT rtxes.
2732 * expmed.h (expand_shift): Take a poly_int64 instead of a
2734 * expmed.c (expand_shift): Likewise.
2735 * rtlanal.c (strip_offset): New function.
2736 (commutative_operand_precedence): Give CONST_POLY_INT the same
2737 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
2739 * rtl-tests.c (const_poly_int_tests): New struct.
2740 (rtl_tests_c_tests): Use it.
2741 * simplify-rtx.c (simplify_const_unary_operation): Handle
2743 (simplify_const_binary_operation): Likewise.
2744 (simplify_binary_operation_1): Fold additions of symbolic constants
2745 and CONST_POLY_INTs.
2746 (simplify_subreg): Handle extensions and truncations of
2748 (simplify_const_poly_int_tests): New struct.
2749 (simplify_rtx_c_tests): Use it.
2750 * wide-int.h (storage_ref): Add default constructor.
2751 (wide_int_ref_storage): Likewise.
2752 (trailing_wide_ints): Use GTY((user)).
2753 (trailing_wide_ints::operator[]): Add a const version.
2754 (trailing_wide_ints::get_precision): New function.
2755 (trailing_wide_ints::extra_size): Likewise.
2757 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2758 Alan Hayward <alan.hayward@arm.com>
2759 David Sherwood <david.sherwood@arm.com>
2761 * emit-rtl.h (gen_int_shift_amount): Declare.
2762 * emit-rtl.c (gen_int_shift_amount): New function.
2763 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
2765 * calls.c (shift_return_value): Likewise.
2766 * cse.c (fold_rtx): Likewise.
2767 * dse.c (find_shift_sequence): Likewise.
2768 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
2769 (expand_shift, expand_smod_pow2): Likewise.
2770 * lower-subreg.c (shift_cost): Likewise.
2771 * optabs.c (expand_superword_shift, expand_doubleword_mult)
2772 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
2773 (expand_vec_perm_var): Likewise.
2774 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2775 (simplify_binary_operation_1): Likewise.
2776 * combine.c (try_combine, find_split_point, force_int_to_mode)
2777 (simplify_shift_const_1, simplify_shift_const): Likewise.
2778 (change_zero_ext): Likewise. Use simplify_gen_binary.
2780 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2782 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2784 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2786 * doc/invoke.texi (ARM Options): Document accepted extension options
2787 for -march=armv8.3-a.
2789 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2792 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2793 or --with-float=softfp, set the default CPU to arm10e.
2795 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2797 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2798 * config/visium/predicates.md (const_shift_operand): Likewise.
2799 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2800 (visium_legitimize_reload_address): Likewise.
2802 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2805 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2807 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2810 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2812 2017-12-20 Julia Koval <julia.koval@intel.com>
2814 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2815 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2816 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2817 * config.gcc: Include vpclmulqdqintrin.h.
2818 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2819 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2821 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2822 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2823 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2824 * config/i386/i386.c (isa2_opts): Add -mcx16.
2825 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2826 (ix86_option_override_internal): Move mcx16 to flags2.
2827 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2828 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2829 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2830 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2831 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2832 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2833 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2834 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2835 * doc/invoke.texi: Add -mvpclmulqdq.
2837 2017-12-20 Tom de Vries <tom@codesourcery.com>
2840 * config/i386/i386.c (ix86_static_chain): Move
2841 DECL_STATIC_CHAIN test ...
2842 * calls.c (rtx_for_static_chain): ... here. New function.
2843 * calls.h (rtx_for_static_chain): Declare.
2844 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2845 instead of targetm.calls.static_chain.
2846 * df-scan.c (df_get_entry_block_def_set): Same.
2848 2017-12-19 Tom de Vries <tom@codesourcery.com>
2850 PR tree-optimization/83493
2851 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2854 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2856 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2857 inform with hardcoded english plural handling.
2859 2017-12-18 Jeff Law <law@redhat.com>
2861 PR tree-optimization/83477
2862 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2863 a non-virtual PHI, always push a new range.
2865 2017-12-19 Martin Sebor <msebor@redhat.com>
2868 * builtins.c (compute_objsize): Handle non-constant offsets.
2870 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2872 PR tree-optimization/83444
2873 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2874 character load case, if get_stridx on MEM_REF's operand doesn't
2875 look usable, retry with get_addr_stridx.
2877 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2880 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2881 (variable_tracking_main_1): Keep markers even when VTA fails.
2884 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2885 even if there are markers before it.
2886 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2888 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2890 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2891 typical order conditions.
2892 * sel-sched.c (extract_new_fences_from): Likewise.
2893 * config/visium/constraints.md (J, K, L): Likewise.
2894 * config/visium/predicates.md (const_shift_operand): Likewise.
2895 * config/visium/visium.c (visium_legitimize_address,
2896 visium_legitimize_reload_address): Likewise.
2897 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2898 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2899 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2900 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2901 SET_DUMP_DETAIL): Likewise.
2902 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2903 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2904 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2905 avr_set_core_architecture, avr_set_current_function,
2906 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2907 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2908 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2909 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2910 avr_map_decompose, avr_fold_builtin): Likewise.
2911 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2912 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2913 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2914 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2915 * config/m32c/m32c.c (m32c_conditional_register_usage,
2916 m32c_address_cost): Likewise.
2917 * config/m32c/predicates.md (shiftcount_operand,
2918 longshiftcount_operand): Likewise.
2919 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2920 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2921 can_use_cdx_ldstw): Likewise.
2922 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2923 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2925 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2926 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2927 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2928 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2929 * config/vax/vax.c (adjacent_operands_p): Likewise.
2930 * config/ft32/constraints.md (L, b, KA): Likewise.
2931 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2933 * cfgexpand.c (expand_stack_alignment): Likewise.
2934 * gcse.c (insert_expr_in_table): Likewise.
2935 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2936 * cgraphunit.c (cgraph_node::expand): Likewise.
2937 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2938 * emit-rtl.c (add_insn): Likewise.
2939 * input.c (dump_location_info): Likewise.
2940 * passes.c (NEXT_PASS): Likewise.
2941 * read-rtl-function.c (parse_note_insn_name,
2942 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2944 * sched-rgn.c (sched_rgn_init): Likewise.
2945 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2946 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2947 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2949 * explow.c (eliminate_constant_term): Likewise.
2950 * final.c (leaf_renumber_regs_insn): Likewise.
2951 * cfgrtl.c (print_rtl_with_bb): Likewise.
2952 * genhooks.c (emit_init_macros): Likewise.
2953 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2954 * tree-data-ref.c (conflict_fn): Likewise.
2955 * selftest.c (assert_streq): Likewise.
2956 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2957 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2958 fold_binary_loc, multiple_of_p): Likewise.
2959 * reload.c (push_reload, find_equiv_reg): Likewise.
2960 * et-forest.c (et_nca, et_below): Likewise.
2961 * dbxout.c (dbxout_symbol_location): Likewise.
2962 * reorg.c (relax_delay_slots): Likewise.
2963 * dojump.c (do_compare_rtx_and_jump): Likewise.
2964 * gengtype-parse.c (type): Likewise.
2965 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2966 simplify_const_relational_operation): Likewise.
2967 * reload1.c (do_output_reload): Likewise.
2968 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2969 * gengtype.c (type_for_name): Likewise.
2970 * gimple-ssa-sprintf.c (format_directive): Likewise.
2972 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2975 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2976 accessing it. Adjust comment.
2978 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2981 * predict.c (zero_one_minusone): New function.
2982 (apply_return_prediction): Avoid return prediction for functions
2983 returning only -1, 0 and 1 values, unless they only return -1 and 0
2986 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2988 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2989 fall-through warning.
2990 (arc_compute_frame_size): Remove unused variables.
2991 (arc_print_operand): Fix fprintif format.
2992 (arc_can_follow_jump): Clean fall-through warning.
2994 2017-12-19 Marek Polacek <polacek@redhat.com>
2997 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
3000 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
3002 * config/arc/arc.c (overriderregs): New variable.
3003 (arc_override_options): Track fixed/call saved/call options.
3004 (arc_conditional_register_usage): Check against overrideregs
3005 variable whenever we change register properties.
3007 2017-12-19 Nathan Sidwell <nathan@acm.org>
3009 * opts.c (finish_options): Don't prefix dump_base_name if it
3010 already contains directories.
3012 2017-12-19 Martin Liska <mliska@suse.cz>
3014 PR rtl-optimization/82675
3015 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
3016 more element in sbitmap.
3018 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
3020 * gimplify.c (gimplify_expr): Use error_operand_p.
3022 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
3025 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
3028 2017-12-19 Daniel Cederman <cederman@gaisler.com>
3030 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
3031 the jump is to a label.
3033 2017-12-19 Jakub Jelinek <jakub@redhat.com>
3035 PR tree-optimization/83444
3036 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
3041 * ipa-inline.c (flatten_remove_node_hook): New function.
3042 (ipa_inline): Keep only nodes with flatten attribute at the end of
3043 the array in the order from ipa_reverse_postorder, only walk that
3044 portion of array for flattening, if there is more than one such
3045 node, temporarily register a removal hook and ignore removed nodes.
3047 PR tree-optimization/80631
3048 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
3049 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
3050 IFN_REDUC_MAX or IFN_REDUC_MIN.
3052 2017-12-18 Jeff Law <law@redhat.com>
3054 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
3055 record symbolic equivalences from backedges in the CFG.
3058 2017-11-19 Jeff Law <law@redhat.com>
3060 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
3061 of degenerates resulting from ignoring an edge.
3063 2017-12-18 Martin Sebor <msebor@redhat.com>
3066 PR tree-optimization/78450
3067 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
3068 (handle_builtin_strlen): Call it.
3070 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
3072 PR rtl-optimization/83424
3073 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
3075 2017-12-18 Marek Polacek <polacek@redhat.com>
3078 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
3079 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
3082 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
3084 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
3085 (umaddsidi4, umaddsidi_split): Likewise.
3087 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
3089 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
3092 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3095 * config/s390/s390.c (s390_option_override): Avoid strncpy.
3097 2017-12-18 Richard Biener <rguenther@suse.de>
3099 PR tree-optimization/81877
3100 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
3101 (outermost_indep_loop): Adjust.
3102 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
3103 (can_sm_ref_p): Adjust.
3105 2017-12-18 Richard Biener <rguenther@suse.de>
3108 * tree.c (array_at_struct_end_p): Return true if the underlying
3109 object has space for at least one element in excess of what
3110 the array domain specifies.
3112 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3114 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
3115 example to avoid overfull hbox.
3116 * doc/invoke.texi (Option Summary): Add missing @gol.
3117 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
3120 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3121 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3123 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
3124 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
3125 (AArch64 Options): Document -mverbose-cost-dump.
3126 (ARM Options): Likewise, plus -mflip-thumb.
3128 2017-12-17 Martin Sebor <msebor@redhat.com>
3131 * gimple-ssa-warn-restrict.c
3132 (builtin_memref::offset_out_of_bounds): Correct the handling of
3135 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3137 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
3139 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
3141 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
3143 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
3145 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
3148 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
3150 * doc/invoke.texi: Fix some typos.
3152 2017-12-16 Martin Sebor <msebor@redhat.com>
3154 PR tree-optimization/78918
3155 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
3156 * builtins.c (check_sizes): Rename...
3157 (check_access): ...to this. Rename function arguments for clarity.
3158 (check_memop_sizes): Adjust names.
3159 (expand_builtin_memchr, expand_builtin_memcpy): Same.
3160 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
3161 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
3162 (check_strncat_sizes, expand_builtin_strncat): Same.
3163 (expand_builtin_strncpy, expand_builtin_memset): Same.
3164 (expand_builtin_bzero, expand_builtin_memcmp): Same.
3165 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
3166 (maybe_emit_sprintf_chk_warning): Same.
3167 (expand_builtin_strcpy): Adjust.
3168 (expand_builtin_stpcpy): Same.
3169 (expand_builtin_with_bounds): Detect out-of-bounds accesses
3170 in pointer-checking forms of memcpy, memmove, and mempcpy.
3171 (gcall_to_tree_minimal, max_object_size): Define new functions.
3172 * builtins.h (max_object_size): Declare.
3173 * calls.c (alloc_max_size): Call max_object_size instead of
3174 hardcoding ssizetype limit.
3175 (get_size_range): Handle new argument.
3176 * calls.h (get_size_range): Add a new argument.
3177 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
3178 * doc/invoke.texi (-Wrestrict): Adjust, add example.
3179 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
3181 (gimple_fold_builtin_memory_chk): Same.
3182 (gimple_fold_builtin_stxcpy_chk): New function.
3183 * gimple-ssa-warn-restrict.c: New source.
3184 * gimple-ssa-warn-restrict.h: New header.
3185 * gimple.c (gimple_build_call_from_tree): Propagate location.
3186 * passes.def (pass_warn_restrict): Add new pass.
3187 * tree-pass.h (make_pass_warn_restrict): Declare.
3188 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
3190 (handle_builtin_strcat): Same.
3191 (strlen_optimize_stmt): Rename...
3192 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
3193 stpncpy, strncpy, and their checking forms.
3195 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
3197 PR rtl-optimization/82849
3198 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
3199 and get_max_loop_iterations_int.
3201 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3202 Alan Hayward <alan.hayward@arm.com>
3203 David Sherwood <david.sherwood@arm.com>
3205 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
3206 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
3207 size as a poly_uint64.
3208 (mode_for_vector, mode_for_int_vector): Take the number of vector
3209 elements as a poly_uint64.
3210 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
3211 size as a poly_uint64.
3212 (mode_for_vector, mode_for_int_vector): Take the number of vector
3213 elements as a poly_uint64.
3215 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3216 Alan Hayward <alan.hayward@arm.com>
3217 David Sherwood <david.sherwood@arm.com>
3219 * machmode.h (MACRO_MODE): New macro.
3220 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
3221 * targhooks.c (default_libcall_value, default_secondary_reload)
3222 (default_memory_move_cost, default_register_move_cost)
3223 (default_class_max_nregs): Likewise.
3225 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3226 Alan Hayward <alan.hayward@arm.com>
3227 David Sherwood <david.sherwood@arm.com>
3229 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
3230 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
3232 * genautomata.c (main): Likewise.
3233 * genconditions.c (write_header): Likewise.
3234 * genemit.c (main): Likewise.
3235 * genextract.c (print_header): Likewise.
3236 * genopinit.c (main): Likewise.
3237 * genoutput.c (output_prologue): Likewise.
3238 * genpeep.c (main): Likewise.
3239 * genpreds.c (write_insn_preds_c): Likewise.
3240 * genrecog.c (writer_header): Likewise.
3241 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
3242 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
3243 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
3244 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
3245 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
3246 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
3247 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
3248 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
3249 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
3250 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
3251 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
3252 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
3253 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
3254 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
3255 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
3256 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
3257 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
3258 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
3259 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
3260 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
3261 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
3262 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
3263 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
3264 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
3265 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
3266 * config/darwin.c (IN_TARGET_CODE): Likewise.
3267 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
3268 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
3269 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
3270 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
3271 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
3272 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
3273 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
3274 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
3275 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
3276 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
3277 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
3278 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
3279 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
3280 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
3281 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
3282 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
3283 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
3284 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
3285 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
3286 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
3287 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
3288 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
3289 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
3290 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
3291 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
3292 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
3293 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
3294 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
3295 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
3296 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
3297 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
3298 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
3299 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
3300 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
3301 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
3302 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
3303 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
3304 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
3305 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
3306 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
3307 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
3308 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
3309 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
3310 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
3311 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
3312 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
3313 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
3314 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
3315 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
3316 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
3317 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
3318 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
3319 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
3320 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
3321 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
3322 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
3323 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
3324 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
3325 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
3326 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
3327 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3328 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
3329 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
3330 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
3331 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
3332 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
3333 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
3334 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
3335 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
3336 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
3337 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
3338 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3339 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
3340 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
3341 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
3342 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
3343 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
3344 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
3345 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
3346 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
3347 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
3348 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
3349 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
3350 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
3351 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
3352 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
3353 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
3354 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
3355 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
3356 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
3357 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
3358 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
3359 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
3360 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
3361 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
3362 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
3363 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
3364 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
3365 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
3366 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
3367 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
3368 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
3369 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
3370 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
3371 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
3372 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
3374 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3375 Alan Hayward <alan.hayward@arm.com>
3376 David Sherwood <david.sherwood@arm.com>
3378 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
3381 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3382 Alan Hayward <alan.hayward@arm.com>
3383 David Sherwood <david.sherwood@arm.com>
3385 * doc/generic.texi (VEC_SERIES_EXPR): Document.
3386 * doc/md.texi (vec_series@var{m}): Document.
3387 * tree.def (VEC_SERIES_EXPR): New tree code.
3388 * tree.h (build_vec_series): Declare.
3389 * tree.c (build_vec_series): New function.
3390 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
3391 * tree-pretty-print.c (dump_generic_node): Likewise.
3392 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
3393 * tree-inline.c (estimate_operator_cost): Likewise.
3394 * expr.c (expand_expr_real_2): Likewise.
3395 * optabs-tree.c (optab_for_tree_code): Likewise.
3396 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
3397 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
3398 * expmed.c (make_tree): Handle VEC_SERIES.
3399 * optabs.def (vec_series_optab): New optab.
3400 * optabs.h (expand_vec_series_expr): Declare.
3401 * optabs.c (expand_vec_series_expr): New function.
3402 * tree-vect-generic.c (expand_vector_operations_1): Check that
3403 the operands also have vector type.
3405 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3406 Alan Hayward <alan.hawyard@arm.com>
3407 David Sherwood <david.sherwood@arm.com>
3409 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
3410 (VEC_COND_EXPR): Add missing @tindex.
3411 * doc/md.texi (vec_duplicate@var{m}): Document.
3412 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
3413 * tree.c (build_vector_from_val): Add stubbed-out handling of
3414 variable-length vectors, using VEC_DUPLICATE_EXPR.
3415 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
3416 * cfgexpand.c (expand_debug_expr): Likewise.
3417 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3418 * tree-inline.c (estimate_operator_cost): Likewise.
3419 * tree-pretty-print.c (dump_generic_node): Likewise.
3420 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
3421 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
3422 (test_vec_duplicate_folding): New function.
3423 (fold_const_c_tests): Call it.
3424 * optabs.def (vec_duplicate_optab): New optab.
3425 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
3426 * optabs.h (expand_vector_broadcast): Declare.
3427 * optabs.c (expand_vector_broadcast): Make non-static. Try using
3428 vec_duplicate_optab.
3429 * expr.c (store_constructor): Try using vec_duplicate_optab for
3431 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
3433 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
3436 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
3437 div/mod latencies a bit.
3439 2017-12-15 Jeff Law <law@redhat.com>
3441 PR tree-optimization/36550
3442 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
3443 (mark_threaded_blocks): Rewrite code to avoid block copying when
3444 optimizing for size. Don't pessimize blocks which will be
3445 copied, but all the statements will be dead.
3447 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
3449 PR tree-optimization/81165
3450 * tree-ssa-threadupdate.c (uses_in_bb): New.
3451 (estimate_threading_killed_stmts): New.
3452 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
3453 * tree-ssa-threadedge.c
3454 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
3457 2017-12-15 Jeff Law <law@redhat.com>
3459 PR tree-optimization/83410
3460 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
3461 threads when parallelizing loops.
3463 2017-12-15 Jakub Jelinek <jakub@redhat.com>
3465 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
3467 * config/alpha/alpha.c (vms_attribute_table): Swap
3468 affects_type_identity and handler fields, adjust comments.
3469 * config/mips/mips.c (mips_attribute_table): Likewise.
3470 * config/visium/visium.c (visium_attribute_table): Likewise.
3471 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
3472 * config/microblaze/microblaze.c (microblaze_attribute_table):
3474 * config/spu/spu.c (spu_attribute_table): Likewise.
3475 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
3476 * config/arc/arc.c (arc_attribute_table): Likewise.
3477 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
3478 * config/v850/v850.c (v850_handle_interrupt_attribute,
3479 v850_handle_data_area_attribute): Formatting fixes.
3480 (v850_attribute_table): Swap affects_type_identity and handler
3481 fields, adjust comments.
3482 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
3483 * config/arm/arm.c (arm_attribute_table): Likewise.
3484 * config/avr/avr.c (avr_attribute_table): Likewise.
3485 * config/s390/s390.c (s390_attribute_table): Likewise.
3486 * config/sh/sh.c (sh_attribute_table): Likewise.
3487 * config/i386/i386.c (ix86_handle_cconv_attribute,
3488 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
3489 (ix86_attribute_table): Swap affects_type_identity and handler
3490 fields, adjust comments.
3491 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3492 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
3493 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
3494 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
3495 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
3496 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3497 * config/rx/rx.c (rx_attribute_table): Likewise.
3498 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
3499 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
3500 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
3501 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
3502 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3503 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
3504 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
3505 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
3506 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3507 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3508 * doc/plugins.texi (user_attr): Likewise. Add NULL for
3510 * attribs.c (empty_attribute_table): Swap affects_type_identity and
3512 (register_scoped_attributes, decl_attributes): Formatting fixes.
3514 PR tree-optimization/83269
3515 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
3516 subtraction in arg0's type if type is signed and arg0 is unsigned.
3520 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
3521 convert? on @0 instead of convert. Check type of @1, not @0.
3522 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
3523 convert? on @0 instead of convert. Check type of @1, not @0.
3524 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
3525 only optimize if either both @1 and @2 types are narrower
3526 precision, or both are wider or equal precision, and in the former
3527 case only if both have undefined overflow.
3529 2017-12-15 Richard Biener <rguenther@suse.de>
3532 * internal-fn.def (IFN_NOP): Add.
3533 * internal-fn.c (expand_NOP): Do nothing.
3534 * lto-streamer-in.c (input_function): Instead of removing
3535 sanitizer calls replace them with IFN_NOP calls.
3537 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
3538 Alan Hayward <alan.hayward@arm.com>
3539 David Sherwood <david.sherwood@arm.com>
3541 * dse.c (store_info, read_info_type): Replace begin and end with
3543 (print_range): New function.
3544 (set_all_positions_unneeded, any_positions_needed_p)
3545 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
3547 (record_store): Likewise. Optimize the case in which all positions
3549 (get_stored_val): Replace read_begin and read_end with read_offset
3551 (replace_read): Update call accordingly.
3553 2017-12-15 Bin Cheng <bin.cheng@arm.com>
3555 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
3556 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
3557 (loop_cand::loop_cand): Initialize above members.
3558 (loop_cand::supported_operations): Delete.
3559 (loop_cand::can_interchange_p): Inline above function.
3560 (loop_cand::classify_simple_reduction): Record number of constant
3561 initialized simple reductions.
3562 (should_interchange_loops): New parameters. Check stmt cost of loops
3564 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
3565 Update call to should_interchange_loops.
3566 (should_interchange_loop_nest): Update call to
3567 should_interchange_loops.
3569 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
3572 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
3573 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
3574 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
3576 2017-12-15 Julia Koval <julia.koval@intel.com>
3578 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
3579 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
3581 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
3582 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
3583 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
3585 2017-12-15 Julia Koval <julia.koval@intel.com>
3587 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
3588 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
3589 * config/i386/sse.md (vaesenc_<mode>): New pattern.
3590 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
3591 _mm_aesenc_epi128): New intrinsics.
3593 2017-12-15 Julia Koval <julia.koval@intel.com>
3595 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
3596 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
3598 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
3599 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
3600 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
3602 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3604 * gimple-ssa-strength-reduction.c (analyze_increments):
3605 Distinguish replacement costs for constant strides from those for
3608 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3610 * var-tracking.c (variable_tracking_main_1): Formatting fix.
3612 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3614 * doc/invoke.texi: Document -Wcast-function-type.
3615 * recog.h (stored_funcptr): Change signature.
3616 * tree-dump.c (dump_node): Avoid warning.
3617 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
3619 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
3622 * gimple-fold.c (get_range_strlen): Add the handling of non-member
3625 2017-12-14 David Malcolm <dmalcolm@redhat.com>
3627 PR tree-optimization/83312
3628 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
3629 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
3630 "val" param, and to cope with arbitrary basic blocks.
3631 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
3632 handle NULL_TREE for "val", dropping "bb" param.
3633 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
3634 drop "bb" param. Handle NULL_TREE for "val".
3635 (find_case_label_for_value): Make "switch_stmt" param const.
3636 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
3638 (vrp_prop::check_all_array_refs): Reimplement as...
3639 (check_array_bounds_dom_walker::before_dom_children): ...this new
3640 vfunc. Replace linear search through BB block list, excluding
3641 those with non-executable in-edges via dominator walk.
3643 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3645 * config/arm/arm.opt (mverbose-cost-dump): New option.
3646 * config/arm/arm.c (arm_rtx_costs): Use it.
3648 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
3651 * reload1.c (emit_input_reload_insns): Skip debug markers.
3653 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
3655 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
3659 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
3660 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
3661 * config/sh/sh.c: Skip debug insns besides notes.
3662 * config/sh/sh.md: Likewise.
3663 * config/sh/sh_treg_combine.cc: Likewise.
3664 * config/sh/sync.md: Likewise.
3666 2017-12-14 Tom de Vries <tom@codesourcery.com>
3668 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3669 Add item for weak_undefined.
3671 2017-12-14 Richard Biener <rguenther@suse.de>
3673 PR tree-optimization/67842
3674 * sese.h (bb_in_region): Remove #if 0'ed code.
3676 2017-12-14 Richard Biener <rguenther@suse.de>
3678 PR tree-optimization/83326
3679 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3680 may_be_zero parameter and handle it by not marking the first
3681 peeled copy as not exiting the loop.
3682 (try_peel_loop): Likewise.
3683 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
3684 to handle the case of constant or zero iterations and perform
3685 loop header copying on-the-fly.
3687 2017-12-14 Richard Biener <rguenther@suse.de>
3689 PR tree-optimization/83418
3690 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
3691 Instead of asserting we don't get unfolded comparisons deal with them.
3693 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3696 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
3697 basic blocks. Assert debug bind insns don't appear outside of bbs,
3698 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
3701 PR tree-optimization/83198
3702 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
3703 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
3704 value if arg is a REAL_CST with incompatible type.
3706 2017-12-14 Sudakshina Das <sudi.das@arm.com>
3707 Bin Cheng <bin.cheng@arm.com>
3710 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
3712 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
3715 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3717 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
3718 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
3720 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
3721 Alan Hayward <alan.hayward@arm.com>
3722 David Sherwood <david.sherwood@arm.com>
3724 * poly-int.h: New file.
3725 * poly-int-types.h: Likewise.
3726 * coretypes.h: Include them.
3727 (POLY_INT_CONVERSION): Define.
3728 * target.def (estimated_poly_value): New hook.
3729 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
3730 * doc/tm.texi: Regenerate.
3731 * doc/poly-int.texi: New file.
3732 * doc/gccint.texi: Include it.
3733 * doc/rtl.texi: Describe restrictions on subreg modes.
3734 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
3735 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
3736 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
3737 * targhooks.h (default_estimated_poly_value): Declare.
3738 * targhooks.c (default_estimated_poly_value): New function.
3739 * target.h (estimated_poly_value): Likewise.
3740 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
3741 (wi::unary_traits): Delete.
3742 (wi::binary_traits::signed_shift_result_type): Define for
3743 offset_int << HOST_WIDE_INT, etc.
3744 (generic_wide_int::operator <<=): Define for all types and use
3745 wi::lshift instead of <<.
3746 (wi::hwi_with_prec): Add a default constructor.
3747 (wi::ints_for): New class.
3748 (operator <<): Define for all wide-int types.
3749 (operator /): New function.
3750 (operator %): Likewise.
3751 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
3752 (ASSERT_MAYBE_NE_AT): New macros.
3754 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
3755 Dominik Vogt <vogt@linux.vnet.ibm.com>
3758 * emit-rtl.c (init_emit): Remove ??? comment.
3759 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3760 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
3761 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
3762 alignment of 3 virtual registers to BITS_PER_WORD.
3764 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
3766 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
3768 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
3769 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
3771 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3774 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
3776 2017-12-13 Jakub Jelinek <jakub@redhat.com>
3778 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
3779 with the exception of debug begin stmt markers appear before
3783 * final.c (rest_of_handle_final): Call variable_tracking_main only
3784 if !flag_var_tracking.
3786 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3787 Jakub Jelinek <jakub@redhat.com>
3791 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3792 labels debug stmts that can only appear after labels.
3794 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3796 PR rtl-optimization/82398
3797 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3798 EXPR_USEFULNESS in priority comparison.
3800 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3802 PR rtl-optimization/83393
3803 * combine.c (move_deaths): If reg_stat points to a too new insn in
3804 last_death, do not use it: find the proper insn instead.
3806 2017-12-12 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3809 * config/rs6000/rs6000-string.c (expand_block_compare,
3810 expand_strn_compare): Fix set_mem_size() calls.
3812 2017-12-12 Jeff Law <law@redhat.com>
3814 PR tree-optimization/83298
3815 PR tree-optimization/83362
3816 PR tree-optimization/83383
3817 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3818 push_value_range a public interface. Add new argument to
3819 record_ranges_from_stmt.
3820 * gimple-ssa-evrp-analyze.c
3821 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3822 Update comments. Handle recording temporary equivalences.
3823 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3824 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3825 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3826 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3827 gimple-ssa-evrp-analyze.h.
3828 (record_temporary_equivalences_from_phis): Add new argument. When
3829 the PHI arg is an SSA_NAME, set the result's range to the range
3831 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3832 from statements too.
3833 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3834 Pass it down to children as needed.
3835 (thread_outgoing_edges): Likewise.
3836 (thread_across_edge): Likewise. Push/pop range state as needed.
3837 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3839 2017-12-12 Julia Koval <julia.koval@intel.com>
3841 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3842 (PTA_CANNONLAKE): Remove PTA_CLWB.
3844 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3847 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3848 (vcondv2div2df): Likewise.
3849 (vconduv2dfv2di): Likewise.
3851 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3853 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3854 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3855 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3856 ATTR_MATHFN_FPROUNDING_ERRNO.
3858 2017-12-12 Richard Biener <rguenther@suse.de>
3860 PR tree-optimization/83385
3861 * graphite-scop-detection.c (get_order, order): Remove.
3862 (bb_to_rpo): New global.
3864 (build_scops): Sort pbbs in RPO order.
3866 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3868 * combine.c (simplify_set): Do not transform subregs to zero_extends
3869 if the destination mode is a vector mode.
3871 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3873 PR tree-optimization/83359
3874 * tree-cfg.h (fold_loop_internal_call): Declare.
3875 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3876 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3877 (find_loop_dist_alias): New function.
3878 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3879 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3882 PR tree-optimization/80631
3883 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3884 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3885 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3886 hardcoding zero as the value if COND_EXPR is never true. For
3887 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3888 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3889 hardcoding MAX_EXPR as the reduction operation.
3890 (is_nonwrapping_integer_induction): Allow negative step.
3891 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3892 vect_create_epilog_for_reduction, if no value is suitable, don't
3893 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3895 2017-12-12 Richard Biener <rguenther@suse.de>
3897 PR tree-optimization/81889
3898 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3899 range info from the non-wrapping IV instead of just the range
3902 2017-12-12 Julia Koval <julia.koval@intel.com>
3904 * config.gcc: Add vaesintrin.h.
3905 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3907 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3908 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3910 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3911 * config/i386/immintrin.h: Include vaesintrin.h.
3912 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3913 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3914 _mm_aesdec_epi128): New intrinsics.
3916 2017-12-12 Julia Koval <julia.koval@intel.com>
3918 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3919 OPTION_MASK_ISA_VAES_UNSET): New.
3920 (ix86_handle_option): Handle -mvaes.
3921 * config/i386/cpuid.h: Define bit_VAES.
3922 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3923 * config/i386/i386-c.c (__VAES__): New.
3924 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3925 (ix86_valid_target_attribute_inner_p): Ditto.
3926 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3927 * config/i386/i386.opt: Add -mvaes.
3928 * doc/invoke.texi: Ditto.
3930 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3932 * debug.h (gcc_debug_hooks): Add inline_entry.
3933 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3934 * debug.c (do_nothing_debug_hooks): Likewise.
3935 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3936 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3937 (dwarf2_lineno_debug_hooks): Likewise.
3939 * common.opt (gstatement-frontiers): New, setting
3940 debug_nonbind_markers_p.
3941 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3942 * toplev.c (process_options): Autodetect value for debug statement
3944 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3945 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3947 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3948 markers. Integrate source bind into debug stmt expand loop.
3949 (pass_expand::execute): Check debug marker limit. Avoid deep
3950 TER and expand debug locations for debug bind insns only.
3951 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3953 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3954 * final.c (reemit_insn_block_notes): Take current block from
3955 nonbind markers. Declare note where it's first set.
3956 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3957 begin stmt markers if enabled.
3958 (notice_source_line): Handle nonbind markers. Fail if their
3959 location is unknown or that of builtins.
3960 (rest_of_handle_final): Convert begin stmt markers to notes if
3961 var-tracking didn't run.
3962 (rest_of_clean_state): Skip begin stmt markers.
3963 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3965 * function.c (allocate_struct_function): Set begin_stmt_markers.
3966 * function.h (struct function): Add debug_marker_count counter
3967 and debug_nonbind_markers flag.
3968 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3969 * gimple-low.c (lower_function_body): Adjust
3970 debug_nonbind_markers.
3971 (lower_stmt): Drop or skip gimple debug stmts.
3972 (lower_try_catch): Skip debug stmts.
3973 * gimple.c (gimple_build_debug_begin_stmt): New.
3974 (gimple_copy): Increment debug_marker_count if copying one.
3975 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3976 * gimplify.c (rexpr_location): New.
3977 (rexpr_has_location): New.
3978 (warn_switch_unreachable_r): Handle gimple debug stmts.
3979 (shortcut_cond_r): Call expr_location.
3981 (find_goto_label): New.
3982 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3984 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3986 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3987 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3988 (LANG_HOOKS_INITIALIZER): ... this.
3989 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3990 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3992 (update_ebb_live_info): Skip debug insn markers.
3993 * lra.c (debug_insn_static_data): Rename to...
3994 (debug_bind_static_data): ... this.
3995 (debug_marker_static_data): New.
3996 (lra_set_insn_recog_data): Select one of the above depending
3998 (lra_update_isn_regno_info): Don't assume debug insns have
4000 (push_insns): Skip debug insns.
4001 * lto-streamer-in.c (input_function): Drop debug stmts
4002 depending on active options. Adjust debug_nonbind_markers.
4003 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
4004 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
4005 begin stmt marker notes.
4006 (print_insn): Likewise.
4007 * recog.c (extract_insn): Recognize rtl for debug markers.
4008 * rtl.def (DEBUG_MARKER): New.
4009 * tree-inline.c: Include params.h.
4010 (remap_gimple_stmt): Handle nonbind markers.
4011 (maybe_move_debug_stmts_to_successors): Likewise.
4012 (copy_debug_stmt): Likewise.
4013 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
4014 markers regardless of no side effects.
4015 (tsi_link_before): Don't update container's side effects when adding
4016 a begin stmt marker.
4017 (tsi_link_after): Likewise.
4018 (expr_first): Skip begin stmt markers.
4019 (expr_last): Likewise.
4020 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
4021 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
4022 Disregard nonbind markers.
4023 * tree.c (make_node_stat): Don't set side effects for begin stmt
4025 (build1_stat): Likewise.
4026 * tree.def (DEBUG_BEGIN_STMT): New.
4027 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
4028 * var-tracking.c (delete_debug_insns): Renamed to...
4029 (delete_vta_debug_insns): ... this.
4030 (reemit_marker_as_note): New.
4031 (vt_initialize): Reemit markers.
4032 (delete_vta_debug_insns): Likewise.
4033 (vt_debug_insns_local): Reemit or delete markers.
4034 (variable_tracking_main_1): Likewise.
4035 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
4036 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
4037 (gimple_debug_nonbind_marker_p): New.
4038 (gimple_build_debug_bind): Adjust.
4039 (gimple_build_debug_begin_stmt): New.
4040 * doc/invoke.texi (max-debug-marker-count): New param.
4041 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
4042 (debug_parameter_ref, debug_marker): New.
4043 (NOTE_INSN_BEGIN_STMT): New.
4044 (DEBUG_INSN): Describe begin stmt markers.
4046 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
4047 without debug insns, we wouldn't, but clean up debug insns
4048 after a control flow insn nevertheless.
4050 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
4051 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
4052 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
4053 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
4054 (rtl_tidy_fallthru_edge): Likewise.
4055 (rtl_verify_fallthru): Likewise.
4056 (rtl_verify_bb_layout): Likewise.
4057 (skip_insns_after_block): Likewise.
4058 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
4059 * dwarf2out.c: Include print-rtl.h.
4060 (dwarf2out_next_real_insn): New.
4061 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
4062 Dump debug binds in asm comments.
4063 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
4064 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
4065 callers to use gsi_start_nondebug_bb instead.
4066 (gsi_after_labels): Skip gimple debug stmts.
4067 (gsi_start_nondebug): New.
4068 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
4069 (proper_loop_form_for_interchange): Adjust.
4070 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
4071 * gimple.h (gimple_seq_last_nondebug_stmt): New.
4072 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
4073 (collect_fallthrough_labels): Likewise.
4074 (should_warn_for_implicit_fallthrough): Likewise.
4075 (warn_implicit_fallthrough_r): Likewise.
4076 (expand_FALLTHROUGH_r): Likewise.
4077 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
4078 (graphite_copy_stmts_from_block): Skip nonbind markers.
4079 * haifa-sched.c (sched_extend_bb): Skip debug insns.
4080 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
4081 * jump.c (clean_barriers): Skip debug insns.
4082 * omp-expand.c (expand_parallel_call): Skip debug insns.
4083 (expand_task_call): Likewise.
4084 (remove_exit_barrier): Likewise.
4085 (expand_omp_taskreg): Likewise.
4086 (expand_omp_for_init_counts): Likewise.
4087 (expand_omp_for_generic): Likewise.
4088 (expand_omp_for_static_nochunk): Likewise.
4089 (expand_omp_for_static_chunk): Likewise.
4090 (expand_omp_simd): Likewise.
4091 (expand_omp_taskloop_for_outer): Likewise.
4092 (expand_omp_taskloop_for_inner): Likewise.
4093 (expand_oacc_for): Likewise.
4094 (expand_omp_sections): Likewise.
4095 (expand_omp_single): Likewise.
4096 (expand_omp_synch): Likewise.
4097 (expand_omp_atomic_load): Likewise.
4098 (expand_omp_atomic_store): Likewise.
4099 (expand_omp_atomic_fetch_op): Likewise.
4100 (expand_omp_atomic_pipeline): Likewise.
4101 (expand_omp_atomic_mutex): Likewise.
4102 (expand_omp_target): Likewise.
4103 (grid_expand_omp_for_loop): Likewise.
4104 (grid_expand_target_grid_body): Likewise.
4105 (build_omp_regions_1): Likewise.
4106 * omp-low.c (check_combined_parallel): Skip debug stmts.
4107 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
4108 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
4109 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
4111 * tree-cfg.c (make_blobs_1): Skip debug stmts.
4112 (make_edges): Likewise.
4113 (cleanup_dead_labels): Likewise.
4114 (gimple_can_merge_blocks_p): Likewise.
4115 (stmt_starts_bb_p): Likewise.
4116 (gimple_block_label): Likewise.
4117 (gimple_redirect_edge_and_branch): Likewise.
4118 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
4120 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
4122 * tree-pretty-print (print_declaration): Omit initializer in slim
4124 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
4126 (eliminate_unnecessary_stmts): Stabilize block removal order.
4127 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
4128 * var-tracking.c (get_first_insn): New.
4129 (vt_emit_notes): Call it.
4130 (vt_initialize): Walk any insns before the first BB.
4131 (delete_debug_insns): Likewise.
4133 * gimple.h (enum gimple_debug_subcode): Add
4134 GIMPLE_DEBUG_BEGIN_STMT.
4135 (gimple_debug_begin_stmt_p): New.
4136 (gimple_debug_nonbind_marker_p): New.
4137 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
4138 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
4139 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
4140 * insn-notes.def (BEGIN_STMT): New.
4141 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
4142 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
4143 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
4144 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
4145 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
4146 (INSN_DEBUG_MARKER_KIND): New.
4147 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
4148 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
4149 (INSN_VAR_LOCATION_PTR): New.
4150 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
4152 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
4153 deps only if debug bind insns are enabled.
4154 (pass_expand::execute): Avoid deep TER and expand
4155 debug locations for debug bind insns only.
4156 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
4157 debug stmts special handling down to debug bind stmts.
4158 * combine.c (try_combine): Narrow debug insns special handling
4159 down to debug bind insns.
4160 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
4161 Narrow debug insns preexisting special handling down to debug
4163 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
4164 handling down to debug bind insns.
4165 * function.c (instantiate_virtual_regs): Skip debug markers,
4166 adjust handling of debug binds.
4167 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
4168 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
4169 * haifa-sched.c (schedule_insn): Narrow special handling of debug
4170 insns to debug bind insns.
4171 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
4172 special handling of debug stmts to debug bind stmts.
4173 * ipa-split.c (split_function): Likewise.
4174 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
4175 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
4177 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
4178 * regrename.c (build_def_use): Likewise.
4179 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
4180 (pass_cprop_hardreg): Narrow special casing of debug insns to
4182 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
4183 * reload1.c (reload): Likewise.
4184 * sese.c (sese_insert_phis_for_liveouts): Narrow special
4185 casing of debug stmts to debug bind stmts.
4186 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
4187 * ssa-iterators.h (num_imm_uses): Likewise.
4188 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
4189 debug stmts to debug bind stmts.
4190 * tree-inline.c (tree_function_versioning): Narrow special casing
4191 of debug stmts to debug bind stmts.
4192 * tree-loop-distribution.c (generate_loops_for_partition):
4193 Narrow special casing of debug stmts to debug bind stmts.
4194 * tree-sra.c (analyze_access_subtree): Narrow special casing
4195 of debug stmts to debug bind stmts.
4196 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
4197 stmts to debug bind stmts.
4198 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
4199 casing of debug stmts to debug bind stmts.
4200 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
4201 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
4202 casing of debug stmts to debug bind stmts.
4203 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
4205 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
4206 debug stmts to debug bind stmts.
4207 (gimple_replace_ssa_lhs): Likewise.
4208 (insert_debug_temp_for_var_def): Likewise.
4209 (insert_debug_temps_for_defs): Likewise.
4210 (reset_debug_uses): Likewise.
4211 * tree-ssanames.c (release_ssa_name_fn): Likewise.
4212 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
4213 (adjust_debug_stmts): Likewise.
4214 (adjust_phi_and_debug_stmts): Likewise.
4215 (vect_do_peeling): Likewise.
4216 * tree-vect-loop.c (vect_transform_loop): Likewise.
4217 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
4218 * var-tracking.c (adjust_mems): Narrow special casing of debug
4219 insns to debug bind insns.
4220 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
4221 (compute_bb_dataflow, vt_find_locations): Likewise.
4222 (vt_expand_loc, emit_notes_for_changes): Likewise.
4223 (vt_init_cfa_base): Likewise.
4224 (vt_emit_notes): Likewise.
4225 (vt_initialize): Likewise.
4226 (vt_finalize): Likewise.
4228 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
4229 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
4230 (next_nonnote_nondebug_insn_bb): New.
4231 (prev_nonnote_nondebug_insn_bb): New.
4232 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
4233 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
4234 (prev_nonnote_nondebug_insn_bb): Declare.
4235 (next_nonnote_nondebug_insn_bb): Declare.
4236 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
4237 * cfgrtl.c (get_last_bb_insn): Likewise.
4238 * lra.c (push_insns): Likewise.
4240 2017-12-11 David Malcolm <dmalcolm@redhat.com>
4243 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
4245 * selftest.c (selftest::test_runner::test_runner): New ctor.
4246 (selftest::test_runner::~test_runner): New dtor.
4247 * selftest.h (class selftest::test_runner): New class.
4249 2017-12-11 Carl Love <cel@us.ibm.com>
4251 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
4252 vec_extract_fp32_from_shortl]): Add #defines.
4253 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
4254 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
4255 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
4256 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
4257 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
4258 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
4259 * doc/extend.texi: Add documentation for the added builtins.
4261 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
4263 PR rtl-optimization/80693
4264 PR rtl-optimization/81019
4265 PR rtl-optimization/81020
4266 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
4267 are not mentioned in i3. Place the REG_UNUSED note on i2,
4268 possibly modified to REG_DEAD, if it did not originate in i3.
4270 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4272 * recog.c (store_data_bypass_p_1): New function.
4273 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
4274 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
4276 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
4278 PR rtl-optimization/83361
4279 * ifcvt.c (if_convert): Call fixup_partitions.
4281 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
4283 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4284 early folding of splat_u{8,16,32}.
4286 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4288 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
4289 output_operand_lossage first argument with capital letter.
4290 (aarch64_override_options): Don't start error and sorry first argument
4291 with capital letter.
4293 2017-12-11 Andi Kleen <ak@linux.intel.com>
4295 PR gcov-profile/83355
4296 * auto-profile.c (string_table::get_index_by_decl): Don't
4297 recurse when abstract origin points to itself.
4299 2017-12-11 Bin Cheng <bin.cheng@arm.com>
4301 PR tree-optimization/83320
4302 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
4303 (prune_datarefs_not_in_loop): Ditto.
4305 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
4307 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
4309 2017-12-10 Jakub Jelinek <jakub@redhat.com>
4311 PR tree-optimization/83337
4312 * gimple-loop-interchange.cc (compute_access_stride): Handle
4313 bitfield DRs properly.
4315 2017-12-09 Jakub Jelinek <jakub@redhat.com>
4317 PR tree-optimization/83338
4318 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
4319 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
4320 vector of unsigned integers to vector of signed integers.
4322 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
4324 PR rtl-optimization/83317
4325 * lra-constraints.c (process_address_1): Add insn code check.
4327 2017-12-08 Michael Matz <matz@suse.de>
4329 Fix PR tree-optimization/83323
4330 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
4331 head-controlled loops and loop BBs.
4332 * common.opt (funroll-and-jam): Remove, instead ...
4333 (floop-unroll-and-jam): ... reuse this option.
4334 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
4335 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
4336 (-floop-unroll-and-jam): ... this option.
4338 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4340 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
4341 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
4342 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
4344 2017-12-08 Richard Biener <rguenther@suse.de>
4347 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
4348 handle accesses outside of zero-sized vars.
4350 2017-12-08 Martin Jambor <mjambor@suse.cz>
4352 PR tree-optimization/83141
4353 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
4354 test for MEM_REFs implicitely changing types with padding. Remove
4356 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
4358 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4360 * config/arc/arc.c (arc_attribute_table): Add exclusions to
4362 * config/avr/avr.c (avr_attribute_table): Likewise.
4363 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
4364 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
4365 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
4366 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
4367 of struct attribute_spec.
4368 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
4370 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
4373 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
4375 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
4378 * profile-count.c (profile_count::from_gcov_type): Move from
4379 profile-count.h; handle overflow.
4380 * profile-count.h (profile_count::from_gcov_type): Move offline.
4382 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
4384 PR rtl-optimization/83304
4385 * combine.c (move_deaths): If we do not know where a register died,
4388 2017-12-08 Richard Biener <rguenther@suse.de>
4390 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
4391 Provide -fopt-info-loop feedback when we interchange in a nest.
4393 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4395 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
4396 for armv6 ARM CPU IDs.
4398 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4400 * common/config/arm/arm-common.c: Include <algorithm>.
4401 (INCLUDE_VECTOR): Define.
4402 (compare_opt_names): New function.
4403 (arm_rewrite_selected_arch): Only strip out extensions that can be
4404 expressed through -mfpu. Sort the remaining extensions
4407 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4409 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
4410 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
4411 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
4412 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
4414 2017-06-08 Tristan Gingold <gindold@adacore.com>
4417 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
4418 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
4420 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4422 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
4423 of struct attribute_spec.
4425 2017-12-08 Julia Koval <julia.koval@intel.com>
4427 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
4428 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
4429 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
4430 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
4431 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
4432 _mm_maskz_dpwssds_epi32): Ditto.
4434 2017-12-08 Richard Biener <rguenther@suse.de>
4436 PR tree-optimization/81303
4437 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
4438 conditions try to create a comparison vector type matching
4439 the data vector type.
4440 (vectorizable_condition): Adjust.
4441 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
4442 Leave invariant conditions alone in case we can vectorize those.
4444 2017-12-08 Julia Koval <julia.koval@intel.com>
4446 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
4447 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
4448 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
4449 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
4450 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
4452 2017-12-08 Julia Koval <julia.koval@intel.com>
4454 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
4455 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
4456 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
4457 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
4458 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
4459 _mm_maskz_dpbusds_epi32): New intrinsics.
4461 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4463 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
4465 (ld<bh>io_signed): Likewise.
4466 (st<bhw>io): Likewise.
4467 * config/nios2/predicates.md (ldstio_memory_operand): Allow
4468 SMALL_INT12 constant integer operand.
4470 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4472 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
4474 (nios2_large_constant_allowed): ...this. Adjust uses.
4475 (nios2_plus_symbolic_constant_p): Rename to...
4476 (nios2_plus_large_constant_p): ...this. Adjust uses.
4477 (nios2_legitimate_address_p): Correct CONST_INT handling.
4478 (nios2_symbolic_memory_operand_p): Rename to...
4479 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
4480 (nios2_large_constant_p): Check for large constant integers too.
4481 (nios2_split_large_constant): Handle constant integers.
4482 (nios2_split_symbolic_memory_operand): Rename to...
4483 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
4484 (nios2_legitimize_constant_address): Handle constant integers.
4485 (r0rel_constant_p): Handle small constant integers.
4486 (nios2_print_operand_address): Handle r0-relative integer addresses.
4487 * config/nios2/nios2-protos.h: Adjust for renamed functions.
4488 * config/nios2/nios2.md: Adjust for renamed functions.
4490 2017-12-07 Andrew Waterman <andrew@sifive.com>
4492 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
4493 (TARGET_HAVE_SRODATA_SECTION): New define.
4494 (riscv_select_section): New function.
4496 2017-12-08 Joseph Myers <joseph@codesourcery.com>
4497 Alexander Monakov <amonakov@ispras.ru>
4498 Jakub Jelinek <jakub@redhat.com>
4501 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
4503 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4505 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
4506 a bool success value. Don't call output_operand_lossage here.
4507 (aarch64_print_ldpstp_address): Return a bool success value.
4508 (aarch64_print_operand_address): Call output_addr_const if
4509 aarch64_print_address_internal fails.
4510 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
4511 'y'; call output_operand_lossage instead. Call output_operand_lossage
4512 if aarch64_print_ldpstp_address fails.
4514 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4516 * tree-vector-builder.h
4517 (tree_vector_builder::binary_encoded_nelts): Declare.
4518 * tree-vector-builder.c
4519 (tree_vector_builder::binary_encoded_nelts): New function.
4520 * fold-const.c (negate_expr_p): Likewise.
4521 (operand_equal_p, fold_checksum_tree): Likewise.
4522 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
4523 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
4524 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
4525 (uniform_vector_p): Likewise.
4526 * varasm.c (const_hash_1, compare_constant): Likewise.
4527 * tree-ssa-ccp.c: Include tree-vector-builder.h.
4528 (valid_lattice_transition): Operate directly on the VECTOR_CST
4530 * ipa-icf.c: Include tree-vector-builder.h.
4531 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
4532 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
4534 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4536 * tree.c (build_vector): Delete.
4537 * tree.h (build_vector): Make static and move into the self-testing
4540 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4542 * vector-builder.h (vector_builder::derived): New const overload.
4543 (vector_builder::elt): New function.
4544 * tree-vector-builder.h (tree_vector_builder::type): New function.
4545 (tree_vector_builder::apply_step): Declare.
4546 * tree-vector-builder.c (tree_vector_builder::apply_step): New
4548 * gimple-fold.h (tree_vector_builder): Declare.
4549 (gimple_build_vector): Take a tree_vector_builder instead of a
4550 type and vector of elements.
4551 * gimple-fold.c (gimple_build_vector): Likewise.
4552 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
4554 (get_initial_defs_for_reduction): Likewise.
4555 (vectorizable_induction): Likewise.
4557 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4559 * tree-vector-builder.h
4560 (tree_vector_builder::new_binary_operation): Declare.
4561 * tree-vector-builder.c
4562 (tree_vector_builder::new_binary_operation): New function.
4563 * fold-const.c (fold_relational_const): Use it.
4564 (const_binop): Likewise. Check that both input vectors have
4565 the same number of elements, thus excluding things like WIDEN_SUM.
4566 Check whether it is possible to operate directly on the encodings
4569 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4571 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
4572 new_unary_operation, operating only on the encoded elements.
4573 (const_unop): Likewise.
4574 (exact_inverse): Likewise.
4575 (distributes_over_addition_p): New function.
4576 (const_binop): Use tree_vector_builder and new_unary_operation
4577 for combinations of VECTOR_CST and INTEGER_CST. Operate only
4578 on the encoded elements unless the encoding is strided and the
4579 operation does not distribute over addition.
4580 (fold_convert_const): Use tree_vector_builder and
4581 new_unary_operation. Operate only on the encoded elements
4582 for truncating integer conversions, or for non-stepped encodings.
4584 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4586 * config/sparc/sparc.c: Include tree-vector-builder.h.
4587 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
4588 * expmed.c: Include tree-vector-builder.h.
4589 (make_tree): Use tree_vector_builder instead of build_vector.
4590 * fold-const.c: Include tree-vector-builder.h.
4591 (const_binop): Use tree_vector_builder instead of build_vector.
4592 (const_unop): Likewise.
4593 (native_interpret_vector): Likewise.
4594 (fold_vec_perm): Likewise.
4595 (fold_ternary_loc): Likewise.
4596 * gimple-fold.c: Include tree-vector-builder.h.
4597 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
4599 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
4600 (simplify_vector_constructor): Use tree_vector_builder instead
4602 * tree-vect-generic.c: Include tree-vector-builder.h.
4603 (add_rshift): Use tree_vector_builder instead of build_vector.
4604 (expand_vector_divmod): Likewise.
4605 (optimize_vector_constructor): Likewise.
4606 * tree-vect-loop.c: Include tree-vector-builder.h.
4607 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
4608 of build_vector. Explicitly use a stepped encoding for
4610 * tree-vect-slp.c: Include tree-vector-builder.h.
4611 (vect_get_constant_vectors): Use tree_vector_builder instead
4613 (vect_transform_slp_perm_load): Likewise.
4614 (vect_schedule_slp_instance): Likewise.
4615 * tree-vect-stmts.c: Include tree-vector-builder.h.
4616 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
4617 (vect_gen_perm_mask_any): Likewise.
4618 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
4619 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
4621 (build_vector_from_val): Likewise. Explicitly use a duplicate
4624 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
4626 * doc/generic.texi (VECTOR_CST): Describe new representation of
4628 * vector-builder.h: New file.
4629 * tree-vector-builder.h: Likewise.
4630 * tree-vector-builder.c: Likewise.
4631 * Makefile.in (OBJS): Add tree-vector-builder.o.
4632 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
4633 * tree-core.h (tree_base): Add a vector_cst field to the u union.
4634 (tree_vector): Change the number of elements to
4635 vector_cst_encoded_nelts.
4636 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
4637 (VECTOR_CST_ELTS): Delete.
4638 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
4639 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
4640 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
4641 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
4642 (VECTOR_CST_ENCODED_ELT): Likewise.
4643 (vector_cst_encoded_nelts): New function.
4644 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4645 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4646 (vector_cst_int_elt, vector_cst_elt): Declare.
4647 * tree.c: Include tree-vector-builder.h.
4648 (tree_code_size): Abort if passed VECTOR_CST.
4649 (tree_size): Update for new VECTOR_CST layout.
4650 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4651 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4652 (build_vector): Use tree_vector_builder.
4653 (vector_cst_int_elt, vector_cst_elt): New functions.
4654 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
4655 encoded elements and then create the vector in the canonical form.
4656 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
4657 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
4658 (tree_c_tests): Call test_vector_cst_patterns.
4659 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
4661 (hash_tree): Likewise.
4662 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
4663 (streamer_write_tree_header): Likewise.
4664 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
4665 (streamer_alloc_tree): Likewise. Update call to make_vector.
4666 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
4668 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4670 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
4671 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
4672 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
4674 2017-12-07 Bin Cheng <bin.cheng@arm.com>
4675 Richard Biener <rguenther@suse.de>
4677 PR tree-optimization/81303
4678 * Makefile.in (gimple-loop-interchange.o): New object file.
4679 * common.opt (floop-interchange): Reuse the option from graphite.
4680 * doc/invoke.texi (-floop-interchange): Ditto. New document for
4681 -floop-interchange and mention it for -O3.
4682 * opts.c (default_options_table): Enable -floop-interchange at -O3.
4683 * gimple-loop-interchange.cc: New file.
4684 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
4685 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
4686 * passes.def (pass_linterchange): New pass.
4687 * timevar.def (TV_LINTERCHANGE): New time var.
4688 * tree-pass.h (make_pass_linterchange): New declaration.
4689 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
4690 interchange. Record IV before/after increment in new parameters.
4691 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
4692 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
4694 (check_reduction_path): ...New function here.
4695 * tree-vectorizer.h (check_reduction_path): New declaration.
4697 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
4700 PR rtl-optimization/80818
4701 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
4702 always early clobbered.
4703 * lra-lives.c (process_bb_lives): Check input hard regs for early
4704 clobbered non-operand hard reg.
4706 2017-12-07 Jakub Jelinek <jakub@redhat.com>
4709 * tree-cfg.c (verify_gimple_assign_binary): Don't require
4710 types_compatible_p, just that TYPE_MODE is the same.
4712 2017-12-07 Martin Sebor <msebor@redhat.com>
4715 * attribs.c (empty_attribute_table): Initialize new member of
4716 struct attribute_spec.
4717 (decl_attributes): Add argument. Handle mutually exclusive
4718 combinations of attributes.
4719 (selftests::test_attribute_exclusions): New function.
4720 (selftests::attribute_c_tests): Ditto.
4721 * attribs.h (decl_attributes): Add default argument.
4722 * selftest.h (attribute_c_tests): Declare.
4723 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
4724 * tree-core.h (attribute_spec::exclusions, exclude): New type and
4726 * doc/extend.texi (Common Function Attributes): Update const and pure.
4727 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
4728 of struct attribute_spec.
4729 * config/arc/arc.c (arc_attribute_table): Same.
4730 * config/arm/arm.c (arm_attribute_table): Same.
4731 * config/avr/avr.c ( avr_attribute_table): Same.
4732 * config/bfin/bfin.c (bfin_attribute_table): Same.
4733 * config/cr16/cr16.c (cr16_attribute_table): Same.
4734 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
4735 * config/h8300/h8300.c (h8300_attribute_table): Same.
4736 * config/i386/i386.c (ix86_attribute_table): Same.
4737 * config/ia64/ia64.c (ia64_attribute_table): Same.
4738 * config/m32c/m32c.c (m32c_attribute_table): Same.
4739 * config/m32r/m32r.c (m32r_attribute_table): Same.
4740 * config/m68k/m68k.c (m68k_attribute_table): Same.
4741 * config/mcore/mcore.c (mcore_attribute_table): Same.
4742 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
4743 * config/mips/mips.c (mips_attribute_table): Same.
4744 * config/msp430/msp430.c (msp430_attribute_table): Same.
4745 * config/nds32/nds32.c (nds32_attribute_table): Same.
4746 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
4747 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
4748 * config/rl78/rl78.c (rl78__attribute_table): Same.
4749 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
4750 * onfig/rx/rx.c (rx_attribute_table): Same.
4751 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
4752 * config/sh/sh.c (sh_attribute_table): Same.
4753 * config/sparc/sparc.c (sparc_attribute_table): Same.
4754 * config/spu/spu.c (spu_attribute_table): Same.
4755 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
4756 * config/v850/v850.c (v850_attribute_table): Same.
4757 * config/visium/visium.c (visium_attribute_table): Same.
4759 2017-12-07 Tamar Christina <tamar.christina@arm.com>
4762 * config/arm/arm.c (INCLUDE_STRING): Define.
4763 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
4764 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
4767 2017-12-07 Michael Matz <matz@suse.de>
4769 Add unroll and jam pass
4771 * gimple-loop-jam.c: New file.
4772 * Makefile.in (OBJS): Add gimple-loop-jam.o.
4773 * common.opt (funroll-and-jam): New option.
4774 * opts.c (default_options_table): Add unroll-and-jam at -O3.
4775 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
4776 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
4777 * passes.def: Add pass_loop_jam.
4778 * timevar.def (TV_LOOP_JAM): Add.
4779 * tree-pass.h (make_pass_loop_jam): Declare.
4780 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
4781 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
4782 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
4783 to flow_loop_tree_node_add.
4784 (duplicate_subloops, copy_loops_to): Append to sibling list.
4785 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
4786 * doc/invoke.texi (-funroll-and-jam): Document new option.
4787 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
4789 2017-12-07 Richard Biener <rguenther@suse.de>
4791 PR tree-optimization/83296
4792 PR tree-optimization/67769
4793 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4794 flow sensitive info in an unrelated BB.
4795 (value_replacement): Use reset_flow_sensitive_info.
4796 (minmax_replacement): Reset flow sensitive info on the def
4797 we move. Do not reset flow sensitive info in the whole BB
4798 we move the stmt to.
4799 (abs_replacement): Likewise.
4801 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4804 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4805 rs6000_cpu to the given -mcpu=, or to the default processor.
4807 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4809 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4810 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4811 instead of rs6000_cpu_attr.
4812 (rs6000_adjust_cost): Ditto.
4813 (is_microcoded_insn): Ditto.
4814 (rs6000_adjust_priority): Ditto.
4815 (rs6000_issue_rate): Ditto.
4816 (rs6000_use_sched_lookahead): Ditto.
4817 (rs6000_use_sched_lookahead_guard): Ditto.
4818 (rs6000_sched_reorder): Ditto.
4819 (force_new_group): Ditto.
4820 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4821 (group_ending_nop): Ditto.
4823 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4825 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4826 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4827 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4828 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4830 (rs6000_reassociation_width): Ditto.
4831 (rs6000_emit_epilogue): Ditto.
4832 (rs6000_adjust_cost): Ditto.
4833 (is_microcoded_insn): Ditto.
4834 (is_cracked_insn): Ditto.
4835 (rs6000_adjust_priority): Ditto.
4836 (rs6000_sched_reorder): Ditto.
4837 (rs6000_sched_reorder2): Ditto.
4838 (insn_must_be_first_in_group): Ditto.
4839 (insn_must_be_last_in_group): Ditto.
4840 (rs6000_register_move_cost): Ditto.
4841 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4844 2017-12-07 Julia Koval <julia.koval@intel.com>
4846 * config.gcc: Add vaesintrin.h.
4847 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4849 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4850 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4852 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4853 * config/i386/immintrin.h: Include vaesintrin.h.
4854 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4855 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4856 _mm_aesdec_epi128): New intrinsics.
4858 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4860 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4861 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4862 c-family/c-spellcheck.cc.
4863 (best_macro_match::best_macro_match): Likewise.
4865 (struct edit_distance_traits<cpp_hashnode *>): Move to
4866 c-family/c-spellcheck.h.
4867 (class best_macro_match): Likewise.
4869 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4871 PR tree-optimization/83293
4872 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4873 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4874 might insert into empty bb.
4877 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4878 simplify for plus with :c added, and pointer_plus without that.
4879 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4880 with undefined overflow and the conversion is not widening,
4881 perform negation in utype and only convert to type afterwards.
4882 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4883 simplify for plus with :c added, and pointer_plus without that.
4884 If type is integral with undefined overflow and the conversion is
4885 not widening, perform minus in utype and only convert to type
4886 afterwards. Move the last pointer_diff_expr simplify into the
4889 2017-12-06 Martin Sebor <msebor@redhat.com>
4891 PR tree-optimization/82646
4892 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4893 strncpy, not maxlen.
4895 2017-12-06 Martin Sebor <msebor@redhat.com>
4897 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4900 PR tree-optimization/83075
4901 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4902 strncat/strncpy don't change length of source string.
4904 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4907 2017-11-29 Martin Aberg <maberg@gaisler.com>
4909 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4910 to prevent b2bst errata sequence.
4911 (sqrtdf2_fix): Likewise.
4913 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4915 PR tree-optimization/81945
4916 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4917 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4918 to dest_cfun has orig_loop_num set, either remap it to the new
4919 loop number if the loop got moved too, or clear it.
4921 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4923 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4925 (thunderx2t99_nothing): New insn reservation.
4926 (thunderx2t99_mrs): New insn reservation.
4927 (thunderx2t99_multiple): New insn reservation.
4928 (thunderx2t99_alu_basi): Add bfx to reservation.
4929 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4931 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4934 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4936 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4938 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4939 tree-ssa-pre.c::remove_dead_inserted_code.
4940 * tree-ssa-dce.h: New file.
4941 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4942 (remove_dead_inserted_code): Move and rename to function
4943 tree-ssa-dce.c::simple_dce_from_worklist.
4944 (pass_pre::execute): Update use.
4946 2017-12-05 Richard Biener <rguenther@suse.de>
4948 PR tree-optimization/83277
4949 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4950 to code-gen liveout vars.
4952 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4954 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4955 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4956 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4959 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4961 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4963 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4964 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4967 2017-12-05 Richard Biener <rguenther@suse.de>
4969 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4970 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4971 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4972 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4974 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4976 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4978 2017-12-05 Martin Liska <mliska@suse.cz>
4979 Jakub Jelinek <jakub@redhat.com>
4981 * doc/invoke.texi: Document the options.
4982 * flag-types.h (enum sanitize_code): Add
4983 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4984 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4985 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4986 * opts.c: Define new sanitizer options.
4987 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4988 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4990 2017-12-05 Julia Koval <julia.koval@intel.com>
4992 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4993 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4994 (ix86_handle_option): Handle -mavx512vnni.
4995 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4996 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4997 * config/i386/i386-c (__AVX512VNNI__): New.
4998 * config/i386/i386.c (ix86_target_string): Handle new option.
4999 (ix86_valid_target_attribute_inner_p): Handle new option.
5000 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
5001 * config/i386/i386.opt (mavx512vnni): New option.
5003 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
5006 * athlon.md: Disable for generic.
5007 * haswell.md: Enable for generic.
5008 * i386.c (ix86_sched_init_global): Add core hooks for generic.
5009 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
5011 (ix86_adjust_cost): Move generic to haswell path.
5013 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
5015 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
5016 instead of MEM_P in a couple more places. Fix formatting issues.
5018 2017-12-04 Jim Wilson <jimw@sifive.com>
5020 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
5021 instead of GP_REG_LAST-1.
5022 (riscv_adjust_libcall_cfi_prologue): Likewise.
5023 (riscv_adjust_libcall_cri_epilogue): Likewise.
5024 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
5027 2017-12-04 Luis Machado <luis.machado@linaro.org>
5029 * ipa-pure-const.c (check_decl): Add missing newline.
5030 (state_from_flags): Likewise.
5032 2017-12-04 Jeff Law <law@redhat.com>
5034 PR tree-optimizatin/78496
5035 * gimple-ssa-evrp-analyze.h
5036 (evrp_range_analyzer::get_vr_values): Simplify.
5037 * gimple-ssa-evrp-analyze.c: Corresponding changes.
5038 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
5039 and gimple-ssa-evrp-analyze.h.
5040 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
5041 (simplify_stmt_for_jump_threading): Copy a blob of code from
5042 tree-vrp.c to use ranges to simplify statements.
5043 (dom_opt_dom_walker::before_dom_children): Call
5044 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
5045 (dom_opt_dom_walker::after_dom_children): Similarly for
5046 evrp_range_analyzer::leave.
5047 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
5050 * gimple-ssa-evrp-analyze.c
5051 (evrp_range_analyzer::extract_range_from_stmt): Always use
5052 vr_values::update_value_range so preexisting range info is
5053 medged with new range info, even if the new range is VR_VARYING.
5055 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
5057 * combine.c: Adjust comment.
5058 (use_crosses_set_p): Delete.
5059 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
5060 (try_combine): Ditto.
5062 2017-12-04 Richard Biener <rguenther@suse.de>
5064 PR tree-optimization/83255
5065 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
5066 Re-add zero-iteration check.
5068 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
5070 PR rtl-optimization/83245
5071 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
5072 hard registers as earlyclobber, also if not in an asm.
5074 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
5078 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5081 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
5082 rs6000_cpu based on cpu_index, not tune_index.
5084 2017-12-04 Richard Biener <rguenther@suse.de>
5086 PR tree-optimization/83238
5087 * graphite-scop-detection.c (scop_detection::merge_sese): Make
5088 code match comment, rejecting invalid SESE regions.
5090 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
5092 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
5093 require base operand is a REG_POINTER prior to reload on targets
5094 with non-equivalent space registers.
5096 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
5098 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
5099 (update_specialized_profile): Fix updating of counts.
5100 (perhaps_add_new_callers): Likewise.
5102 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
5105 * x86-tune.def: Remove obsolette FIXMEs.
5106 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
5107 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
5108 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
5110 (X86_TUNE_PAD_RETURNS): Disable for generic.
5112 2017-12-02 Jakub Jelinek <jakub@redhat.com>
5114 PR tree-optimization/83170
5115 PR tree-optimization/83241
5116 * gimple-ssa-store-merging.c
5117 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
5118 gimple_vuse (ins_stmt) in case it has changed.
5119 (imm_store_chain_info::output_merged_store): Likewise.
5121 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
5125 * tree-cfg.c (pass_warn_function_return::execute): Handle
5126 __builtin_ubsan_handle_missing_return like __builtin_unreachable
5127 with BUILTINS_LOCATION.
5131 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
5132 is BLKmode for vector field with vector raw mode, use TYPE_MODE
5133 instead of DECL_MODE.
5135 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
5136 last argument to rtx pointer.
5137 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
5138 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
5139 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
5140 depending on the chosen ISAs.
5141 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
5142 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
5143 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
5145 * config/i386/sse.md (mov<mode>_internal): Likewise.
5146 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
5148 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5150 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
5152 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5154 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
5155 parameters from prototype.
5156 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
5157 parameters. Don't print a comment.
5158 (emit_fusion_gpr_load): Adjust.
5159 (emit_fusion_load_store): Adjust.
5160 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
5161 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
5162 comment on the second line.
5164 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5167 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
5168 rs6000_cpu based on cpu_index, not tune_index.
5170 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5172 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
5173 which_alternative instead of which_alternative + 1.
5174 (output_asm_insn): Print an extra tab if the template is short.
5176 2017-12-01 Jim Wilson <jimw@sifive.com>
5178 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
5180 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
5181 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
5183 * doc/tm.texi: Regenerate.
5185 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
5188 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
5189 whether we can allocate pseudos before trying to fix an address.
5190 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
5191 memory address is indexed or indirect.
5192 (floatuns_<mode>si2_hw2): Likewise.
5194 2017-12-01 Jason Merrill <jason@redhat.com>
5196 * Makefile.in (TAGS): Add c-family/*.cc.
5198 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
5200 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
5201 (call_value_insn): Likewise.
5202 (sibcall_insn): Likewise.
5203 (sibcall_value_insn): Likewise.
5204 (movsi_aarch64): Likewise.
5205 (movdi_aarch64): Likewise.
5206 (add_losym_): Likewise.
5207 (ldr_got_small_): Likewise.
5208 (ldr_got_small_sidi): Likewise.
5209 (ldr_got_small_28k_): Likewise.
5210 (ldr_got_small_28k_sidi): Likewise.
5211 * config/aarch64/aarch64.c (aarch64_print_address_internal):
5212 Move output_addr_const to symbolic case. Add error check.
5214 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5216 * config/s390/predicates.md (plus16_Q_operand): New predicate.
5217 * config/s390/s390.md: Disable MVC merging peephole if it would
5218 disable operand forwarding.
5219 (new peephole2): Split MVCs if it would turn them into up to 2
5222 2017-12-01 Richard Biener <rguenther@suse.de>
5224 PR tree-optimization/83232
5225 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
5226 detection of same access. Instead of breaking the group here
5227 do not consider the duplicate. Add comment explaining real fix.
5229 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
5231 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
5233 2017-12-01 Sudakshina Das <sudi.das@arm.com>
5235 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
5237 2017-12-01 Jakub Jelinek <jakub@redhat.com>
5239 * function.h (struct function): Remove cilk_frame_decl,
5240 is_cilk_function and calls_cilk_spawn fields.
5241 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
5243 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
5244 cilk_elemental field.
5245 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
5246 * target.def: Adjust comment.
5247 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
5248 Don't test cilk_elemental.
5250 PR tree-optimization/83233
5251 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
5252 bswap_stat name for the struct.
5255 * tree.h (SWITCH_BREAK_LABEL_P): Define.
5256 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
5257 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
5258 SWITCH_BREAK_LABEL_P set on the label.
5259 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
5260 added for default case if it was missing and not all cases covered.
5261 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
5262 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
5264 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
5265 to avoid -Wimplicit-fallthrough warning.
5266 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
5267 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
5269 PR tree-optimization/83221
5270 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
5272 (init_reassoc): Formatting fix.
5275 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
5276 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
5280 * tree-cfg.c: Include asan.h.
5281 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
5283 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5285 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
5287 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5289 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
5290 vpcmpeqd instruction.
5292 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5294 * config/i386/i386.c (standard_sse_constant_opcode): Fix
5295 registers type for 128bit mode.
5297 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5299 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
5301 * gengtype-state.c (read_state_pair): Likewise.
5302 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
5303 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
5304 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
5305 * ifcvt.c (noce_try_store_flag_constants): Likewise.
5306 * tree-ssa-ccp.c (ccp_finalize): Likewise.
5307 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
5308 * builtins.c (fold_builtin_3): Likewise.
5309 * graphite-scop-detection.c
5310 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
5311 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
5314 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
5317 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
5318 parameter. Return true if ignore_decl_rtl_set_p is true and other
5319 conditions are satisfied.
5320 * asan.h (asan_protect_global): Add new parameter.
5321 * varasm.c (categorize_decl_for_section): Pass true as second parameter
5322 to asan_protect_global calls.
5324 2017-11-30 Jim Wilson <jimw@sifive.com>
5326 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
5327 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
5328 -mstrict-align, add info on default value. Delete redundant lines for
5329 -mabi. Add missing -mexplicit-relocs docs.
5331 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5333 * config/arc/arc.md (trap): New pattern.
5335 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5337 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
5338 instruction to end into a delay slot.
5339 * config/arc/arc.md (cond_delay_insn): Check if the instruction
5340 can be placed into a delay slot against reg_note.
5342 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5344 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
5345 labels number of usages.
5347 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5349 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
5351 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
5353 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5356 * internal-fn.c (expand_mul_overflow): Optimize unsigned
5357 multiplication by power of 2 constant into two shifts + comparison.
5359 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
5362 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
5364 2017-11-30 Richard Biener <rguenther@suse.de>
5366 PR tree-optimization/83202
5367 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
5368 allow_peel argument and guard peeling.
5369 (canonicalize_loop_induction_variables): Likewise.
5370 (canonicalize_induction_variables): Pass false.
5371 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
5372 peeling from cunrolli.
5374 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5376 * combine.c (try_combine): Print a message to dump file whenever
5377 I0, I1, or I2 cannot be combined into I3.
5379 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5381 PR rtl-optimization/83156
5382 PR rtl-optimization/82621
5383 * combine.c (try_combine): Don't split an I2 if one of the dests is
5384 set again before I3. Allow unused dests.
5386 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5388 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
5390 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
5392 PR rtl-optimization/80818
5393 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
5394 recursively. Use insn code for clobber.
5395 (lra_set_insn_recog_data): Pass the new arg to
5396 collect_non_operand_hard_regs.
5397 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
5399 (lra_update_insn_regno_info): Pass insn to
5400 add_regs_to_insn_regno_info.
5402 2017-11-29 Jim Wilson <jimw@sifive.com>
5403 Andrew Waterman <andrew@sifive.com>
5405 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
5406 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
5407 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
5408 use SINGLE_SHIFT_COST.
5409 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
5410 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
5412 2017-11-29 Julia Koval <julia.koval@intel.com>
5414 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
5415 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
5416 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
5417 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
5418 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
5419 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
5420 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
5421 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
5422 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
5423 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
5424 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
5425 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
5426 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
5427 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
5428 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
5429 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
5430 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
5431 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
5432 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
5433 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
5434 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
5435 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
5436 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
5437 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
5438 __builtin_ia32_vpshldv_v2di_mask,
5439 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
5440 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
5441 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
5443 2017-11-29 Julia Koval <julia.koval@intel.com>
5445 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
5446 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
5447 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
5448 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
5449 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
5450 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
5451 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
5452 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
5453 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
5454 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
5455 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
5456 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
5457 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
5458 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
5459 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
5460 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
5461 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
5462 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
5463 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5464 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
5465 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
5467 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5469 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
5470 movsi_pic_gotdata_op instruction as a load for the UT699 errata
5473 2017-11-29 Martin Aberg <maberg@gaisler.com>
5475 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
5476 to prevent b2bst errata sequence.
5477 (sqrtdf2_fix): Likewise.
5479 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5481 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
5482 (div_sqrt_insn_p): New function.
5483 (sparc_do_work_around_errata): Insert NOP instructions to
5484 prevent sequences that could trigger the TN-0013 errata for
5485 certain LEON3 processors.
5486 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
5487 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
5488 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
5489 (in_branch_delay): Prevent div and sqrt in delay slot if
5491 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
5493 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5495 * config/sparc/sparc.c (atomic_insn_p): New function.
5496 (sparc_do_work_around_errata): Insert NOP instructions to
5497 prevent sequences that could trigger the TN-0010 errata for
5499 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
5500 instruction referable in atomic_insns_p.
5502 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5504 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
5505 (atomic_compare_and_swap_leon3_1): Likewise.
5508 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5510 * config/sparc/sparc.c (fpop_insn_p): New function.
5511 (sparc_do_work_around_errata): Insert NOP instructions to
5512 prevent sequences that could trigger the TN-0012 errata for
5514 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
5515 * config/sparc/sparc.md (fix_gr712rc): New attribute.
5516 (in_branch_annul_delay): Prevent floating-point instructions
5517 in delay slot of annulled integer branch.
5519 2017-11-29 Richard Biener <rguenther@suse.de>
5521 PR tree-optimization/83202
5522 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
5524 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
5525 nodes vectorized to the same stmts multiple times.
5526 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
5527 (vect_analyze_slp_instance): Adjust.
5528 (scalar_stmts_to_slp_tree_map_t): New typedef.
5529 (vect_schedule_slp_instance): Add a map recording the SLP node
5530 representing the vectorized stmts for a set of scalar stmts.
5531 Avoid code-generating redundancies.
5532 (vect_schedule_slp): Allocate map and pass it down.
5534 2017-11-29 Nathan Sidwell <nathan@acm.org>
5537 * tree.c (build_complex_type): Fix canonicalization. Only fill in
5540 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
5542 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
5543 cases for printing LDP/STP memory addresses.
5544 (aarch64_print_address_internal): Renamed from
5545 aarch64_print_operand_address, added parameter, add Pmode check.
5546 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
5547 (aarch64_print_operand_address): Indirect to
5548 aarch64_print_address_internal.
5549 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
5550 'y' operand output specifier.
5552 2017-11-29 Jakub Jelinek <jakub@redhat.com>
5555 * tree.c (build_simple_mem_ref_loc): Handle
5556 get_addr_base_and_unit_offset returning a MEM_REF.
5559 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
5562 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
5565 2017-11-28 David Malcolm <dmalcolm@redhat.com>
5567 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
5569 (layout::move_to_column): Likewise.
5571 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5574 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
5575 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
5577 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5578 Martin Jambor <mjambor@suse.cz>
5581 * tree.h (expr_type_first_operand_type_p): Declare
5582 * tree.c (expr_type_first_operand_type_p): New function.
5583 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
5584 (ipa_value_from_jfunc): Adjust declaration.
5585 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
5586 Use it as result type for arithmetics, unless it is NULL in which case
5587 be more conservative.
5588 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
5589 ipa_get_jf_pass_through_result.
5590 (propagate_vals_across_pass_through): Likewise.
5591 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
5592 is to propagate_vals_across_pass_through.
5593 (propagate_constants_across_call): Pass PARM_TYPE to
5594 propagate_scalar_across_jump_function.
5595 (find_more_scalar_values_for_callers_subset): Pass parameter type to
5596 ipa_value_from_jfunc.
5597 (cgraph_edge_brings_all_scalars_for_node): Likewise.
5598 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
5599 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
5600 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
5601 target_type, pass it to ipa_value_from_jfunc.
5602 (update_indirect_edges_after_inlining): Pass parameter type to
5603 try_make_edge_direct_simple_call.
5605 2017-11-28 Jeff Law <law@redhat.com>
5607 * gimple-ssa-evrp-analyze.c
5608 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
5609 refine ranges if scev_initialized_p returns true.
5610 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
5612 2017-11-28 Julia Koval <julia.koval@intel.com>
5614 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
5615 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
5616 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
5617 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
5618 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
5619 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
5620 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
5621 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
5622 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
5623 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
5624 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
5625 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
5626 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
5627 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
5628 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
5629 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
5630 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
5631 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
5632 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
5633 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
5634 __builtin_ia32_vpshrd_v2di_mask): New builtins.
5635 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
5637 2017-11-28 Julia Koval <julia.koval@intel.com>
5639 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
5640 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
5641 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
5642 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
5643 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
5644 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
5645 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
5646 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
5647 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
5648 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
5649 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
5650 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
5651 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
5652 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
5653 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
5654 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
5655 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
5656 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
5657 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
5658 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
5659 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
5660 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
5661 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
5662 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
5663 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
5664 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
5665 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
5666 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
5667 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
5668 __builtin_ia32_vpshld_v2di_mask): New builtins.
5669 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5670 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
5672 2017-11-28 Richard Biener <rguenther@suse.de>
5674 PR tree-optimization/80776
5675 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
5677 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
5679 (evrp_range_analyzer::record_ranges_from_incoming_edges):
5680 If the incoming edge is an effective fallthru because the other
5681 edge only reaches a __builtin_unreachable () then record ranges
5682 derived from the controlling condition in SSA info.
5683 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
5684 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5686 2017-11-28 Olivier Hainque <hainque@adacore.com>
5688 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
5689 on mingw build hosts.
5691 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5694 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
5696 2017-11-28 Richard Biener <rguenther@suse.de>
5699 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
5700 copies generated from memcpy use a character array as reference
5703 2017-11-28 Julia Koval <julia.koval@intel.com>
5704 Sebastian Peryt <sebastian.peryt@intel.com>
5706 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
5707 c-family/c-cilkplus.o, c-family/array-notation-common.o,
5708 cilk-common.o, cilk.h, cilk-common.c): Remove.
5710 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
5711 * builtins.c (is_builtin_name): Remove cilkplus condition.
5712 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
5713 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
5714 cilk-builtins.def, cilkplus.def): Remove.
5715 * cif-code.def (CILK_SPAWN): Remove.
5716 * cilk-builtins.def: Delete.
5717 * cilk-common.c: Ditto.
5719 * cilkplus.def: Ditto.
5720 * config/darwin.h (fcilkplus): Delete.
5721 * cppbuiltin.c: Ditto.
5722 * doc/extend.texi: Remove cilkplus doc.
5723 * doc/generic.texi: Ditto.
5724 * doc/invoke.texi: Ditto.
5725 * doc/passes.texi: Ditto.
5726 * gcc.c (fcilkplus): Remove.
5727 * gengtype.c (cilk.h): Remove.
5728 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
5730 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
5732 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
5734 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
5735 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
5736 cilkplus conditions.
5737 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
5738 inline_read_section): Ditto.
5739 * ipa-inline-analysis.c (cilk.h): Remove.
5740 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
5741 * lto-wrapper.c (merge_and_complain, append_compiler_options,
5742 append_linker_options): Remove condition for fcilkplus.
5743 * lto/lto-lang.c (cilk.h): Remove.
5744 (lto_init): Remove condition for fcilkplus.
5745 * omp-expand.c (expand_cilk_for_call): Delete.
5746 (expand_omp_taskreg, expand_omp_for_static_chunk,
5747 expand_omp_for): Remove cilkplus
5749 (expand_cilk_for): Delete.
5750 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
5751 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
5752 execute_lower_omp, diagnose_sb_0): Ditto.
5753 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
5754 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5755 * tree-nested.c: Ditto.
5756 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
5757 (dump_generic_node): Ditto.
5758 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5759 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
5761 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
5763 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5765 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
5766 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
5767 for vector mode and !TARGET_SIMD.
5769 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5771 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
5773 * tree.h (SWITCH_LABELS): Remove.
5774 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
5775 assert SWITCH_BODY is non-NULL.
5776 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
5778 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
5780 PR tree-optimization/80788
5781 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
5782 has TREE_OVERFLOW set, call drop_tree_overflow.
5784 2017-11-28 Richard Biener <rguenther@suse.de>
5786 PR tree-optimization/83158
5787 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
5789 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5792 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5793 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5795 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5797 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5798 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5800 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5803 * varasm.c (compare_constant): Take the mode of the constants into
5804 account when comparing floating point constants.
5806 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5808 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5809 from explicit instantiation of debug_helper.
5810 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5812 2017-11-27 Richard Biener <rguenther@suse.de>
5814 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5817 2017-11-27 Richard Biener <rguenther@suse.de>
5819 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5820 (type_hash_canon): Also clear int_cst_hash_table entry for
5822 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5824 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5826 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5827 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5828 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5830 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5832 * hash-map.h (gt_cleare_cache): Avoid UB.
5834 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5836 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5837 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5838 (decide_unroll_runtime_iterations): Likewise.
5839 (decide_unroll_stupid): Likewise.
5841 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5844 * config/i386/i386.md: Add a loop with incssp.
5846 2017-11-27 Martin Jambor <mjambor@suse.cz>
5848 PR tree-optimization/81248
5849 * tree-sra.c (splice_param_accesses): Remove size check.
5850 (decide_one_param_reduction): Fix size check.
5851 * gimple-pretty-print.c (dump_profile): Silence warning.
5852 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5854 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5857 * dbxout.c (lastlineno): New variable.
5858 (dbx_debug_hooks): Use dbxout_switch_text_section as
5859 switch_text_section debug hook.
5860 (dbxout_function_end): Switch to current_function_section
5861 rather than function_section. If crtl->has_bb_partition,
5862 output just one N_FUN, depending on in_cold_section_p.
5863 (dbxout_source_line): Remember last lineno in lastlineno.
5864 (dbxout_switch_text_section): New function.
5865 (dbxout_function_decl): Adjust dbxout_block caller.
5866 (dbx_block_with_cold_children): New function.
5867 (dbxout_block): Return true if any LBRAC/RBRAC have been
5868 emitted. Use dbx_block_with_cold_children at depth == 0
5869 in second partition. Add PARENT_BLOCKNUM argument, pass
5870 it optionally adjusted to children. Output LBRAC/RBRAC
5871 around recursive call only if the block is in the current
5872 partition, if not and anything was output, emit empty
5874 * final.c (final_scan_insn): Compute cold_function_name
5875 before calling switch_text_section debug hook. Call
5876 that hook even if dwarf2out_do_frame if not emitting
5880 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5881 TREE_READONLY decls.
5883 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5885 PR rtl-optimization/82488
5886 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5888 2017-11-26 Julia Koval <julia.koval@intel.com>
5890 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5892 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5895 2017-11-26 Julia Koval <julia.koval@intel.com>
5897 * config/i386/driver-i386.c (host_detect_local_cpu):
5898 Detect skylake-avx512.
5900 2017-11-26 Julia Koval <julia.koval@intel.com>
5902 * config.gcc: Add -march=cannonlake.
5903 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5904 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5905 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5906 (PTA_CANNONLAKE): New.
5907 (processor_target_table): Add cannonlake.
5908 (ix86_option_override_internal): Ditto.
5909 (fold_builtin_cpu): Ditto.
5910 (get_builtin_code_for_version): Handle cannonlake.
5911 (M_INTEL_COREI7_CANNONLAKE): New.
5912 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5913 * doc/invoke.texi: Add -march=cannonlake.
5915 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5917 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5918 (try_init_one_plugin): Alternative implementation for MinGW.
5919 * Makefile.in (plugin_implib): New.
5920 (gengtype-lex.c): Fix broken AIX workaround.
5921 * configure: Regenerate.
5922 * doc/plugins.texi: Document support for MinGW.
5924 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5926 PR rtl-optimization/81553
5927 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5928 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5929 is a shift where C1 has different mode than the whole shift, use C1's
5930 mode for MULT rather than the shift's mode.
5933 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5934 builtins not enabled in the currently selected ISA.
5936 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5938 PR tree-optimization/71026
5939 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5940 (insert_reciprocals): Change to insert reciprocals before a division
5941 by a square and to insert the square of a reciprocal.
5942 (execute_cse_reciprocals_1): Change to consider division by a square.
5943 (register_division_in): Add importance parameter.
5945 2017-11-24 Richard Biener <rguenther@suse.de>
5947 PR tree-optimization/82402
5948 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5949 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5951 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5953 * match.pd (0-ptr): New transformation.
5955 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5958 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5959 when skipping checking.
5961 2017-11-24 Richard Biener <rguenther@suse.de>
5963 PR tree-optimization/83128
5964 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5965 (vn_reference_lookup_3): Likewise.
5967 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5970 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5971 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5974 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5975 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5976 check for the latter with gimple_call_builtin_p. Do not handle
5977 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5979 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5981 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5984 2017-11-23 Julia Koval <julia.koval@intel.com>
5986 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5987 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5988 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5989 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5990 _mm512_maskz_expandloadu_epi16): New intrinsics.
5991 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5992 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5993 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5994 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5995 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5996 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5997 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5998 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5999 _mm256_maskz_expandloadu_epi8): New intrinsics.
6000 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
6001 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
6002 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
6003 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
6004 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
6005 * config/i386/sse.md (VI248_VLBW): New iterator.
6006 (expand<mode>_mask, expand<mode>_maskz): New patterns.
6008 2017-11-23 Julia Koval <julia.koval@intel.com>
6010 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
6011 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
6012 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
6013 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
6014 _mm512_mask_compressstoreu_epi16): New.
6015 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
6016 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
6017 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
6018 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
6019 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
6020 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
6021 _mm256_mask_compressstoreu_epi8): New.
6022 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
6023 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
6024 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
6025 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
6026 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
6027 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
6028 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
6029 __builtin_ia32_compresshi128_mask,
6030 __builtin_ia32_compressstoreuqi512_mask,
6031 __builtin_ia32_compressstoreuhi512_mask,
6032 __builtin_ia32_compressstoreuqi256_mask,
6033 __builtin_ia32_compressstoreuqi128_mask,
6034 __builtin_ia32_compressstoreuhi256_mask,
6035 __builtin_ia32_compressstoreuhi128_mask): New builtins.
6036 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
6038 (ix86_expand_special_args_builtin): Handle new types.
6039 (s4fma_expand): Handle new builtin array.
6040 * config/i386/immintrin.h: Include new headers.
6041 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
6042 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
6044 2017-11-23 Jakub Jelinek <jakub@redhat.com>
6046 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
6047 fixes. Declare temp and g variables at the top in order to avoid
6048 {} in most of the cases.
6050 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
6052 * match.pd (ptr-0): New transformation.
6054 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
6056 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
6057 (struct addr_mode_cost_table): New.
6058 (struct tune_params): Add field addr_mode_costs.
6059 * config/arm/arm.c (generic_addr_mode_costs): New.
6060 (arm_slowmul_tune): Initialise addr_mode_costs field.
6061 (arm_fastmul_tune): Likewise.
6062 (arm_strongarm_tune): Likewise.
6063 (arm_xscale_tune): Likewise.
6064 (arm_9e_tune): Likewise.
6065 (arm_marvell_pj4_tune): Likewise.
6066 (arm_v6t2_tune): Likewise.
6067 (arm_cortex_tune): Likewise.
6068 (arm_cortex_a8_tune): Likewise.
6069 (arm_cortex_a7_tune): Likewise.
6070 (arm_cortex_a15_tune): Likewise.
6071 (arm_cortex_a35_tune): Likewise.
6072 (arm_cortex_a53_tune): Likewise.
6073 (arm_cortex_a57_tune): Likewise.
6074 (arm_exynosm1_tune): Likewise.
6075 (arm_xgene1_tune): Likewise.
6076 (arm_cortex_a5_tune): Likewise.
6077 (arm_cortex_a9_tune): Likewise.
6078 (arm_cortex_a12_tune): Likewise.
6079 (arm_cortex_a73_tune): Likewise.
6080 (arm_v7m_tune): Likewise.
6081 (arm_cortex_m7_tune): Likewise.
6082 (arm_v6m_tune): Likewise.
6083 (arm_fa726te_tune): Likewise.
6084 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
6086 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
6088 * config/arm/arm.c (arm_mem_costs): New function.
6089 (arm_rtx_costs_internal): Use arm_mem_costs.
6091 2017-11-23 Mark Wielaard <mark@klomp.org>
6093 * dwarf2out.c (init_sections_and_labels): Use generation to create
6094 unique ranges_section_label and ranges_base_label. Return generation.
6095 (output_rnglists): Add generation argument. Use generation to create
6096 unique ranges labels.
6097 (dwarf2out_finish): Get generation from init_sections_and_labels
6098 and pass generation to output_rnglists.
6100 2017-11-23 Mike Stump <mikestump@comcast.net>
6101 Eric Botcazou <ebotcazou@adacore.com>
6103 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
6104 * cfgloop.h (struct loop): Add unroll field.
6105 * function.h (struct function): Add has_unroll bitfield.
6106 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
6107 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
6108 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
6109 (pass_rtl_unroll_loops::gate): Likewise.
6110 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
6111 for which loop->unroll==1.
6112 (decide_unroll_constant_iterations): Use note for consistency and
6113 take loop->unroll into account. Return early if loop->unroll is set.
6114 Fix thinko in existing test.
6115 (decide_unroll_runtime_iterations): Use note for consistency and
6116 take loop->unroll into account.
6117 (decide_unroll_stupid): Likewise.
6118 * lto-streamer-in.c (input_cfg): Read loop->unroll.
6119 * lto-streamer-out.c (output_cfg): Write loop->unroll.
6120 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
6122 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
6123 (print_loop): Print loop->unroll if set.
6124 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
6125 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
6126 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
6128 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
6129 loop->unroll is set and smaller than the trip count. Otherwise bypass
6130 entirely the heuristics if loop->unroll is set. Remove dead note.
6131 Fix off-by-one bug in other note.
6132 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
6133 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
6135 (tree_unroll_loops_completely): Make static.
6136 (pass_complete_unroll::execute): Use correct type for variable.
6137 (pass_complete_unrolli::execute): Fix formatting.
6138 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
6140 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
6142 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
6143 enable when TARGET_PREFER_AVX128 is set.
6145 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6147 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
6149 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
6150 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
6152 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6154 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
6155 frequencies. Fix estimation of aggregate parameters.
6157 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6159 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
6162 2017-11-23 Tom de Vries <tom@codesourcery.com>
6164 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
6166 2017-11-23 Tom de Vries <tom@codesourcery.com>
6168 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
6171 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
6174 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
6175 sibcall_value_pcrel_fdpic): Use local variable instead of
6177 (calli_tbr_rel): Add missing operand 2.
6178 (call_valuei_tbr_rel): Add missing operand 3.
6180 2017-11-23 Jakub Jelinek <jakub@redhat.com>
6183 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
6184 bitpos/bitsize covering the whole destination, use store_expr only if
6185 the complex mode is the same. Otherwise, use expand_normal and if
6186 it returns CONCAT, subreg each part separately instead of trying to
6187 subreg the whole result.
6189 2017-11-23 Richard Biener <rguenther@suse.de>
6191 PR tree-optimization/23094
6192 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
6193 come here from walking over backedges in the first iteration.
6194 (vn_reference_lookup_3): Skip clobbers that store the same value.
6196 2017-11-23 Richard Biener <rguenther@suse.de>
6198 PR tree-optimization/81403
6199 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
6200 a block we need a leader relative to.
6201 (phi_translate_1): For nary processing require a leader from
6202 get_representative_for given we run expression simplification
6203 using match-and-simplify. Remove previous fix.
6205 2017-11-22 Jeff Law <law@redhat.com>
6207 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
6208 Use new method allocate_value_range rather than accessing the
6209 vrp_value_range_pool data member directly.
6210 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
6211 to use extract_range_from_stmt method to avoid need for
6212 extract_range_from_assignment method.
6213 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
6214 method rather than setting values_propgated data member directly.
6215 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
6216 and values propagated data members and extract_range_from_assignment
6217 method. Reorder private data members to conform to standards.
6218 Add new methods set_lattice_propagation_complete and
6219 allocate_value_range.
6221 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
6223 PR rtl-optimization/83030
6224 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
6225 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
6226 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
6227 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
6228 CROSSING_JUMP_P flag is already set before setting it.
6229 * cfgrtl.c (fixup_partition_crossing): Likewise.
6230 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
6233 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6235 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
6236 Handle the case where both arguments are using gen_const_vec_series.
6238 2017-11-22 David Malcolm <dmalcolm@redhat.com>
6241 * pretty-print.c (pp_format): Move quoting implementation to
6242 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
6243 to pass address of "quote" local.
6244 (pp_begin_quote): New function.
6245 (pp_end_quote): New function.
6246 * pretty-print.h (printer_fn): Convert penultimate param from bool
6248 (pp_begin_quote): New decl.
6249 (pp_end_quote): New decl.
6250 * tree-diagnostic.c (default_tree_printer): Convert penultimate
6251 param from bool to bool *.
6252 * tree-diagnostic.h (default_tree_printer): Likewise.
6254 2017-11-22 Jeff Law <law@redhat.com>
6256 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
6257 Initialize vr_values.
6258 (evrp_range_analyzer::try_find_new_range): Call methods attached to
6259 vr_values via vr_values class instance rather than delegators.
6260 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
6261 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
6262 (evrp_range_analyzer::push_value_range): Likewise.
6263 (evrp_range_analyzer::pop_value_range): Likewise.
6264 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
6265 most delegators. Those remaining are exposed as public interfaces.
6266 Make vr_values a pointer and private.
6267 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
6269 (evrp_range_analyzer::get_vr_value): New method.
6270 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
6271 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
6272 (class evrp_dom_walker): Attach evrp_folder class, initialize
6273 it in the ctor. Remove temporary delegators.
6274 (evrp_dom_walker::before_dom_children): Call methods in attached
6275 evrp_range_analyzer class via class instance pointer. Use
6276 free value_range_constant_singleton to remove need for
6277 op_with_constant_singleton_value delegator method. Do not
6278 create a vrp_prop class instance for every call! Narrow
6279 scope of a couple variables.
6280 (evrp_dom_walker::cleanup): Call methods in attached
6281 evrp_range_analyzer class via class instance pointer.
6282 * vr-values.h (class vr_values): Privatize many methods and
6285 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6287 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
6288 * cfgexpand.c (expand_debug_expr): Remove handling for them.
6289 * expr.c (expand_expr_real_2): Likewise.
6290 * fold-const.c (const_unop): Likewise.
6291 * optabs-tree.c (optab_for_tree_code): Likewise.
6292 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
6293 * tree-inline.c (estimate_operator_cost): Likewise.
6294 * tree-pretty-print.c (dump_generic_node): Likewise.
6295 (op_code_prio): Likewise.
6296 (op_symbol_code): Likewise.
6297 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
6298 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
6299 * internal-fn.c (direct_internal_fn_optab): New function.
6300 (direct_internal_fn_array, direct_internal_fn_supported_p
6301 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
6302 * fold-const-call.c (fold_const_reduction): New function.
6303 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
6305 * tree-vect-loop.c: Include internal-fn.h.
6306 (reduction_code_for_scalar_code): Rename to...
6307 (reduction_fn_for_scalar_code): ...this and return an internal
6309 (vect_model_reduction_cost): Take an internal_fn rather than
6311 (vect_create_epilog_for_reduction): Likewise. Build calls rather
6313 (vectorizable_reduction): Use internal functions rather than tree
6314 codes for the reduction operation. Update calls to the functions
6316 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
6317 Use calls to internal functions rather than REDUC tree codes.
6318 * config/aarch64/aarch64-simd.md: Update comment accordingly.
6320 2017-11-22 Olivier Hainque <hainque@adacore.com>
6322 * config/vxworks.c (vxworks_override_options): Pick default
6323 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
6324 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
6325 DWARF_GNAT_ENCODINGS_DEFAULT.
6326 * config/vxworksae.h: Likewise.
6328 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
6330 PR tree-optimization/83104
6331 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
6334 2017-11-22 Marek Polacek <polacek@redhat.com>
6335 H.J. Lu <hongjiu.lu@intel.com>
6336 Jason Merrill <jason@redhat.com>
6341 * calls.c (initialize_argument_information): Call
6342 warn_parameter_passing_abi target hook.
6343 (store_one_arg): Use 0 for empty record size. Don't push 0 size
6344 argument onto stack.
6345 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
6346 * common.opt: Update -fabi-version description.
6347 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
6348 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
6350 (ix86_is_empty_record): New function.
6351 (ix86_warn_parameter_passing_abi): New function.
6352 (TARGET_EMPTY_RECORD_P): Redefine.
6353 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
6354 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
6355 * doc/tm.texi: Regenerated.
6356 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
6357 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
6358 * dwarf2out.c (get_ultimate_context): Move to tree.c.
6359 * explow.c (hard_function_value): Call arg_int_size_in_bytes
6360 instead of int_size_in_bytes.
6361 * expr.c (copy_blkmode_to_reg): Likewise.
6362 * function.c (aggregate_value_p): Return 0 for empty types.
6363 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
6365 (locate_and_pad_parm): Call arg size_in_bytes instead
6367 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
6368 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
6369 * target.def (empty_record_p, warn_parameter_passing_abi): New target
6371 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
6372 (std_gimplify_va_arg_expr): Skip empty records. Call
6373 arg_size_in_bytes instead size_in_bytes.
6374 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
6375 * tree-core.h (tree_type_common): Add empty_flag.
6376 (tree_decl_common): Update comments.
6377 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
6379 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6380 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
6382 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6383 * tree.c (default_is_empty_type): New function.
6384 (default_is_empty_record): New function.
6385 (arg_int_size_in_bytes): New function.
6386 (arg_size_in_bytes): New function.
6387 (get_ultimate_context): New function.
6388 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
6389 TRANSLATION_UNIT_WARN_EMPTY_P.
6390 (default_is_empty_record, arg_int_size_in_bytes,
6391 arg_size_in_bytes, get_ultimate_context): Declare.
6393 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6395 * config/arm/arm.c (cmse_clear_registers): New function.
6396 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
6397 code by call to cmse_clear_registers.
6398 (cmse_nonsecure_entry_clear_before_return): Likewise.
6400 2017-11-22 Tamar Christina <tamar.christina@arm.com>
6402 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
6403 (vdot_s32, vdotq_s32): New.
6404 (vdot_lane_u32, vdotq_lane_u32): New.
6405 (vdot_lane_s32, vdotq_lane_s32): New.
6408 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6411 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
6412 for unsigned values with fewer HWIs than the precision.
6413 (test_overflow): New function.
6414 (wide_int_cc_tests): Call it.
6416 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6418 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
6419 has the same mode class as Pmode.
6421 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6423 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
6424 padding_bits_to_clear_ptr.
6425 (cmse_nonsecure_entry_clear_before_return): Likewise.
6427 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6429 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
6430 auto_sbitap instead of integer bitfield to control register needing
6433 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6435 PR tree-optimization/83044
6436 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
6437 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
6438 that need the upper bound. Subtract offset from
6439 get_addr_base_and_unit_offset only if positive and subtract it
6440 before division by eltsize rather than after it.
6443 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
6444 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
6445 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
6449 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
6450 expand_binop with *mul_widen_optab, make sure at least one of the
6451 operands doesn't have VOIDmode.
6454 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
6456 PR rtl-optimization/82044
6457 PR tree-optimization/82042
6458 * dse.c (record_store): Check for overflow.
6459 (check_mem_read_rtx): Properly check for overflow if width == -1, call
6460 add_wild_read instead of clear_rhs_from_active_local_stores on
6461 overflow and log it into dump_file.
6463 2017-11-22 Richard Biener <rguenther@suse.de>
6465 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
6466 fake edges to exit when looking for a place to insert.
6467 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
6469 (insert_into_preds_of_block): Commit edge insertion immediately,
6470 assert that doesn't require new BBs.
6471 (fini_pre): Release expressions.
6472 (pass_pre::execute): Shuffle things around a bit, if the fn
6473 is too large do not compute AVAIL either as this is really the
6476 2017-11-22 Richard Biener <rguenther@suse.de>
6478 PR tree-optimization/83089
6479 * tree-if-conv.c (pass_if_conversion::execute): If anything
6480 changed reset SCEV and free the number of iteration estimates.
6482 2017-11-21 Martin Sebor <msebor@redhat.com>
6484 PR tree-optimization/82945
6485 * calls.h (warn_nonstring_bound): Remove unused function.
6487 2017-11-21 Martin Sebor <msebor@redhat.com>
6489 PR tree-optimization/82945
6490 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
6491 * calls.h (maybe_warn_nonstring_arg): Declare new function.
6492 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
6494 (initialize_argument_information): Call maybe_warn_nonstring_arg.
6495 * calls.h (get_attr_nonstring_decl): Declare new function.
6496 * doc/extend.texi (attribute nonstring): Update.
6497 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
6498 get_attr_nonstring_decl and handle it.
6499 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
6500 detection of nul-termination.
6501 (strlen_to_stridx): Change to a pointer.
6502 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
6503 (pass_strlen::execute): Same.
6505 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
6507 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
6508 for the new option -mprefer-vector-width=[none|128|256|512].
6509 * config/i386/i386.c (ix86_target_string): remove old style options
6510 -mprefer-avx256 and make -mprefer-avx128 as alias.
6511 (ix86_option_override_internal): Apply defaults for the
6512 -mprefer-vector-width=[128|256] option.
6513 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
6514 Implement macros to work with -mprefer-vector-width=.
6515 * config/i386/i386.opt: Implemented option
6516 -mprefer-vector-width=[none|128|256|512].
6517 * doc/invoke.texi: Documentation for
6518 -mprefer-vector-width=[none|128|256|512].
6520 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
6522 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
6524 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
6526 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
6528 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
6530 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
6532 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
6533 * expr.c (expand_expr_real_2): Likewise.
6534 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
6535 fold_binary_loc): Likewise.
6536 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
6537 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
6538 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
6539 MINUS_EXPR transformations.
6540 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
6541 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
6542 * tree-inline.c (estimate_operator_cost): Likewise.
6543 * tree-pretty-print.c (dump_generic_node, op_code_prio,
6544 op_symbol_code): Likewise.
6545 * tree-vect-stmts.c (vectorizable_operation): Likewise.
6546 * vr-values.c (extract_range_from_binary_expr): Likewise.
6547 * varasm.c (initializer_constant_valid_p_1): Likewise.
6548 * tree.def: New tree code POINTER_DIFF_EXPR.
6550 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
6552 * config/i386/i386.md (*bswap<mode>2_movbe): Add
6553 integer suffix to movbe mnemonic.
6554 (*bswaphi2_movbe): Ditto.
6555 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
6557 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6560 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
6561 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
6562 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
6564 2017-11-21 Martin Liska <mliska@suse.cz>
6566 * tree-inline.c (expand_call_inline): Remove not needed
6569 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
6570 Jakub Jelinek <jakub@redhat.com>
6573 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
6574 Remove static keyword from f variable.
6576 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6578 PR tree-optimization/83086
6579 * gimple-ssa-store-merging.c
6580 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
6581 rather than n.base_addr.
6583 2017-11-21 Martin Liska <mliska@suse.cz>
6585 PR rtl-optimization/82044
6586 PR tree-optimization/82042
6587 * dse.c (check_mem_read_rtx): Check for overflow.
6589 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6591 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
6594 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6596 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
6597 and its corresponding call arg location note.
6599 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6600 Andrew Burgess <andrew.burgess@embecosm.com>
6602 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
6604 (arc_return_slot_offset): Likewise.
6605 (arc_eh_return_address_location): New declaration.
6606 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
6607 (MUST_SAVE_REGISTER): Add exception handler case.
6608 (MUST_SAVE_RETURN_ADDR): Likewise.
6609 (arc_frame_pointer_required): Likewise.
6610 (arc_frame_pointer_needed): New function.
6611 (arc_compute_frame_size): Changed.
6612 (arc_expand_prologue): Likewise.
6613 (arc_expand_epilogue): Likewise.
6614 (arc_initial_elimination_offset): Likewise.
6615 (arc_return_slot_offset): Delete.
6616 (arc_eh_return_address_location): New function.
6617 (arc_builtin_setjmp_frame_value): Likewise.
6618 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
6619 (EH_RETURN_STACKADJ_RTX): Define.
6620 (EH_RETURN_HANDLER_RTX): Likewise.
6621 * config/arc/arc.md (eh_return): Delete.
6623 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6625 * print-tree.h (debug_vec_tree): Remove prototype.
6626 * gdbinit.in (pvt): Remove macro.
6628 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6630 PR tree-optimization/83047
6631 * gimple-ssa-store-merging.c
6632 (imm_store_chain_info::output_merged_store): If the loads with the
6633 same vuse are in different basic blocks, for load_gsi pick a load
6634 location that is dominated by the other loads.
6637 * config/i386/i386.c (ix86_memmodel_check): Start
6638 -Winvalid-memory-model diagnostics with lowercase letter.
6641 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
6642 set in_cold_section_p to the partition containing loc_list->first.
6643 When seeing loc_list->last_before_switch node, update secname and
6644 perform range_across_switch second partition handling only after that.
6647 * run-rtl-passes.c: Include debug.h.
6648 (run_rtl_passes): Call debug_hooks->assembly_start.
6649 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
6653 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
6654 OPTAB_DIRECT in calls to expand_simple_binop.
6656 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6659 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
6661 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6664 * spellcheck.h (best_match::blithely_get_best_candidate): New
6667 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6669 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
6670 with lower case letter.
6672 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
6674 * config/i386/i386.md (bswaphi2): New expander.
6675 (*bswaphi2_movbe): New insn pattern.
6676 (bswaphi -> rorhi pepehole2): New peephole pattern.
6678 2017-11-20 Jeff Law <law@redhat.com>
6680 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
6681 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
6682 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
6683 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
6684 gimple-ssa-evrp-analyze.h.
6686 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
6687 set BB_VISITED here.
6688 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
6690 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6692 * tree-predcom.c: Add general comment on Store-Store chains.
6693 (split_data_refs_to_components): Postpone clearing eliminate_store_p
6695 (get_chain_last_ref_at): Rename into...
6696 (get_chain_last_write_at): ...this.
6697 (get_chain_last_write_before_load): New function.
6698 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
6699 CT_STORE_STORE when write reference is added.
6700 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
6701 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
6702 (initialize_root_vars_store_elim_1): Ditto.
6703 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
6704 definition is created.
6705 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
6706 chain by replacing it with dominant stored value.
6708 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6710 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
6712 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
6714 * vr-values.c (extract_range_from_binary_expr): Use a full range
6717 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
6719 * config/arm/arm.md (R4_REGNUM): Define constant.
6720 (nonsecure_call_internal): Remove r4 clobber.
6721 (nonsecure_call_value_internal): Likewise.
6722 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
6723 clobber and resequence match_operands.
6724 (nonsecure_call_value_reg_thumb1_v5): Likewise.
6725 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
6726 (nonsecure_call_value_reg_thumb2): Likewise.
6728 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6730 PR tree-optimization/78821
6731 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
6732 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
6733 to the address of the base rather than the base itself.
6734 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
6735 (find_bswap_or_nop_finalize): New function.
6736 (find_bswap_or_nop): Use it.
6737 (bswap_replace): Return a tree rather than bool, change first
6738 argument from gimple * to gimple_stmt_iterator, allow inserting
6739 into an empty sequence, allow ins_stmt to be NULL - then emit
6740 all stmts into gsi. Fix up MEM_REF address gimplification.
6741 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
6742 (struct store_immediate_info): Add N and INS_STMT non-static
6744 (store_immediate_info::store_immediate_info): Initialize them
6745 from newly added ctor args.
6746 (merged_store_group::apply_stores): Formatting fixes. Sort by
6748 (stmts_may_clobber_ref_p): For stores call also
6749 refs_anti_dependent_p.
6750 (gather_bswap_load_refs): New function.
6751 (imm_store_chain_info::try_coalesce_bswap): New method.
6752 (imm_store_chain_info::coalesce_immediate_stores): Use it.
6753 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
6754 (imm_store_chain_info::output_merged_store): Fail if number of
6755 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
6756 and NOP_EXPR rhs_code.
6757 (pass_store_merging::process_store): Compute n and ins_stmt, if
6758 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
6759 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
6761 (pass_store_merging::execute): Calculate dominators.
6763 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
6764 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
6765 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
6766 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6767 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
6768 class pass_optimize_bswap, bswap_replace,
6769 pass_optimize_bswap::execute): Moved to ...
6770 * gimple-ssa-store-merging.c: ... this file.
6771 Include optabs-tree.h.
6772 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
6773 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6774 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
6775 anonymous namespace, remove static keywords.
6776 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
6777 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
6779 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
6782 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6784 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
6786 * vec.h (debug_helper): New function.
6787 (DEFINE_DEBUG_VEC): New macro.
6788 * hash-set.h (debug_helper): New function.
6789 (DEFINE_DEBUG_HASH_SET): New macro.
6790 * cfg.c (debug_slim (edge)): New function.
6791 Call DEFINE_DEBUG_VEC for edges.
6792 Call DEFINE_DEBUG_HASH_SET for edges.
6793 * cfghooks.c (debug_slim (basic_block)): New function.
6794 Call DEFINE_DEBUG_VEC for basic blocks.
6795 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6796 * print-tree.c (debug_slim): New function to handle trees.
6797 Call DEFINE_DEBUG_VEC for trees.
6798 Call DEFINE_DEBUG_HASH_SET for trees.
6799 (debug (vec<tree, va_gc>) &): Remove.
6800 (debug (<vec<tree, va_gc>) *): Remove.
6801 * print-rtl.c (debug_slim): New function to handle const_rtx.
6802 Call DEFINE_DEBUG_VEC for rtx_def.
6803 Call DEFINE_DEBUG_VEC for rtx_insn.
6804 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6805 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6806 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6807 (debug (vec<rtx_insn *> *ptr): Remove.
6808 (debug_insn_vector): Remove.
6809 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6811 2017-11-20 Tom de Vries <tom@codesourcery.com>
6813 PR rtl-optimization/82020
6814 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6815 IF_THEN_ELSE condition.
6817 2017-11-19 Jeff Law <law@redhat.com>
6819 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6820 of degenerates resulting from ignoring an edge.
6822 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6825 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6827 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6830 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6831 for uninitialied counts.
6833 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6836 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6838 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6841 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6843 2017-11-19 Tom de Vries <tom@codesourcery.com>
6845 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6848 2017-11-19 Tom de Vries <tom@codesourcery.com>
6850 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6853 2017-11-19 Tom de Vries <tom@codesourcery.com>
6855 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6858 2017-11-19 Tom de Vries <tom@codesourcery.com>
6860 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6862 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6864 2017-11-19 Tom de Vries <tom@codesourcery.com>
6866 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6868 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6869 semicolon after MCORE_EXPORT_NAME call.
6871 2017-11-19 Tom de Vries <tom@codesourcery.com>
6874 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6875 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6877 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6879 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6880 (cgraph_edge::clone): Cleanup updating of profile.
6881 * ipa-cp.c (update_profiling_info): Likewise.
6882 * ipa-inline-transform.c (inline_transform): Likewise.
6883 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6884 * ipa-split.c (execute_split_functions): Do not split when function
6886 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6887 * profile-count.c (profile_count::dump): Add global0.
6888 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6890 (profile_count::to_sreal_scale): Likewise.
6891 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6892 (profile_count::combine_with_ipa_count): New function.
6893 * profile-count.h (profile_guessed_global0adjusted): New.
6894 (profile_count::adjusted_zero): New.
6895 (profile_count::global0adjusted): New.
6896 (profile_count::combine_with_ipa_count): New.
6897 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6898 correct profile of return block of split functions.
6899 (copy_cfg_body): Remove unused profile_count.
6900 (copy_body): Likewise.
6901 (expand_call_inline): Update.
6902 (tree_function_versioning): Update.
6904 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6906 * hash-set.h (hash_set::empty): New.
6907 * tree-ssa-threadbackward.h: Delete.
6908 * tree-ssa-threadbackward.c (class thread_jumps): New.
6909 Move max_threaded_paths into class.
6910 (fsm_find_thread_path): Remove arguments that are now in class.
6911 (profitable_jump_thread_path): Rename to...
6912 (thread_jumps::profitable_jump_thread_path): ...this.
6913 (convert_and_register_jump_thread_path): Rename to...
6914 (thread_jumps::convert_and_register_current_path): ...this.
6915 (check_subpath_and_update_thread_path): Rename to...
6916 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6917 (register_jump_thread_path_if_profitable): Rename to...
6918 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6919 (handle_phi): Rename to...
6920 (thread_jumps::handle_phi): ...this.
6921 (handle_assignment): Rename to...
6922 (thread_jumps::handle_assignment): ...this.
6923 (fsm_find_control_statement_thread_paths): Rename to...
6924 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6925 (find_jump_threads_backwards): Rename to...
6926 (thread_jumps::find_jump_threads_backwards): ...this.
6927 Initialize path local data.
6928 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6929 from within thread_jumps class.
6930 (pass_early_thread_jumps::execute): Same.
6932 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6934 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6936 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6938 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6939 to canonical location.
6941 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6944 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6946 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6948 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6950 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6951 rcrt1.o%s/grcrt1.o%s for -static-pie.
6953 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6955 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6956 ix86_shift_rotate_cost): Break out from ...
6957 (ix86_rtx_costs): ... here.
6958 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6961 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6963 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6964 when entry block was promoted unlikely.
6965 (estimate_bb_frequencies): Increase frequency scale.
6966 * profile-count.h (profile_count): Export precision info.
6968 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6970 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6971 disturbin profile of entry block.
6973 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6975 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6978 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6980 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6981 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6983 2017-11-17 Jeff Law <law@redhat.com>
6985 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6986 from evrp_dom_walker class. Various methods moved into new class.
6987 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6988 (evrp_range_analyzer::enter): New method.
6989 (evrp_range_analyzer::leave): New method.
6990 (evrp_dom_walker): Remove delegators no longer needed by this class.
6991 Replace vr_values data member with evrp_range_analyzer
6993 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6994 method extracted from evrp_dom_walker::before_dom_children.
6995 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6996 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6998 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6999 Add private copy constructor and move assignment operators.
7000 Privatize methods and class data where trivially possible.
7001 (evrp_dom_walker::cleanup): New function, extracted from
7002 execute_early_vrp. Simplify access to class data.
7004 * vr-values.h (get_output_for_vrp): Prototype.
7005 * vr-values.c (get_output_for_vrp): New function extracted from
7006 vrp_visit_assignment_or_call and extract_range_from_stmt.
7007 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
7009 2017-11-17 Luis Machado <luis.machado@linaro.org>
7011 * config/aarch64/aarch64.c
7012 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
7013 (qdf24xx_tunings) <autoprefetcher_model>: Set to
7014 tune_params::AUTOPREFETCHER_WEAK.
7016 2017-11-17 Tamar Christina <tamar.christina@arm.com>
7019 * config/arm/arm.c (arm_valid_target_attribute_rec):
7020 Parse "arch=" and "+<ext>".
7021 (arm_valid_target_attribute_tree): Re-init global options.
7022 (arm_option_override): Make non-static.
7023 (arm_options_perform_arch_sanity_checks): Make errors fatal.
7024 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
7025 (__ARM_FEATURE_CRC32): Support undef.
7026 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
7027 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
7029 2017-11-17 David Malcolm <dmalcolm@redhat.com>
7031 * gdbinit.in (break-on-diagnostic): New command.
7033 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7035 * config/i386/i386.c (ix86_expand_epilogue): Change simple
7036 return to indirect jump for EH return if control-flow
7037 protection is enabled. Change explicit 'false' argument in
7038 pro_epilogue_adjust_stack with a value of
7040 * config/i386/i386.md (simple_return_indirect_internal):
7041 Remove SImode restriction to support 64-bit.
7043 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
7045 * combine.c (added_notes_insn): New.
7046 (try_combine): Handle added_notes_insn like added_links_insn.
7047 Rewrite return value code.
7048 (distribute_notes): Set added_notes_insn to the earliest insn we added
7051 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
7053 PR rtl-optimization/82621
7054 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
7055 dest of one of those SETs is unused.
7057 2017-11-17 Richard Biener <rguenther@suse.de>
7060 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
7061 * tree-pretty-print.c (dump_generic_node): Handle
7062 annot_expr_parallel_kind.
7063 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
7064 * gimplify.c (gimple_boolify): Likewise.
7066 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7068 * config.gcc (extra_headers): Add cet.h for x86 targets.
7069 * config/i386/cet.h: New file.
7070 * doc/install.texi: Add --enable-cet/--disable-cet.
7072 2017-11-17 Richard Biener <rguenther@suse.de>
7074 PR tree-optimization/83017
7075 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
7076 (gen_parallel_loop): Properly count iterations.
7077 (parallelize_loops): Handle loop->can_be_parallel independent
7078 of flag_loop_parallelize_all. Make static profitability test match
7080 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
7081 * invoke.texi (parloops-min-per-thread): Document.
7083 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
7085 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
7086 upstreaming review comments.
7088 2017-11-17 Tamar Christina <tamar.christina@arm.com>
7090 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
7091 with fast unaligned access.
7092 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
7094 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
7096 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
7097 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
7099 2017-11-17 Richard Biener <rguenther@suse.de>
7101 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
7102 folding of references.
7104 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
7107 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
7108 of replacing call to strncmp with corresponding call to strcmp when
7111 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
7113 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
7114 option prefer-avx256 for skylake-avx512 configuration.
7115 * config/i386/i386.c (ix86_option_override_internal): Ditto.
7116 (get_builtin_code_for_version): Ditto.
7118 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
7119 Monk Chiang <sh.chiang04@gmail.com>
7121 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
7122 (FIXED_REGISTERS): Reserve more register numbers.
7123 (CALL_USED_REGISTERS): Likewise.
7124 (REG_ALLOC_ORDER): Likewise.
7125 (REG_CLASS_CONTENTS): Likewise.
7126 (REGISTER_NAMES): Likewise.
7128 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
7129 Kito Cheng <kito.cheng@gmail.com>
7131 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
7133 * config/nds32/iterators.md: Add vector mode iterators and attributes.
7135 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
7137 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
7138 parameter list for vec_splats.
7140 2017-11-16 Joseph Myers <joseph@codesourcery.com>
7142 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
7144 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
7146 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
7149 * doc/install.texi: Document incompatibility of Perl >=5.6.26
7150 with the required version of automake 1.11.6.
7152 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
7154 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
7156 (DU_C2_3_power9): Correct reservation combinations.
7157 (FP_DIV_power9, VEC_DIV_power9): New.
7158 (power9-alu): Split out rotate/shift...
7159 (power9-rot): ...to here, correct dispatch resource.
7160 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
7162 (power9-fp): Correct latency.
7163 (power9-sdiv): Add div/sqrt resource.
7164 (power9-ddiv): Correct latency, add div/sqrt resource.
7165 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
7166 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
7168 (power9-qpdiv, power9-qpmul): Adjust resource usage.
7170 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
7172 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
7173 switch statement mapping KF built-ins to TF built-ins if we don't
7174 have the proper ISA 3.0 assembler support.
7176 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7178 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
7179 (lower_emutls_function_body): Do not compute it.
7181 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7183 * ipa-split.c (split_bb_info): Turn time to sreal.
7184 (split_point): Likewise.
7185 (dump_split_point): Likewise.
7186 (fine_split_points): Likewise.
7187 (execute_split_functions): Only zero split_bbs; turn time to sreals.
7189 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7191 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
7194 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7196 * predict.c (combine_predictions_for_bb): Preserve zero predicted
7198 (expensive_function_p): Remove useless assert.
7199 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
7201 2017-11-16 Martin Sebor <msebor@redhat.com>
7203 PR tree-optimization/82588
7204 PR tree-optimization/82583
7205 * tree-vrp.c (check_array_ref): Handle flexible array members,
7206 string literals, and inner indices.
7207 (search_for_addr_array): Add detail to diagnostics.
7209 2017-11-16 Nathan Sidwell <nathan@acm.org>
7213 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
7214 (SET_DECL_ASSEMBLER_NAME): Forward to
7215 overwrite_decl_assembler_name.
7216 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
7217 (overwrite_decl_assembler_name): Declare.
7218 * tree.c (overwrite_decl_assembler_name): New.
7219 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
7220 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
7221 (LANG_HOOKS_INITIALIZER): Add it.
7222 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
7223 * langhooks.c (lhd_set_decl_assembler_name): Use
7224 SET_DECL_ASSEMBLER_NAME.
7225 (lhd_overwrite_decl_assembler_name): Default implementation.
7227 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7228 Jackson Woodruff <jackson.woodruff@arm.com>
7230 PR tree-optimization/71026
7231 * match.pd: Canonicalize constant multiplies in division.
7233 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7235 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
7236 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
7237 Remove OPT_fomit_frame_pointer entry.
7238 * common/config/arc/arc-common.c: Likewise.
7239 * common/config/arm/arm-common.c: Likewise.
7240 * common/config/avr/avr-common.c: Likewise.
7241 * common/config/c6x/c6x-common.c: Likewise.
7242 * common/config/cr16/cr16-common.c: Likewise.
7243 * common/config/cris/cris-common.c: Likewise.
7244 * common/config/epiphany/epiphany-common.c: Likewise.
7245 * common/config/fr30/fr30-common.c: Likewise.
7246 * common/config/frv/frv-common.c: Likewise.
7247 * common/config/ia64/ia64-common.c: Likewise.
7248 * common/config/iq2000/iq2000-common.c: Likewise.
7249 * common/config/lm32/lm32-common.c: Likewise.
7250 * common/config/m32r/m32r-common.c: Likewise.
7251 * common/config/mcore/mcore-common.c: Likewise.
7252 * common/config/microblaze/microblaze-common.c: Likewise.
7253 * common/config/mips/mips-common.c: Likewise.
7254 * common/config/mmix/mmix-common.c: Likewise.
7255 * common/config/mn10300/mn10300-common.c: Likewise.
7256 * common/config/nios2/nios2-common.c: Likewise.
7257 * common/config/pa/pa-common.c: Likewise.
7258 * common/config/pdp11/pdp11-common.c: Likewise.
7259 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
7260 * common/config/riscv/riscv-common.c: Likewise.
7261 * common/config/rs6000/rs6000-common.c: Likewise.
7262 * common/config/rx/rx-common.c: Likewise.
7263 * common/config/s390/s390-common.c: Likewise.
7264 * common/config/sh/sh-common.c: Likewise.
7265 * common/config/sparc/sparc-common.c: Likewise.
7266 * common/config/tilegx/tilegx-common.c: Likewise.
7267 * common/config/tilepro/tilepro-common.c: Likewise.
7268 * common/config/v850/v850-common.c: Likewise.
7269 * common/config/visium/visium-common.c: Likewise.
7270 * common/config/xstormy16/xstormy16-common.c: Likewise.
7271 * common/config/xtensa/xtensa-common.c: Likewise.
7272 * invoke.texi (-fomit-frame-pointer): Update documentation.
7274 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7276 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
7278 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7280 * cfg.c (scale_bbs_frequencies_int,
7281 cale_bbs_frequencies_gcov_type): Remove.
7282 * cfg.h (scale_bbs_frequencies_int,
7283 cale_bbs_frequencies_gcov_type): Remove.
7285 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7287 * tree-ssa-loop-manip.c
7288 (scale_dominated_blocks_in_loop): Update to profile counts.
7289 (tree_transform_and_unroll_loop): Likewise.
7291 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7293 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
7294 scale_bbs_frequencies_int.
7296 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7298 * final.c (compute_alignments): Use counts rather than frequencies.
7300 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7302 * cfgloopanal.c: Include sreal.h
7303 (average_num_loop_insns): Use counts and sreal for accounting.
7305 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7307 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
7310 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7311 Kito Cheng <kito.cheng@gmail.com>
7313 * config/nds32/constraints.md: Provide more constraints.
7314 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
7315 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
7316 support constraints usage.
7318 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7320 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
7322 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7323 Kito Cheng <kito.cheng@gmail.com>
7325 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
7326 * config/nds32/nds32.opt: Refine the layout.
7327 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
7328 TARGET_EXT_STRING): Support new options.
7329 * config/nds32/nds32.h: Likewise.
7330 * config/nds32/nds32.md: Likewise.
7331 * config/nds32/nds32-predicates.c: Likewise.
7332 * config/nds32/constraints.md: Likewise.
7333 * common/config/nds32/nds32-common.c: Likewise.
7335 2017-11-16 Julia Koval <julia.koval@intel.com>
7338 * config/i386/gfniintrin.h: Add sse check.
7339 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
7341 2017-11-16 Julia Koval <julia.koval@intel.com>
7343 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
7344 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
7345 (ix86_handle_option): Handle -mavx512vbmi2.
7346 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
7347 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
7348 * config/i386/i386-c.c (__AVX512VBMI2__): New.
7349 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
7350 (ix86_valid_target_attribute_inner_p): Ditto.
7351 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
7352 * config/i386/i386.opt (mavx512vbmi2): New option.
7353 * doc/invoke.texi: Add new option.
7355 2017-11-16 Julia Koval <julia.koval@intel.com>
7357 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
7358 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
7359 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
7360 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
7361 _mm512_gf2p8mul_epi8): New intrinsics.
7362 * config/i386/i386-builtin-types.def
7363 (V64QI_FTYPE_V64QI_V64QI): New type.
7364 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
7365 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
7366 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
7367 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
7368 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
7369 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
7371 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
7373 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
7374 explicitly as a stream of bytes.
7376 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7378 * config/rs6000/altivec.h (vec_xst_be): New #define.
7379 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
7380 and externalize from *altivec_vperm_<mode>_internal.
7381 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
7383 (XL_BE_V8HI): Likewise.
7384 (XL_BE_V4SI): Likewise.
7385 (XL_BE_V4SI): Likewise.
7386 (XL_BE_V2DI): Likewise.
7387 (XL_BE_V4SF): Likewise.
7388 (XL_BE_V2DF): Likewise.
7389 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
7390 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
7391 all array entries with these keys: VSX_BUILTIN_VEC_XL,
7392 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
7393 VSX_BUILTIN_VEC_XST_BE.
7394 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
7395 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
7397 (altivec_init_builtins): Replace conditional calls to def_builtin
7398 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
7399 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
7400 with unconditional calls. Remove calls to def_builtin for
7401 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
7402 __builtin_vec_xst_be.
7403 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
7404 to define_expand, and add alternate RTL generation for P8.
7405 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
7406 vsx_ld_elemrev_v8hi.
7407 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
7408 add alternate RTL generation for P8.
7409 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
7410 vsx_ld_elemrev_v16qi.
7411 (vsx_st_elemrev_v8hi): Convert define_insn
7412 to define_expand, and add alternate RTL generation for P8.
7413 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
7414 vsx_st_elemrev_v8hi.
7415 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
7416 add alternate RTL generation for P8.
7417 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
7418 vsx_st_elemrev_v16qi.
7420 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
7423 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
7424 TARGET_AVX512ER check.
7425 (ix86_option_override_internal): Set MASK_VZEROUPPER if
7426 neither -mzeroupper nor -mno-zeroupper is used and
7427 TARGET_EMIT_VZEROUPPER is set.
7428 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
7429 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
7431 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
7433 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7434 folding of vector compares.
7435 (fold_build_vec_cmp): New helper function.
7436 (fold_compare_helper): New helper function.
7437 (builtin_function_type): Add compare builtins to the list of functions
7438 having unsigned arguments. Cosmetic updates to comment indentation.
7439 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
7440 the not+eq combination.
7442 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7444 PR tree-optimization/82726
7445 PR tree-optimization/70754
7446 * tree-predcom.c (order_drefs_by_pos): New function.
7447 (combine_chains): Move code setting has_max_use_after to...
7448 (try_combine_chains): ...here. New parameter. Sort combined chains
7449 according to position information.
7450 (tree_predictive_commoning_loop): Update call to above function.
7451 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
7453 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7455 PR tree-optimization/82726
7457 2017-01-23 Bin Cheng <bin.cheng@arm.com>
7459 PR tree-optimization/70754
7460 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
7461 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
7462 combined stmt before it if not NULL.
7463 (combine_chains): Process refs reversely and compute dominance point
7467 2017-02-23 Bin Cheng <bin.cheng@arm.com>
7469 PR tree-optimization/79663
7470 * tree-predcom.c (combine_chains): Process refs in reverse order
7471 only for ZERO length chains, and add explaining comment.
7473 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7475 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
7476 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
7477 * doc/invoke.texi (ARM Options): Add armv8.3-a.
7479 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7481 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
7483 2017-11-15 Martin Liska <mliska@suse.cz>
7485 * tree-cfg.c (pass_warn_function_return::execute):
7486 Compare warn_return_type for greater than zero.
7488 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
7492 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
7493 to return true on Xeon and not on Xeon Phi.
7494 (ix86_check_avx256_register): Changed to ...
7495 (ix86_check_avx_upper_register): ... this. Add extra check for
7496 VALID_AVX512F_REG_OR_XI_MODE.
7497 (ix86_avx_u128_mode_needed): Changed
7498 ix86_check_avx256_register to ix86_check_avx_upper_register.
7499 (ix86_check_avx256_stores): Changed to ...
7500 (ix86_check_avx_upper_stores): ... this. Changed
7501 ix86_check_avx256_register to ix86_check_avx_upper_register.
7502 (ix86_avx_u128_mode_after): Changed
7503 avx_reg256_found to avx_upper_reg_found. Changed
7504 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
7505 (ix86_avx_u128_mode_entry): Changed
7506 ix86_check_avx256_register to ix86_check_avx_upper_register.
7507 (ix86_avx_u128_mode_exit): Ditto.
7508 * config/i386/i386.h: (host_detect_local_cpu): New define.
7510 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7512 * config/arm/xgene1.md (xgene1): Split into automatons
7513 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
7514 (xgene1_f_load): Adjust reservations and/or types.
7515 (xgene1_f_store): Likewise.
7516 (xgene1_load_pair): Likewise.
7517 (xgene1_store_pair): Likewise.
7518 (xgene1_fp_load1): Likewise.
7519 (xgene1_load1): Likewise.
7520 (xgene1_store1): Likewise.
7521 (xgene1_move): Likewise.
7522 (xgene1_alu): Likewise.
7523 (xgene1_simd): Likewise.
7524 (xgene1_bfm): Likewise.
7525 (xgene1_neon_load1): Likewise.
7526 (xgene1_neon_store1): Likewise.
7527 (xgene1_neon_logic): Likewise.
7528 (xgene1_neon_st1): Likewise.
7529 (xgene1_neon_ld1r): Likewise.
7530 (xgene1_alu_cond): Added.
7531 (xgene1_shift_reg): Likwise.
7532 (xgene1_bfx): Likewise.
7533 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
7535 2017-11-15 Jakub Jelinek <jakub@redhat.com>
7538 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
7540 (can_widen_mult_without_libcall): New function.
7541 (expand_mul_overflow): If only checking unsigned mul overflow,
7542 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
7543 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
7544 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
7545 (expand_DIVMOD): Formatting fix.
7546 * expmed.h (expand_mult): Add NO_LIBCALL argument.
7547 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
7548 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
7550 PR tree-optimization/82977
7551 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
7552 constructed temporary to strlen_to_stridx.put.
7554 2017-11-15 Martin Liska <mliska@suse.cz>
7556 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
7557 * configure: Regenerate.
7559 2017-11-15 Martin Liska <mliska@suse.cz>
7562 * config/sh/sh-mem.cc: Use proper probability for
7565 2017-11-14 Jeff Law <law@redhat.com>
7567 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
7568 the red zone for stack_clash_protection_final_dynamic_probe targets
7569 when the total dynamic stack size is zero bytes.
7571 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
7572 blocks is post order.
7574 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
7576 * dumpfile.h (TDF_COMPARE_DEBUG): New.
7577 * final.c (rest_of_clean_state): Set it for the
7578 -fcompare-debug dump.
7579 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
7580 class when TDF_COMPARE_DEBUG is set.
7582 * dwarf2out.c (gen_producer_string): Discard
7585 2017-11-15 Joseph Myers <joseph@codesourcery.com>
7588 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
7589 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
7590 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
7591 (__TGMATH_REAL_2_3): Remove macros.
7592 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
7593 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
7595 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
7597 (remquo): Define using __TGMATH_REAL_3.
7599 2017-11-14 Jeff Law <law@redhat.com>
7601 * vr-values.c: New file with contents extracted from tree-vrp.c.
7602 * Makefile.in (OBJS): Add vr-values.o
7603 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
7604 (set_value_range, set_and_canonicalize_value_range): Likewise.
7605 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
7606 (value_range_constant_singleton, symbolic_range_p): Likewise.
7607 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
7608 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
7609 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
7610 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
7611 (find_case_label_range, find_case_label_index): Likewise.
7612 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
7613 (range_int_cst_singleton_p, value_inside_range): Likewise.
7614 (get_single_symbol): Likewise.
7615 (switch_update): Move structure definition here.
7616 (to_remove_edges, to_update_switch_stmts): Provide externs.
7617 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
7618 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
7619 (vrp_val_is_min, set_value_range): Likewise.
7620 (set_and_canonicalize_value_range, copy_value_range): Likewise.
7621 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
7622 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
7623 (range_is_nonnull, range_int_cst_p): Likewwise.
7624 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
7625 (get_single_symbol, operand_less_p): Likewise
7626 (compare_values_warnv, compare_values): Likewise.
7627 (value_inside_range, value_range_constant_singleton): Likewise.
7628 (zero_nonzero_bitgs_from_vr): Likewise.
7629 (extract_range_from_binary_expr_1): Likewise.
7630 (overflow_comparison_p): Likewise.
7631 (to_remove_edges, to_update_switch_stmts): Likewise.
7632 (find_case_label-index, find_case_label_range): Likewise.
7633 (switch_update, set_value_range_to_nonnegative): Remove.
7634 (set_value_range_to_truthvalue): Likewise.
7635 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
7636 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
7637 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
7638 (find_case_label_ranges, test_for_singularity): Likewise.
7639 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
7640 (x_vr_values): Move to its remaining use site.
7642 2017-11-10 Jeff Law <law@redhat.com>
7644 * vr-values.h (VR_INITIALIZER): Move #define here.
7645 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
7646 * Makefile.in (OBJS): Add tree-evrp.o
7647 * tree-vrp.h (assert_info): Move structure definition here.
7648 (set_value_range_to_varying): Prototype.
7649 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7650 (infer_value_range, register_edge_assert_for): Likewise.
7651 (stmt_interesting_for_vrp): Likewise.
7652 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
7653 (set_value_range_to_varying): No longer static.
7654 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7655 (infer_value_range, register_edge_assert_for): Likewise.
7657 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7659 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
7660 generate the XXBRD instruction.
7662 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
7663 (rs6000_builtin_type_compatible): Treat _Float128 and long double
7664 as being compatible if -mabi=ieeelongdouble.
7665 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
7666 to setup float128 built-ins with hardware support.
7667 (BU_FLOAT128_HW_2): Likewise.
7668 (BU_FLOAT128_HW_3): Likewise.
7669 (BU_FLOAT128_HW_VSX_1): Likewise.
7670 (BU_FLOAT128_HW_VSX_2): Likewise.
7671 (scalar_extract_expq): Change float128 built-in functions to
7672 accommodate having both KFmode and TFmode functions. Use the
7673 KFmode variant as the default.
7674 (scalar_extract_sigq): Likewise.
7675 (scalar_test_neg_qp): Likewise.
7676 (scalar_insert_exp_q): Likewise.
7677 (scalar_insert_exp_qp): Likewise.
7678 (scalar_test_data_class_qp): Likewise.
7679 (sqrtf128_round_to_odd): Delete processing the round to odd
7680 built-in functions as special built-in functions, and define them
7681 as float128 built-ins. Use the KFmode variant as the default.
7682 (truncf128_round_to_odd): Likewise.
7683 (addf128_round_to_odd): Likewise.
7684 (subf128_round_to_odd): Likewise.
7685 (mulf128_round_to_odd): Likewise.
7686 (divf128_round_to_odd): Likewise.
7687 (fmaf128_round_to_odd): Likewise.
7688 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7689 support for KFmode and TFmode xststdcqp calls.
7690 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
7691 point, switch the built-in handlers for the get/set float128
7692 exponent, get float128 mantissa, float128 test built-ins, and the
7693 float128 round to odd built-in functions. Eliminate creating the
7694 float128 round to odd built-in functions as special built-ins.
7695 (rs6000_init_builtins): Eliminate special creation of the float128
7696 round to odd built-in functions.
7697 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
7698 function insns to support both TFmode and KFmode variants.
7699 (xsxsigqp_<mode>): Likewise.
7700 (xsiexpqpf_<mode>): Likewise.
7701 (xsiexpqp_<mode>): Likewise.
7702 (xststdcqp_<mode>): Likewise.
7703 (xststdcnegqp_<mode>): Likewise.
7704 (xststdcqp_<mode>): Likewise.
7706 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7708 * tree-ssa-threadupdate.c (compute_path_counts): Remove
7709 unused path_in_freq_ptr parameter.
7710 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
7712 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7714 * ipa-inline.c (edge_badness): Dump sreal frequency.
7715 (compute_inlined_call_time): Match natural implementaiton ...
7716 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
7717 forgotten division by CGRAPH_FREQ_BASE.
7719 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7721 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
7722 Solaris 11. Update comment.
7723 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
7725 * config/sol2.h (STARTFILE_SPEC): Likewise.
7726 * configure: Regenerate.
7728 2017-11-14 Carl Love <cel@us.ibm.com>
7730 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
7731 le_ and be_ prefixes to swap* variables. Remove
7732 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
7735 2017-11-14 Jason Merrill <jason@redhat.com>
7737 Support GTY((cache)) on hash_map.
7738 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
7739 (ggc_cache_remove): Override it instead of ggc_mx.
7740 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
7741 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
7742 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
7743 (simple_cache_map_traits): Override maybe_mx.
7744 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
7745 (hash_map): Friend gt_cleare_cache.
7746 (gt_cleare_cache): New.
7747 * tree.h (tree_cache_traits): New hash_map traits class.
7748 (tree_cache_map): New typedef.
7750 2017-11-14 Richard Biener <rguenther@suse.de>
7752 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
7753 paramter and handling.
7754 (cleanup_control_flow_bb): Likewise.
7755 (cleanup_control_flow_pre): New helper performing a DFS walk
7756 to call cleanup_control_flow_bb in PRE order.
7757 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
7758 via cleanup_control_flow_pre.
7760 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
7762 * config/aarch64/aarch64-simd.md
7763 (aarch64_simd_bsl<mode>_internal): Remove DImode.
7764 (*aarch64_simd_bsl<mode>_alt): Likewise.
7765 (aarch64_simd_bsldi_internal): New.
7766 (aarch64_simd_bsldi_alt): Likewise.
7768 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7770 * tracer.c (better_p): Do not compare frequencies.
7771 * reg-stack.c (better_edge): Likewise.
7772 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
7775 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7777 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
7778 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
7779 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
7780 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7781 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
7782 * predict.c (maybe_hot_frequency_p): Inline to ...
7783 (maybe_hot_count_p): ... here; rewrite to counts.
7784 (counts_to_freqs): Rename to ...
7785 (update_max_bb_count): ... this one.
7786 (expensive_function_p): Use counts.
7787 (estimate_bb_frequencies): Update.
7788 (rebuild_frequencies): Update.
7789 * predict.h (counts_to_freqs): Rename to ...
7790 (update_max_bb_count): ... this one.
7791 * profile.c (compute_branch_probabilities): Add debug info
7792 * tree-inline.c (expand_call_inline): Update debug info.
7793 (optimize_inline_calls): Use update_max_bb_count..
7794 (tree_function_versioning): Use update_max_bb_count..
7795 * value-prof.c (gimple_value_profile_transformations):
7796 Do not use update_max_bb_count.
7798 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7800 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7801 always use frequencies.
7803 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7805 * bb-reorder.c: Remove frequencies from comments.
7806 (better_edge_p): Use profile counts.
7807 (find_traces): Dump profile counts.
7808 (rotate_loop): Use profile counts.
7809 (find_traces_1_round): Likewise.
7810 (connect_better_edge_p): Use counts instead of probabilities for
7812 (copy_bb_p): Drop early check for non-0 frequency.
7813 (sanitize_hot_paths): Update comments.
7815 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7817 * ipa-split.c (struct split_point): Add count.
7818 (consider_split): Do not compute incoming frequency; compute incoming
7819 count and store it to split_point.
7820 (split_function): Set count of the call to split part correctly.
7822 2017-11-13 Carl Love <cel@us.ibm.com>
7824 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7826 2017-11-13 Tom Tromey <tom@tromey.com>
7828 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7830 2017-11-13 Carl Love <cel@us.ibm.com>
7832 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7833 Add support for builtins:
7834 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7835 vector {un,}signed {char,int,short},
7836 vector {un,}signed {char,int,short}) arguments.
7837 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7838 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7839 Add BU_P9V_AV_2 expansions for the builtins.
7840 * config/rs6000/altivec.h (vec_first_match_index,
7841 vec_first_mismatch_index, vec_first_match_or_eos_index,
7842 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7843 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7844 new extern declaration.
7845 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7846 * config/rs6000/vsx.md (first_match_index_<mode>,
7847 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7848 first_mismatch_or_eos_index_<mode>): Add define expand.
7849 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7850 * doc/extend.texi: Update the built-in documenation file for the new
7853 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7855 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7856 into the min/max operations for _Float<N> and _Float<N>X types.
7858 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7861 * dwarf2out.c (do_eh_frame): New static variable.
7862 (dwarf2out_begin_prologue): Set it.
7863 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7865 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7867 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7869 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7871 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7874 * debug.h (dwarf2out_do_eh_frame): Declare.
7875 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7876 (dwarf2out_do_frame): Use it.
7877 (dwarf2out_do_cfi_asm): Likewise.
7878 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7879 (dwarf2out_assembly_start): Likewise.
7880 (dwarf2out_begin_prologue): Fix comment.
7881 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7882 if the target needs either debug or unwind DWARF2 info.
7883 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7884 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7885 -ftrapping-math, -ftrapv and -fwrapv.
7887 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7889 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7890 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7891 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7892 (estimate_edge_size_and_time): Likewise.
7893 (ipa_merge_fn_summary_after_inlining): Likewise.
7894 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7895 (compute_uninlined_call_time): Use sreal_frequency.
7896 (compute_inlined_call_time): Likewise.
7897 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7898 * profile-count.c: Include sreal.h.
7899 (profile_count::to_sreal_scale): New.
7900 * profile-count.h: Forward declare sreal.
7901 (profile_count::to_sreal_scale): Declare.
7903 2017-11-13 Nathan Sidwell <nathan@acm.org>
7905 * diagnostic.c (maybe_line_and_column): New.
7906 (diagnostic_get_location_text): Use it.
7907 (diagnostic_report_current_module): Likewise.
7908 (test_diagnostic_get_location_text): Add tests.
7910 2017-11-13 Luis Machado <luis.machado@linaro.org>
7912 * doc/md.texi (Specifying processor pipeline description): Fix
7913 incorrect latency for the div instruction example.
7915 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7917 PR tree-optimization/78821
7918 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7919 that bit_not_p is the same.
7920 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7921 (split_group): Count precisely bit_not_p bits in each statement.
7922 (invert_op): New function.
7923 (imm_store_chain_info::output_merged_store): Use invert_op to
7924 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7925 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7927 2017-11-13 Martin Liska <mliska@suse.cz>
7929 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7930 (struct source_info): Likewise.
7931 (add_branch_counts): Likewise.
7932 (add_line_counts): Likewise.
7933 (function_summary): Likewise.
7934 (output_intermediate_line): Likewise.
7935 (generate_results): Likewise.
7937 2017-11-13 Martin Liska <mliska@suse.cz>
7939 * gcov.c (struct block_info): Remove typedef for block_t.
7940 (struct line_info): Likewise.
7941 (line_info::has_block): Likewise.
7942 (EXIT_BLOCK): Likewise.
7943 (unblock): Likewise.
7944 (circuit): Likewise.
7945 (get_cycles_count): Likewise.
7946 (process_file): Likewise.
7947 (read_graph_file): Likewise.
7948 (solve_flow_graph): Likewise.
7949 (find_exception_blocks): Likewise.
7950 (add_line_counts): Likewise.
7951 (accumulate_line_info): Likewise.
7952 (output_line_details): Likewise.
7954 2017-11-13 Martin Liska <mliska@suse.cz>
7956 * gcov.c (struct arc_info): Remove typedef for arc_t.
7957 (struct line_info): Likewise.
7958 (add_branch_counts): Likewise.
7959 (output_branch_count): Likewise.
7960 (function_info::~function_info): Likewise.
7961 (circuit): Likewise.
7962 (output_intermediate_line): Likewise.
7963 (read_graph_file): Likewise.
7964 (solve_flow_graph): Likewise.
7965 (find_exception_blocks): Likewise.
7966 (add_line_counts): Likewise.
7967 (accumulate_line_info): Likewise.
7968 (output_line_details): Likewise.
7969 (output_function_details): Likewise.
7971 2017-11-13 Martin Liska <mliska@suse.cz>
7973 * gcov.c (struct function_info): Remove typedef for function_t.
7974 (struct source_info): Likewise.
7975 (source_info::get_functions_at_location): Likewise.
7976 (solve_flow_graph): Likewise.
7977 (find_exception_blocks): Likewise.
7978 (add_line_counts): Likewise.
7979 (output_intermediate_file): Likewise.
7980 (process_file): Likewise.
7981 (generate_results): Likewise.
7982 (release_structures): Likewise.
7983 (read_graph_file): Likewise.
7984 (read_count_file): Likewise.
7985 (accumulate_line_counts): Likewise.
7986 (output_lines): Likewise.
7988 2017-11-13 Martin Liska <mliska@suse.cz>
7990 * gcov.c (function_info::function_info): Remove num_counts
7991 and add vector<gcov_type>.
7992 (function_info::~function_info): Use the vector.
7993 (process_file): Likewise.
7994 (read_graph_file): Likewise.
7995 (read_count_file): Likewise.
7996 (solve_flow_graph): Likewise.
7998 2017-11-13 Martin Liska <mliska@suse.cz>
8000 * gcov.c (function_info::is_artificial): New function.
8001 (process_file): Erase all artificial early.
8002 (generate_results): Skip as all artificial are already
8005 2017-11-13 Martin Liska <mliska@suse.cz>
8007 * gcov.c (read_graph_file): Store to global vector of functions.
8008 (read_count_file): Iterate the vector.
8009 (process_file): Likewise.
8010 (generate_results): Likewise.
8011 (release_structures): Likewise.
8013 2017-11-13 Jakub Jelinek <jakub@redhat.com>
8015 PR tree-optimization/82954
8016 * gimple-ssa-store-merging.c
8017 (imm_store_chain_info::coalesce_immediate_stores): If
8018 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
8020 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
8022 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
8023 Upddate call to ENDIAN_LANE_N.
8024 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
8025 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
8026 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
8027 and use aarch64_endian_lane_rtx.
8028 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
8030 2017-11-12 Tom de Vries <tom@codesourcery.com>
8032 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
8034 2017-11-12 Tom de Vries <tom@codesourcery.com>
8036 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
8038 2017-11-12 Tom de Vries <tom@codesourcery.com>
8040 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
8042 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
8043 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
8044 * defaults.h (ASM_OUTPUT_LABELREF): Same.
8046 2017-11-11 Martin Sebor <msebor@redhat.com>
8049 * doc/extend.texi (attribute nonstring): Remove spurious argument.
8052 * prefic.c (translate_name): Replace strncpy with memcpy to
8053 avoid -Wstringop-truncation.
8055 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8057 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
8059 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8061 * predict.c (maybe_hot_frequency_p): Do not use cfun.
8063 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8065 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
8068 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
8070 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
8071 enable generating XXBRH if the value is in a vector register.
8072 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
8073 value is in a vector register.
8074 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
8075 register to register bswap64's instead of doing the GPR sequence
8076 used on previous machines.
8077 (bswapdi2_xxbrd): New insn.
8078 (bswapdi2_reg): Disallow on ISA 3.0.
8079 (register to register bswap64 splitter): Do not split the insn on
8080 ISA 3.0 systems that use XXBRD.
8082 2017-11-10 Martin Sebor <msebor@redhat.com>
8085 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
8087 (find_subframework_file): Same.
8089 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8091 * auto-profile.c (afdo_indirect_call): Drop frequency.
8092 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
8093 (cgraph_node::create_edge): Drop frequency argument.
8094 (cgraph_node::create_indirect_edge): Drop frequency argument.
8095 (cgraph_edge::make_speculative): Drop frequency arguments.
8096 (cgraph_edge::resolve_speculation): Do not update frequencies
8097 (cgraph_edge::dump_edge_flags): Do not dump frequency.
8098 (cgraph_node::dump): Check consistency in IPA mode.
8099 (cgraph_edge::maybe_hot_p): Use IPA counter.
8100 (cgraph_edge::verify_count_and_frequency): Rename to ...
8101 (cgraph_edge::verify_count): ... this one; drop frequency checking.
8102 (cgraph_node::verify_node): Update.
8103 * cgraph.h (struct cgraph_edge): Drop frequency.
8104 (cgraph_edge::frequency): New function.
8105 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
8107 (cgraph_edge::rebuild_edges): Likewise.
8108 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
8109 (duplicate_thunk_for_node): Do not pass frequency.
8110 (cgraph_node::create_clone): Scale only counts.
8111 (cgraph_node::create_virtual_clone): Do not pass frequency.
8112 (cgraph_node::create_edge_including_clones): Do not pass frequency.
8113 (cgraph_node::create_version_clone): Do not pass frequency.
8114 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
8115 (cgraph_node::expand_thunk): Do not pass frequency.
8116 (cgraph_node::create_wrapper): Do not pass frequency.
8117 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
8119 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
8120 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
8121 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
8122 (gather_caller_stats): Use frequency function.
8123 (ipcp_cloning_candidate_p): Use frequency function.
8124 (ipcp_propagate_stage): Use frequency function.
8125 (get_info_about_necessary_edges): Use frequency function.
8126 (update_profiling_info): Update only IPA profile.
8127 (update_specialized_profile): Use frequency functoin.
8128 (perhaps_add_new_callers): Update only IPA profile.
8129 * ipa-devirt.c (ipa_devirt): Use IPA profile.
8130 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
8131 (dump_ipa_call_summary): Use frequency function.
8132 (estimate_edge_size_and_time): Use frequency function.
8133 (ipa_merge_fn_summary_after_inlining): Use frequency function.
8134 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
8135 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
8136 (update_noncloned_counts): ... ths one; scale counts only.
8137 (clone_inlined_nodes): Do not scale frequency.
8138 (inline_call): Do not pass frequency.
8139 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
8140 (compute_inlined_call_time): Use IPA profile.
8141 (want_inline_small_function_p): Use IPA profile.
8142 (want_inline_self_recursive_call_p): Use IPA profile.
8143 (edge_badness): Use IPA profile.
8144 (lookup_recursive_calls): Use IPA profile.
8145 (recursive_inlining): Do not pass frequency.
8146 (resolve_noninline_speculation): Do not update frequency.
8147 (inline_small_functions): Collect max of IPA profile.
8148 (dump_overall_stats): Dump IPA porfile.
8149 (dump_inline_stats): Dump IPA porfile.
8150 (ipa_inline): Collect IPA stats.
8151 * ipa-inline.h (clone_inlined_nodes): Update prototype.
8152 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
8153 (ipa_propagate_frequency): Use frequency function.
8154 (ipa_profile): Cleanup.
8155 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
8156 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
8157 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
8158 (input_node): Do not stream frequency.
8159 (input_edge): Do not stream frequency.
8160 (merge_profile_summaries): Scale only IPA profiles.
8161 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
8162 * predict.c (drop_profile): Do not recompute frequency.
8163 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
8164 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
8165 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
8166 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
8167 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
8168 * tree-inline.c (copy_bb): Do not scale frequency.
8169 (expand_call_inline): Do not scale frequency.
8170 (tree_function_versioning): Do not scale frequency.
8171 * ubsan.c (ubsan_create_edge): Do not pass frequency.
8173 2017-11-10 Julia Koval <julia.koval@intel.com>
8175 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
8176 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
8177 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
8178 (_mm256_maskz_gf2p8affine_epi64_epi8)
8179 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
8180 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
8181 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
8182 (__builtin_ia32_vgf2p8affineqb_v32qi)
8183 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
8184 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
8186 2017-11-10 Tamar Christina <tamar.christina@arm.com>
8190 (arm_option_override): Refactor.
8191 (arm_option_reconfigure_globals): New.
8192 (arm_options_perform_arch_sanity_checks): New.
8193 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
8195 (arm_options_perform_arch_sanity_checks): Likewise
8197 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
8199 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
8200 (power9-qpmul): New.
8201 * rs6000/rs6000.md ("type" attr): Add qmul.
8202 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
8203 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
8204 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
8206 2017-11-10 Martin Sebor <msebor@redhat.com>
8209 * builtins.c (compute_objsize): Handle arrays that
8210 compute_builtin_object_size likes to fail for. Make extern.
8211 * builtins.h (compute_objsize): Declare.
8212 (check_strncpy_sizes): New function.
8213 (expand_builtin_strncpy): Call check_strncpy_sizes.
8214 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
8215 -Wstringop-truncation.
8216 (gimple_fold_builtin_strncat): Same.
8217 * gimple.c (gimple_build_call_from_tree): Set call location.
8218 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
8219 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
8220 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
8221 (handle_builtin_strlen): Use strlen_to_stridx.
8222 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
8224 Use strlen_to_stridx.
8225 (pass_strlen::execute): Release strlen_to_stridx.
8226 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
8227 (-Wstringop-truncation): Document new option.
8229 2017-11-10 Martin Liska <mliska@suse.cz>
8231 PR gcov-profile/82702
8232 * gcov.c (main): Handle intermediate files in a different
8234 (get_gcov_intermediate_filename): New function.
8235 (output_gcov_file): Remove support of intermediate files.
8236 (generate_results): Allocate intermediate file.
8237 (release_structures): Clean-up properly fn_end.
8238 (output_intermediate_file): Start iterating with line 1.
8240 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8242 PR tree-optimization/82929
8243 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8244 ops_swapped_p non-static data member.
8245 (store_immediate_info::store_immediate_info): Clear it.
8246 (imm_store_chain_info::coalesce_immediate_stores): If swapping
8247 ops set ops_swapped_p.
8248 (count_multiple_uses): Handle ops_swapped_p.
8250 2017-11-10 Martin Liska <mliska@suse.cz>
8252 * coverage.c (coverage_init): Stream information about
8253 support of has_unexecuted_blocks.
8254 * doc/gcov.texi: Document that.
8255 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
8256 * gcov.c (read_graph_file): Likewise.
8257 (output_line_beginning): Fix a small issue with
8260 2017-11-10 Bin Cheng <bin.cheng@arm.com>
8262 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
8263 reference of trivial component.
8265 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8268 * gimple-ssa-store-merging.c
8269 (pass_store_merging::terminate_all_aliasing_chains): For
8270 gimple_store_p stmts also call refs_output_dependent_p.
8272 PR rtl-optimization/82913
8273 * compare-elim.c (try_merge_compare): Punt if def_insn is not
8276 2017-11-09 Jeff Law <law@redhat.com>
8278 * vr-values.h: New file with vr_values class.
8279 * tree-vrp.c: Include vr-values.h
8280 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
8281 data objects into the vr_values class.
8282 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
8283 (get_value_range): Make it a member function within vr_values class.
8284 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
8285 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
8286 (op_with_constant_singleton_value_range): Likewise.
8287 (extract_range_for_var_from_comparison_expr): Likewise.
8288 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
8289 (extract_range_from_binary_expr): Likewise.
8290 (extract_range_from_unary_expr): Likewise.
8291 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
8292 (check_for_binary_op_overflow, extract_range_basic): Likewise.
8293 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
8294 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
8295 (compare_name_with_value, compare_names): Likewise.
8296 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
8297 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
8298 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
8299 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
8300 (extract_range_from_phi_node): Likewise.
8301 (simplify_truth_ops_using_ranges): Likewise.
8302 (simplify_div_or_mod_using_ranges): Likewise.
8303 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
8304 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
8305 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
8306 (simplify_float_conversion_using_ranges): Likewise.
8307 (simplify_internal_call_using_ranges): Likewise.
8308 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
8309 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
8310 poitner via x_vr_values for calls into gimple folder.
8311 (vrp_initialize_lattice): Make this the vr_values ctor.
8312 (vrp_free_lattice): Make this the vr_values dtor.
8313 (set_vr_value): New function.
8314 (class vrp_prop): Add vr_values data member. Add various member
8315 functions as well as member functions that delegate to vr_values.
8316 (check_array_ref): Make a member function within vrp_prop class.
8317 (search_for_addr_array, vrp_initialize): Likewise.
8318 (vrp_finalize): Likewise. Revamp to avoid direct access to
8319 vr_value, values_propagated, etc.
8320 (check_array_bounds): Extract vrp_prop class instance pointer from
8321 walk info structure. Use it to call member functions.
8322 (check_all_array_refs): Make a member function within vrp_prop class.
8323 Smuggle class instance pointer via walk info structure.
8324 (x_vr_values): New local static.
8325 (vrp_valueize): Use x_vr_values to get class instance.
8326 (vr_valueize_1): Likewise.
8327 (class vrp_folder): Add vr_values data member. Add various member
8328 functions as well as member functions that delegate to vr_values.
8329 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
8330 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
8331 class instance from vr_values. Use it to call member functions.
8332 (vrp_dom_walker): Add vr_values data member.
8333 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
8334 instance via x_vr_values.
8335 (identify_jump_threads): Accept vr_values as argument. Store
8336 it into the walker structure.
8337 (evrp_dom_walker): Add vr_values class data member. Add various
8339 (evrp_dom_walker::try_find_new_range): Use vr_values data
8340 member to access the memory allocator.
8341 (evrp_dom_walker::before_dom_children): Store vr_values class
8342 instance into the vrp_folder class.
8343 (evrp_dom_walker::push_value_range): Rework to avoid direct
8344 access to num_vr_values and vr_value.
8345 (evrp_dom_walker::pop_value_range): Likewise.
8346 (execute_early_vrp): Remove call to vrp_initialize_lattice.
8347 Use vr_values to get to dump_all_value_ranges member function.
8348 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
8349 and simplify_cond_using_ranges_2 via vrp_prop class instance.
8350 Pass vr_values class instance down to identify_jump_threads.
8351 Remove call to vrp_free_lattice.
8352 (debug_all_value_ranges): Remove.
8354 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
8355 (vrp_folder): Likewise.
8357 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
8358 Get it from the existing bitmap instead.
8359 (vrp_intersect_ranges_1): Likewise.
8361 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8363 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8365 (store_immediate_info::store_immediate_info): Add bitnotp argument,
8366 set bit_not_p to it.
8367 (imm_store_chain_info::coalesce_immediate_stores): Break group
8368 if bit_not_p is different.
8369 (count_multiple_uses, split_group,
8370 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
8371 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
8372 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
8373 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
8374 to store_immediate_info ctor.
8376 2017-11-09 Jim Wilson <jimw@sifive.com>
8378 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
8379 (scan_prog_file): Likewise.
8381 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8383 * bb-reorder.c (max_entry_frequency): Remove.
8384 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
8385 connect_traces, push_to_next_round_p): Remove prototypes.
8386 (find_traces_1_round): Use counts only.
8387 (push_to_next_round_p): Likewise.
8388 (find_traces): Likewise.
8389 (rotate_loop): Likewise.
8390 (find_traces_1_round): Likewise.
8391 (connect_traces): Likewise.
8392 (edge_order): Likewise.
8394 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
8396 * config/arm/arm.c (output_return_instruction): Add comments to
8397 indicate requirement for cmse_nonsecure_entry return to account
8398 for the size of clearing instruction output here.
8399 (thumb_exit): Likewise.
8400 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
8401 return in hardfloat mode.
8403 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
8405 * config/rs6000/rs6000.c (machine_function): Add a bool,
8406 "toc_is_wrapped_separately".
8407 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
8408 if it wasn't explicitly set or unset, we are optimizing for speed, and
8409 doing separate shrink-wrapping.
8410 (rs6000_get_separate_components): Enable the TOC component if
8411 saving the TOC register in the prologue.
8412 (rs6000_components_for_bb): Handle the TOC component.
8413 (rs6000_emit_prologue_components): Store the TOC register where needed.
8414 (rs6000_set_handled_components): Mark TOC as handled, if handled.
8415 (rs6000_emit_prologue): Don't save the TOC if that is already done.
8417 2017-11-09 Martin Jambor <mjambor@suse.cz>
8419 * ipa-param-manipulation.c: New file.
8420 * ipa-param-manipulation.h: Likewise.
8421 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
8422 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
8423 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
8424 (ipa_parm_adjustment): Likewise.
8425 (ipa_parm_adjustment_vec): Likewise.
8426 (ipa_get_vector_of_formal_parms): Moved declaration to
8427 ipa-param-manipulation.h.
8428 (ipa_get_vector_of_formal_parm_types): Likewise.
8429 (ipa_modify_formal_parameters): Likewise.
8430 (ipa_modify_call_arguments): Likewise.
8431 (ipa_combine_adjustments): Likewise.
8432 (ipa_dump_param_adjustments): Likewise.
8433 (ipa_modify_expr): Likewise.
8434 (ipa_get_adjustment_candidate): Likewise.
8435 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
8436 ipa-param-manipulation.c.
8437 (ipa_get_vector_of_formal_parm_types): Likewise.
8438 (ipa_modify_formal_parameters): Likewise.
8439 (ipa_modify_call_arguments): Likewise.
8440 (ipa_modify_expr): Likewise.
8441 (get_ssa_base_param): Likewise.
8442 (ipa_get_adjustment_candidate): Likewise.
8443 (index_in_adjustments_multiple_times_p): Likewise.
8444 (ipa_combine_adjustments): Likewise.
8445 (ipa_dump_param_adjustments): Likewise.
8446 * tree-sra.c: Also include ipa-param-manipulation.h
8447 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
8450 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8451 Alan Hayward <alan.hayward@arm.com>
8452 David Sherwood <david.sherwood@arm.com>
8454 * doc/sourcebuild.texi (vect_masked_store): Document.
8456 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8457 Alan Hayward <alan.hayward@arm.com>
8458 David Sherwood <david.sherwood@arm.com>
8460 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
8462 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8463 Alan Hayward <alan.hayward@arm.com>
8464 David Sherwood <david.sherwood@arm.com>
8466 * doc/sourcebuild.texi (vect_variable_length): Document.
8468 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8469 Alan Hayward <alan.hayward@arm.com>
8470 David Sherwood <david.sherwood@arm.com>
8472 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
8474 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8475 Alan Hayward <alan.hayward@arm.com>
8476 David Sherwood <david.sherwood@arm.com>
8478 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
8480 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8481 Alan Hayward <alan.hayward@arm.com>
8482 David Sherwood <david.sherwood@arm.com>
8484 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
8485 previously undocumented selectors.
8486 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
8488 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8490 * doc/rtl.texi (const_vector): Say that elements can be
8491 const_wide_ints too.
8492 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
8493 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
8494 (gen_vec_duplicate): Use it instead of CONSTANT_P.
8495 * optabs.c (expand_vector_broadcast): Likewise.
8497 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8498 Alan Hayward <alan.hayward@arm.com>
8499 David Sherwood <david.sherwood@arm.com>
8501 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
8502 scaled index even if the unscaled address was invalid.
8503 Don't increase the complexity of using a scale in that case.
8505 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8506 Alan Hayward <alan.hayward@arm.com>
8507 David Sherwood <david.sherwood@arm.com>
8509 * doc/rtl.texi: Rewrite the subreg rules so that they partition
8510 the inner register into REGMODE_NATURAL_SIZE bytes rather than
8511 UNITS_PER_WORD bytes.
8512 * emit-rtl.c (validate_subreg): Divide subregs into blocks
8513 based on REGMODE_NATURAL_SIZE of the inner mode.
8514 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
8515 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
8516 * expmed.c (lowpart_bit_field_p): Divide the value up into
8517 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
8518 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
8519 whether something is likely to occupy more than one register.
8521 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8524 * ipa-inline-transform.c (update_noncloned_frequencies): Use
8525 profile_count::adjust_for_ipa_scaling.
8526 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
8527 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
8529 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
8531 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8533 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
8534 (split_group): Add total_orig and total_new arguments, estimate the
8535 number of statements related to the store group without store merging
8536 and with store merging.
8537 (imm_store_chain_info::output_merged_store): Adjust split_group
8538 callers, punt if estimated number of statements with store merging
8539 is not smaller than estimated number of statements without it.
8541 (handled_load): Remove has_single_use checks.
8542 (pass_store_merging::process_store): Likewise.
8544 2017-11-09 Richard Biener <rguenther@suse.de>
8546 PR tree-optimization/82902
8547 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
8549 2017-11-09 Martin Liska <mliska@suse.cz>
8552 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
8555 2017-11-09 Martin Liska <mliska@suse.cz>
8557 PR tree-optimization/82669
8558 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
8560 2017-11-09 Martin Liska <mliska@suse.cz>
8562 PR gcov-profile/48463
8563 * coverage.c (coverage_begin_function): Output also end locus
8564 of a function and information whether the function is
8566 * gcov-dump.c (tag_function): Parse and print the information.
8567 * gcov.c (INCLUDE_MAP): Add include.
8568 (INCLUDE_SET): Likewise.
8569 (struct line_info): Move earlier in the source file because
8570 of vector<line_info> in function_info structure.
8571 (line_info::line_info): Likewise.
8572 (line_info::has_block): Likewise.
8573 (struct source_info): Add new member index.
8574 (source_info::get_functions_at_location): New function.
8575 (function_info::group_line_p): New function.
8576 (output_intermediate_line): New function.
8577 (output_intermediate_file): Use the mentioned function.
8578 (struct function_start): New.
8579 (struct function_start_pair_hash): Likewise.
8580 (process_file): Add code that identifies group functions.
8581 Assign lines either to global or function scope.
8582 (generate_results): Skip artificial functions.
8583 (find_source): Assign index for each source file.
8584 (read_graph_file): Read new flag artificial and end_line.
8585 (add_line_counts): Assign it either to global of function scope.
8586 (accumulate_line_counts): Isolate core of the function to
8587 accumulate_line_info and call it for both function and global
8589 (accumulate_line_info): New function.
8590 (output_line_beginning): Fix GNU coding style.
8591 (print_source_line): New function.
8592 (output_line_details): Likewise.
8593 (output_function_details): Likewise.
8594 (output_lines): Iterate both source (global) scope and function
8596 (struct function_line_start_cmp): New class.
8597 * doc/gcov.texi: Reflect changes in documentation.
8599 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8602 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
8603 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
8604 and similarly for not instead of neg.
8606 2017-11-08 Andi Kleen <ak@linux.intel.com>
8608 * config/i386/i386.opt: Add -mforce-indirect-call.
8609 * config/i386/predicates.md: Check for flag_force_indirect_call.
8610 * doc/invoke.texi: Document -mforce-indirect-call
8612 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
8614 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
8616 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
8617 config/riscv/riscv.c (predict.h): New include.
8618 (riscv_slow_unaligned_access_p): No longer static.
8619 (riscv_block_move_straight): Add require.
8620 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8622 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8625 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
8626 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
8627 nonimmediate_operand predicate for operand 1 instead of
8630 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8632 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
8634 * config/aarch64/constraints.md (Uml): New constraint.
8635 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
8638 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8640 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
8641 of two vec_duplicates into a vec_concat.
8643 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8645 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8646 Simplify vec_merge of vec_duplicate and vec_concat.
8647 * config/aarch64/constraints.md (Utq): New constraint.
8648 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
8651 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8653 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8654 Simplify vec_merge of vec_duplicate and const_vector.
8655 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
8657 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
8658 mode iterator. Update predicate on operand 1 to
8659 handle non-const_vec constants. Delete constraints.
8660 (*aarch64_combinez_be<mode>): Likewise for operand 2.
8662 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8664 PR tree-optimization/78821
8665 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
8667 (store_operand_info::store_operand_info): Initialize it to false.
8668 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
8669 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
8670 store in the group, and if chain_info is non-NULL, to ignore altogether
8672 (compatible_load_p): Fail if bit_not_p does not match.
8673 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
8674 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
8675 (pass_store_merging::process_store): Adjust
8676 terminate_all_aliasing_chains calls to pass NULL in all current spots,
8677 call terminate_all_aliasing_chains newly when adding a store into
8678 a chain with non-NULL chain_info.
8680 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8682 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
8684 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8686 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8688 (aarch64_layout_frame): Initialise emit_frame_chain.
8689 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
8690 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
8692 2017-11-08 Martin Liska <mliska@suse.cz>
8694 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
8695 of gimple_call_internal_p.
8697 2017-11-07 Tom de Vries <tom@codesourcery.com>
8699 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
8701 2017-11-07 Tom de Vries <tom@codesourcery.com>
8703 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
8706 2017-11-08 Martin Liska <mliska@suse.cz>
8709 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
8711 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8713 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
8715 2017-11-07 Jakub Jelinek <jakub@redhat.com>
8718 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
8719 (*cmp<mode>_ccz_1): New insn with $k alternative.
8722 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
8723 RTX_COMM_COMPARE as commutative as well.
8724 (ix86_binary_operator_ok): Formatting fix.
8725 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
8726 *<code><mode>3<mask_name><round_saeonly_name>,
8727 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
8728 *<s>mul<mode>3_highpart<mask_name>,
8729 *vec_widen_umult_even_v16si<mask_name>,
8730 *vec_widen_umult_even_v8si<mask_name>,
8731 *vec_widen_umult_even_v4si<mask_name>,
8732 *vec_widen_smult_even_v16si<mask_name>,
8733 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
8734 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
8735 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
8736 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
8737 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
8738 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
8739 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
8740 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
8741 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
8742 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
8743 ix86_binary_operator_ok. Formatting fixes.
8744 (*<plusminus_insn><mode>3<mask_name><round_name>,
8745 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
8748 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
8750 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
8751 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
8752 eq and ne if TARGET_ISEL.
8753 (cmp): New code_iterator.
8754 (UNS, UNSU_, UNSIK): New code_attrs.
8755 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
8756 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
8758 ("eq<mode>3"): ... this.
8759 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
8761 ("ne<mode>3"): ... this.
8763 2017-11-07 Julia Koval <julia.koval@intel.com>
8766 * common/config/i386/i386-common.c
8767 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
8768 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
8769 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
8770 * config/i386/i386.opt: Ditto.
8771 * config/i386/i386.c (ix86_target_string): Ditto.
8772 (ix86_option_override_internal): Ditto.
8773 (ix86_init_mpx_builtins): Move MPX to args2.
8774 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
8775 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8776 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8777 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8778 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8779 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8780 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
8782 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
8785 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
8786 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
8787 (zero-extendsidi peephole2): Remove peephole.
8789 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8792 * stor-layout.c: Include attribs.h.
8793 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8794 explicit lookup of "aligned" attribute.
8796 2017-11-07 Andrew Waterman <andrew@sifive.com>
8798 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8799 (riscv_expand_block_move): Likewise.
8800 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8802 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8803 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8804 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8805 (riscv_adjust_block_mem): Likewise.
8806 (riscv_block_move_loop): Likewise.
8807 (riscv_expand_block_move): Likewise.
8808 * config/riscv/riscv.md (movmemsi): New pattern.
8810 2017-11-07 Michael Clark <michaeljclark@mac.com>
8812 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8813 (MUSL_DYNAMIC_LINKER): Likewise.
8815 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8817 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8818 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8821 2017-11-07 Richard Biener <rguenther@suse.de>
8823 * match.pd: Fix build.
8825 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8826 Jackson Woodruff <jackson.woodruff@arm.com>
8828 PR tree-optimization/71026
8829 * match.pd: Canonicalize negate in division.
8831 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8834 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8836 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8838 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8839 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8841 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8843 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8844 non-scalar integral types.
8845 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8846 (-(A-B), -(~A)): New transformations.
8848 2017-11-07 Tom de Vries <tom@codesourcery.com>
8850 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8851 semicolon after "do {} while (0)".
8852 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8853 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8854 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8855 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8856 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8858 2017-11-07 Tom de Vries <tom@codesourcery.com>
8860 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8861 after "do {} while (0)".
8862 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8863 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8864 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8865 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8866 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8868 2017-11-07 Tom de Vries <tom@codesourcery.com>
8871 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8873 (arm_rtx_costs_internal): Add missing semicolon after
8874 HANDLE_NARROW_SHIFT_ARITH call.
8876 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8878 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8879 disable isel if it was not set explicitly.
8881 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8883 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8884 (add_type_attribute) likewise.
8886 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8888 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8889 of struct ix86_frame.
8890 (ix86_initial_elimination_offset): Likewise.
8891 (ix86_expand_split_stack_prologue): Likewise.
8893 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8895 * tree-vrp.h (enum value_range_type): Update stale comment.
8897 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8898 Alan Hayward <alan.hayward@arm.com>
8899 David Sherwood <david.sherwood@arm.com>
8901 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8902 (aarch64_expand_vec_perm_const): Take the number of units too.
8903 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8904 (aarch64_expand_vec_perm_const): Likewise.
8905 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8906 (vec_perm<mode>): Update accordingly.
8908 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8909 Alan Hayward <alan.hayward@arm.com>
8910 David Sherwood <david.sherwood@arm.com>
8912 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8913 Take the number of units too.
8914 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8915 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8916 but check for a vector mode before rather than after the call.
8917 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8918 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8919 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8920 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8921 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8922 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8923 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8924 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8925 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8926 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8927 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8928 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8929 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8930 (aarch64_sqdmull2_n<mode>): Update accordingly.
8932 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8933 Alan Hayward <alan.hayward@arm.com>
8934 David Sherwood <david.sherwood@arm.com>
8936 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8937 the number of units too.
8938 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8939 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8940 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8941 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8942 (vec_store_lanesxi<mode>): Update accordingly.
8944 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8945 Alan Hayward <alan.hayward@arm.com>
8946 David Sherwood <david.sherwood@arm.com>
8948 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8949 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8950 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8951 of units rather than the mode.
8952 * config/aarch64/iterators.md (nunits): New mode attribute.
8953 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8954 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8955 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8956 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8957 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8958 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8959 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8960 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8961 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8962 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8963 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8964 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8965 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8966 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8967 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8968 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8969 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8970 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8971 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8972 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8973 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8974 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8975 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8976 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8977 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8978 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8979 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8980 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8981 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8982 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8983 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8984 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8985 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8986 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8987 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8988 (aarch64_simd_vec_setv2di): Likewise.
8990 2017-11-06 Carl Love <cel@us.ibm.com>
8992 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8994 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8995 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8997 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8999 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
9000 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
9001 Add power 8 macro expansions.
9002 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
9003 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
9004 power 8 instructions. (VSX_XXBR): Add iterator.
9006 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
9008 * config/arm/arm.md (predicable_short_it): Change default to "no",
9009 improve documentation, remove uses that are identical to the default.
9010 (enabled_for_depr_it): Rename to enabled_for_short_it.
9011 * gcc/config/arm/arm-fixed.md (predicable_short_it):
9012 Remove default uses.
9013 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
9014 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
9015 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
9016 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
9018 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
9021 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
9022 float128 helper macros, which are no longer used after deleting
9023 the old 'q' built-in functions, and moving the round to odd
9024 built-in functions to being special built-in functions.
9025 (BU_FLOAT128_2): Likewise.
9026 (BU_FLOAT128_1_HW): Likewise.
9027 (BU_FLOAT128_2_HW): Likewise.
9028 (BU_FLOAT128_3_HW): Likewise.
9029 (FABSQ): Delete old 'q' built-in functions.
9030 (COPYSIGNQ): Likewise.
9031 (SQRTF128_ODD): Move round to odd built-in functions to be
9032 special built-in functions, so that we can handle
9033 -mabi=ieeelongdouble.
9034 (TRUNCF128_ODD): Likewise.
9035 (ADDF128_ODD): Likewise.
9036 (SUBF128_ODD): Likewise.
9037 (MULF128_ODD): Likewise.
9038 (DIVF128_ODD): Likewise.
9039 (FMAF128_ODD): Likewise.
9040 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
9041 built-in names to 'f128'.
9042 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
9043 old 'q' built-in functions, as the machine independent code for
9044 'f128' built-in functions handles this.
9045 (rs6000_expand_builtin): Add expansion for float128 round to odd
9046 functions, keying off on -mabi=ieeelongdouble of whether to use
9047 the KFmode or TFmode variant.
9048 (rs6000_init_builtins): Initialize the _Float128 round to odd
9050 * doc/extend.texi (PowerPC Built-in Functions): Document the old
9051 _Float128 'q' built-in functions are now mapped into the new
9052 'f128' built-in functions.
9054 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
9056 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
9057 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
9059 2017-11-06 David Malcolm <dmalcolm@redhat.com>
9062 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
9063 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
9064 * toplev.c: Include "ipa-fnsummary.h".
9065 (toplev::finalize): Call ipa_fnsummary_c_finalize.
9067 2017-11-06 Jakub Jelinek <jakub@redhat.com>
9069 PR tree-optimization/82838
9070 * gimple-ssa-store-merging.c
9071 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
9072 on a separate gimple_seq which is then appended to seq.
9074 2017-11-06 Jeff Law <law@redhat.com>
9077 * config/i386/i386.c (PROBE_INTERVAL): Remove.
9078 (get_probe_interval): New functions.
9079 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
9080 (ix86_adjust_stack_and_probe): Likewise.
9081 (output_adjust_stack_and_probe): Likewise.
9082 (ix86_emit_probe_stack_range): Likewise.
9083 (ix86_expand_prologue): Likewise.
9085 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
9087 PR tree-optimization/82816
9088 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
9089 if the modes of the two types are the same.
9090 (convert_plusminus_to_widen): Likewise.
9092 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9094 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
9096 (usadv16qi): New define_expand.
9097 (usadv8hi): New define_expand.
9099 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
9102 * ipa-inline-transform.c (update_noncloned_frequencies): Always
9104 (inline_transform): Likewise.
9105 * predict.c (counts_to_freqs): Remove useless conditional.
9106 * profile-count.h (profile_count::apply_scale): Move sanity check.
9107 * tree-inline.c (copy_bb): Always scale.
9108 (copy_cfg_body): Likewise.
9110 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
9113 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
9115 (*cmp_ite0): Add enabled_for_depr_it attribute.
9116 (*cmp_ite1): Likewise.
9118 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
9120 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
9123 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
9125 * tree-vrp.c (vrp_int_const_binop): Return true on success and
9126 return the value by pointer.
9127 (extract_range_from_multiplicative_op_1): Update accordingly.
9128 Return as soon as an operation fails.
9130 2017-11-05 Tom de Vries <tom@codesourcery.com>
9133 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
9134 (DEF_SANITIZER_BUILTIN): ... here.
9135 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
9136 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
9138 2017-11-05 Tom de Vries <tom@codesourcery.com>
9141 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
9143 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
9144 ASM_OUTPUT_BEFORE_CASE_LABEL call.
9145 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
9147 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9148 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9149 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9151 2017-11-05 Tom de Vries <tom@codesourcery.com>
9154 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
9157 2017-11-04 Michael Clark <michaeljclark@mac.com>
9159 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
9160 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
9162 (*addsi3_extended): Likewise.
9163 (*addsi3_extended2): Likewise.
9164 (<optab>si3): Likewise.
9165 (<optab>di3): Likewise.
9166 (<optab><mode>3): Likewise.
9167 (<*optabe>si3_internal): Likewise.
9168 (zero_extendqi<SUPERQI:mode>2): Likewise.
9169 (*add<mode>hi3): Likewise.
9170 (*xor<mode>hi3): Likewise.
9171 (<optab>di3): Likewise.
9172 (*<optab>si3_extend): Likewise.
9173 (*sge<u>_<X:mode><GPR:mode>): Likewise.
9174 (*slt<u>_<X:mode><GPR:mode>): Likewise.
9175 (*sle<u>_<X:mode><GPR:mode>): Likewise.
9177 2017-11-04 Andrew Waterman <andrew@sifive.com>
9179 * config/riscv/riscv.c (riscv_option_override): Conditionally set
9180 TARGET_STRICT_ALIGN based upon -mtune argument.
9182 2017-11-04 Andrew Waterman <andrew@sifive.com>
9184 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
9186 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
9188 * config/i386/i386.c (choose_basereg): Use optional scratch
9189 register and add assertion.
9190 (x86_emit_outlined_ms2sysv_save): Use scratch register when
9191 needed, and don't allocate stack.
9192 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
9193 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
9194 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
9196 2017-11-03 Jeff Law <law@redhat.com>
9198 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
9199 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
9200 to probe at the start of a noreturn function.
9202 2017-11-03 Jakub Jelinek <jakub@redhat.com>
9204 PR tree-optimization/78821
9205 * gimple-ssa-store-merging.c: Update the file comment.
9206 (MAX_STORE_ALIAS_CHECKS): Define.
9207 (struct store_operand_info): New type.
9208 (store_operand_info::store_operand_info): New constructor.
9209 (struct store_immediate_info): Add rhs_code and ops data members.
9210 (store_immediate_info::store_immediate_info): Add rhscode, op0r
9211 and op1r arguments to the ctor, initialize corresponding data members.
9212 (struct merged_store_group): Add load_align_base and load_align
9214 (merged_store_group::merged_store_group): Initialize them.
9215 (merged_store_group::do_merge): Update them.
9216 (merged_store_group::apply_stores): Pick the constant for
9217 encode_tree_to_bitpos from one of the two operands, or skip
9218 encode_tree_to_bitpos if neither operand is a constant.
9219 (class pass_store_merging): Add process_store method decl. Remove
9220 bool argument from terminate_all_aliasing_chains method decl.
9221 (pass_store_merging::terminate_all_aliasing_chains): Remove
9222 var_offset_p argument and corresponding handling.
9223 (stmts_may_clobber_ref_p): New function.
9224 (compatible_load_p): New function.
9225 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
9226 if there is overlap and rhs_code is not INTEGER_CST. For
9227 non-overlapping stores terminate group if rhs is not mergeable.
9228 (get_alias_type_for_stmts): Change first argument from
9229 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
9230 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
9231 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
9233 (get_location_for_stmts): Change first argument from
9234 auto_vec<gimple *> & to vec<gimple *> &.
9235 (struct split_store): Remove orig_stmts data member, add orig_stores.
9236 (split_store::split_store): Create orig_stores rather than orig_stmts.
9237 (find_constituent_stmts): Renamed to ...
9238 (find_constituent_stores): ... this. Change second argument from
9239 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
9240 to info structures rather than the statements.
9241 (split_group): Rename ALLOW_UNALIGNED argument to
9242 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
9243 it. Adjust find_constituent_stores caller.
9244 (imm_store_chain_info::output_merged_store): Handle rhs_code other
9245 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
9246 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
9247 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
9248 (mem_valid_for_store_merging): New function.
9249 (handled_load): New function.
9250 (pass_store_merging::process_store): New method.
9251 (pass_store_merging::execute): Use process_store method. Adjust
9252 terminate_all_aliasing_chains caller.
9254 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9256 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9257 Return true for more constants, symbols and label references.
9258 (aarch64_valid_floating_const): Remove unused function.
9260 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9263 * config/aarch64/aarch64.c (aarch64_layout_frame):
9264 Undo forcing of LR at bottom of frame.
9266 2017-11-03 Jeff Law <law@redhat.com>
9269 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
9270 for int_registers_saved.
9272 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
9273 extracted from tree-ssa-dom.c.
9274 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
9275 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
9276 (record_equivalences_from_incoming_edge): Add additional argument
9277 to single_pred_edge_ignoring_loop_edges call.
9278 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
9279 (uncprop_dom_walker::before_dom_children): Add additional argument
9280 to single_pred_edge_ignoring_loop_edges call.
9281 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
9282 single_pred_edge_ignoring_loop_edges rather than open coding.
9283 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
9285 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
9287 * match.pd (-(-A)): Rewrite.
9289 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9291 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
9292 (rs6000_emit_int_cmove): New declaration.
9293 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
9294 (rs6000_emit_sISEL): Delete.
9295 (rs6000_emit_int_cmove): Make non-static.
9296 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
9297 instead of rs6000_emit_sISEL.
9299 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
9301 * asan.c (create_cond_insert_point): Maintain profile.
9302 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
9304 * basic-block.h (struct basic_block_def): Remove frequency.
9305 (EDGE_FREQUENCY): Use to_frequency
9306 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
9308 (find_traces): Update to use to_frequency.
9309 (find_traces_1_round): Likewise; use only IPA counts.
9310 (bb_to_key): Likewise.
9311 (connect_traces): Use IPA counts only.
9312 (copy_bb_p): Update to use to_frequency.
9313 (fix_up_crossing_landing_pad): Likewise.
9314 (sanitize_hot_paths): Likewise.
9315 * bt-load.c (basic_block_freq): Likewise.
9316 * cfg.c (init_flow): Set count_max to uninitialized.
9317 (check_bb_profile): Remove frequencies; check counts.
9318 (dump_bb_info): Do not dump frequencies.
9319 (update_bb_profile_for_threading): Update counts only.
9320 (scale_bbs_frequencies_int): Likewise.
9321 (MAX_SAFE_MULTIPLIER): Remove.
9322 (scale_bbs_frequencies_gcov_type): Update counts only.
9323 (scale_bbs_frequencies_profile_count): Update counts only.
9324 (scale_bbs_frequencies): Update counts only.
9325 * cfg.h (struct control_flow_graph): Add count-max.
9326 (update_bb_profile_for_threading): Update prototype.
9327 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
9328 (find_many_sub_basic_blocks): Likewise.
9329 * cfgcleanup.c (try_forward_edges): Likewise.
9330 (try_crossjump_to_edge): Likewise.
9331 * cfgexpand.c (expand_gimple_cond): Likewise.
9332 (expand_gimple_tailcall): Likewise.
9333 (construct_init_block): Likewise.
9334 (construct_exit_block): Likewise.
9335 * cfghooks.c (verify_flow_info): Check consistency of counts.
9336 (dump_bb_for_graph): Do not dump frequencies.
9337 (split_block_1): Do not update frequencies.
9338 (split_edge): Do not update frequencies.
9339 (make_forwarder_block): Do not update frequencies.
9340 (duplicate_block): Do not update frequencies.
9341 (account_profile_record): Do not update frequencies.
9342 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
9343 for global heuristics.
9344 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
9345 (expected_loop_iterations_unbounded): Use counts only.
9346 * cfgloopmanip.c (scale_loop_profile): Simplify.
9347 (create_empty_loop_on_edge): Simplify
9349 (duplicate_loop_to_header_edge): Simplify
9350 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
9351 (update_br_prob_note): Take care of removing note when profile
9353 (relink_block_chain): Do not dump frequency.
9354 (rtl_account_profile_record): Use to_frequency.
9355 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
9356 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
9357 (cgraph_update_edges_for_call_stmt_node): Likewise.
9358 (cgraph_edge::verify_count_and_frequency): Update.
9359 (cgraph_node::verify_node): Temporarily disable frequency verification.
9360 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
9361 to_cgraph_frequency.
9362 (cgraph_edge::rebuild_edges): Convert to ipa counts.
9363 * cgraphunit.c (init_lowered_empty_function): Do not initialize
9365 (cgraph_node::expand_thunk): Update profile.
9366 * except.c (dw2_build_landing_pads): Do not update frequency.
9367 * final.c (compute_alignments): Use to_frequency.
9368 (dump_basic_block_info): Do not dump frequency.
9369 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
9370 (dump_gimple_bb_header): Do not dump frequency.
9371 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
9373 * gimple-streamer-in.c (input_bb): Do not stream frequency.
9374 * gimple-streamer-out.c (output_bb): Do not stream frequency.
9375 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
9376 (init_before_recovery): Do not update frequency.
9377 (sched_create_recovery_edges): Do not update frequency.
9378 * hsa-gen.c (convert_switch_statements): Do not update frequency.
9379 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
9380 (ipa_cp_c_finalize): Set max_count to uninitialized.
9381 * ipa-fnsummary.c (get_minimal_bb): Use counts.
9382 (param_change_prob): Use counts.
9383 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
9385 * ipa-split.c (consider_split): Use to_frequency.
9386 (split_function): Use to_frequency.
9387 * ira-build.c (loop_compare_func): Likewise.
9388 (mark_loops_for_removal): Likewise.
9389 (mark_all_loops_for_removal): Likewise.
9390 * loop-doloop.c (doloop_modify): Do not update frequency.
9391 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
9393 * lto-streamer-in.c (input_function): Update count_max.
9394 * omp-expand.c (expand_omp_taskreg): Update count_max.
9395 * omp-simd-clone.c (simd_clone_adjust): Update profile.
9396 * predict.c (maybe_hot_frequency_p): Use to_frequency.
9397 (maybe_hot_count_p): Use ipa counts only.
9398 (maybe_hot_bb_p): Simplify.
9399 (maybe_hot_edge_p): Simplify.
9400 (probably_never_executed): Do not take frequency argument.
9401 (probably_never_executed_bb_p): Do not pass frequency.
9402 (probably_never_executed_edge_p): Likewise.
9403 (combine_predictions_for_bb): Check that profile is nonzero.
9404 (propagate_freq): Do not set frequency.
9405 (drop_profile): Simplify.
9406 (counts_to_freqs): Simplify.
9407 (expensive_function_p): Use to_frequency.
9408 (propagate_unlikely_bbs_forward): Simplify.
9409 (determine_unlikely_bbs): Simplify.
9410 (estimate_bb_frequencies): Add hack to silence graphite issues.
9411 (compute_function_frequency): Use ipa counts.
9412 (pass_profile::execute): Update.
9413 (rebuild_frequencies): Use counts only.
9414 (force_edge_cold): Use counts only.
9415 * profile-count.c (profile_count::dump): Dump new count types.
9416 (profile_count::differs_from_p): Check compatiblity.
9417 (profile_count::to_frequency): New function.
9418 (profile_count::to_cgraph_frequency): New function.
9419 * profile-count.h (struct function): Declare.
9420 (enum profile_quality): Add profile_guessed_local and
9421 profile_guessed_global0.
9422 (class profile_proability): Decrease number of bits to 29;
9423 update from_reg_br_prob_note and to_reg_br_prob_note.
9424 (class profile_count: Update comment; decrease number of bits
9425 to 61. Check compatibility.
9426 (profile_count::compatible_p): New private member function.
9427 (profile_count::ipa_p): New member function.
9428 (profile_count::operator<): Handle global zero correctly.
9429 (profile_count::operator>): Handle global zero correctly.
9430 (profile_count::operator<=): Handle global zero correctly.
9431 (profile_count::operator>=): Handle global zero correctly.
9432 (profile_count::nonzero_p): New member function.
9433 (profile_count::force_nonzero): New member function.
9434 (profile_count::max): New member function.
9435 (profile_count::apply_scale): Handle IPA scalling.
9436 (profile_count::guessed_local): New member function.
9437 (profile_count::global0): New member function.
9438 (profile_count::ipa): New member function.
9439 (profile_count::to_frequency): Declare.
9440 (profile_count::to_cgraph_frequency): Declare.
9441 * profile.c (OVERLAP_BASE): Delete.
9442 (compute_frequency_overlap): Delete.
9443 (compute_branch_probabilities): Do not use compute_frequency_overlap.
9444 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
9445 * sched-ebb.c (rank): Use counts only.
9446 * shrink-wrap.c (handle_simple_exit): Use counts only.
9447 (try_shrink_wrapping): Use counts only.
9448 (place_prologue_for_one_component): Use counts only.
9449 * tracer.c (find_best_predecessor): Use to_frequency.
9450 (find_trace): Use to_frequency.
9451 (tail_duplicate): Use to_frequency.
9452 * trans-mem.c (expand_transaction): Do not update frequency.
9453 * tree-call-cdce.c: Do not update frequency.
9454 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
9455 (gimple_merge_blocks): Likewise.
9456 (gimple_split_edge): Likewise.
9457 (gimple_duplicate_sese_region): Likewise.
9458 (gimple_duplicate_sese_tail): Likewise.
9459 (move_sese_region_to_fn): Likewise.
9460 (gimple_account_profile_record): Likewise.
9461 (insert_cond_bb): Likewise.
9462 * tree-complex.c (expand_complex_div_wide): Likewise.
9463 * tree-eh.c (lower_resx): Update profile.
9464 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
9466 (initialize_cfun): Do not initialize frequencies
9467 (freqs_to_counts): Delete.
9468 (copy_cfg_body): Ignore count parameter.
9469 (copy_body): Update.
9470 (expand_call_inline): Update count_max.
9471 (optimize_inline_calls): Update count_max.
9472 (tree_function_versioning): Update count_max.
9473 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
9474 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
9476 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
9477 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
9478 (try_peel_loop): Likewise.
9479 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
9481 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
9482 (tree_transform_and_unroll_loop): Do not use frequencies
9483 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
9484 Use reliable prediction only.
9485 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
9486 * tree-ssa-sink.c (select_best_block): Use to_frequency.
9487 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
9488 probability scaling.
9489 * tree-ssa-threadupdate.c (create_block_for_threading): Do
9490 not update frequency
9491 (any_remaining_duplicated_blocks): Likewise.
9492 (update_profile): Likewise.
9493 (estimated_freqs_path): Delete.
9494 (freqs_to_counts_path): Delete.
9495 (clear_counts_path): Delete.
9496 (ssa_fix_duplicate_block_edges): Likewise.
9497 (duplicate_thread_path): Likewise.
9498 * tree-switch-conversion.c (gen_inbound_check): Use counts.
9499 * tree-tailcall.c (decrease_profile): Do not update frequency.
9500 (eliminate_tail_call): Likewise.
9501 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
9502 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
9503 (optimize_mask_stores): Likewise.
9504 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9505 * ubsan.c (ubsan_expand_null_ifn): Update profile.
9506 (ubsan_expand_ptr_ifn): Update profile.
9507 * value-prof.c (gimple_ic): Simplify.
9508 * value-prof.h (gimple_ic): Update prototype.
9509 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
9510 * ipa-inline.c (compute_uninlined_call_time): Be sure that
9512 (want_inline_self_recursive_call_p): Likewise.
9513 (resolve_noninline_speculation): Only cummulate defined counts.
9514 (inline_small_functions): Use nonzero_p.
9515 (ipa_inline): Do not access freed node.
9517 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9519 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9520 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
9522 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
9524 * config/riscv/riscv.c (riscv_legitimize_move): Handle
9525 non-legitimate address.
9527 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9529 * config/rs6000/rs6000.md (*lt0_disi): Delete.
9530 (*lt0_<mode>di, *lt0_<mode>si): New.
9532 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9534 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
9535 TARGET_PAIRED_FLOAT.
9537 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
9538 Jim Wilson <jim.wilson@linaro.org>
9540 * config/aarch64/aarch64-cores.def (saphira): New CPU.
9541 * config/aarch64/aarch64-tune.md: Regenerated.
9542 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
9543 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
9545 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
9547 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
9548 (arc_expand_prologue): Restore blink for millicode.
9549 * config/arc/linux.h (LINK_EH_SPEC): Defined.
9551 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
9554 * config/i386/i386.c (ix86_vector_duplicate_value): Use
9555 gen_vec_duplicate after forcing the scalar into a register.
9557 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
9559 * combine (try_combine): Print the insns input to try_combine to the
9562 2017-11-02 Steve Ellcey <sellcey@cavium.com>
9565 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
9566 Remove second argument from aarch64_process_target_attr call.
9567 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
9569 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
9571 (aarch64_handle_attr_arch): Remove second argument.
9572 (aarch64_handle_attr_cpu): Ditto.
9573 (aarch64_handle_attr_tune): Ditto.
9574 (aarch64_handle_attr_isa_flags): Ditto.
9575 (aarch64_process_one_target_attr): Ditto.
9576 (aarch64_process_target_attr): Ditto.
9577 (aarch64_option_valid_attribute_p): Remove second argument.
9578 on aarch64_process_target_attr call.
9580 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9582 * diagnostic.c: Include "selftest-diagnostic.h".
9583 (selftest::assert_location_text): New function.
9584 (selftest::test_diagnostic_get_location_text): New function.
9585 (selftest::diagnostic_c_tests): Call it.
9587 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9589 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
9590 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
9591 (class selftest::test_diagnostic_context): Move to...
9592 * selftest-diagnostic.c: New file.
9593 * selftest-diagnostic.h: New file.
9595 2017-11-02 James Bowman <james.bowman@ftdichip.com>
9597 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
9598 offset range for FT32B.
9599 * config/ft32/ft32.h: option "mcompress" enables relaxation.
9600 * config/ft32/ft32.md: Add TARGET_NOPM.
9601 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
9602 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
9604 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
9606 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
9608 2017-11-02 Jeff Law <law@redhat.com>
9610 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
9611 virtual keyword on FINAL OVERRIDE members.
9613 * tree-ssa-propagate.h (ssa_propagation_engine): Group
9614 virtuals together. Add virtual destructor.
9615 (substitute_and_fold_engine): Similarly.
9617 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
9619 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
9621 2017-11-02 Richard Biener <rguenther@suse.de>
9623 PR tree-optimization/82795
9624 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
9626 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9628 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
9630 (gcc_GAS_CHECK_FEATURE): Remove.
9631 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
9632 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
9633 * configure: Regenerate.
9635 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
9637 * config/arc/arc.c (hwloop_optimize): Account for empty
9640 2017-11-02 Richard Biener <rguenther@suse.de>
9643 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
9644 Truncate ARRAY_REF index and element size.
9646 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9648 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
9650 2017-11-01 Jeff Law <law@redhat.com>
9652 * tree-ssa-ccp.c (ccp_folder): New class derived from
9653 substitute_and_fold_engine.
9654 (ccp_folder::get_value): New member function.
9655 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
9656 (ccp_fold_stmt): Remove prototype.
9657 (ccp_finalize): Call substitute_and_fold from the ccp_class.
9658 * tree-ssa-copy.c (copy_folder): New class derived from
9659 substitute_and_fold_engine.
9660 (copy_folder::get_value): Renamed from get_value.
9661 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
9662 * tree-vrp.c (vrp_folder): New class derived from
9663 substitute_and_fold_engine.
9664 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
9665 (vrp_folder::get_value): New member function.
9666 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
9667 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
9668 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
9669 provide a class interface to folder/substitute routines.
9670 (ssa_prop_fold_stmt_fn): Remove typedef.
9671 (ssa_prop_get_value_fn): Likewise.
9672 (subsitute_and_fold): Remove prototype.
9673 (replace_uses_in): Likewise.
9674 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
9675 Renamed from replace_uses_in. Call the virtual member function
9676 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
9677 (substitute_and_fold_dom_walker): Remove initialization of
9678 data member entries for calbacks. Add substitute_and_fold_engine
9679 member and initialize it.
9680 (substitute_and_fold_dom_walker::before_dom_children0: Use the
9681 member functions for get_value, replace_phi_args_in c
9682 replace_uses_in, and fold_stmt calls.
9683 (substitute_and_fold_engine::substitute_and_fold): Renamed from
9684 substitute_and_fold. Remove assert. Update ctor call.
9686 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
9687 (ssa_prop_visit_phi_fn): Likewise.
9688 (class ssa_propagation_engine): New class to provide an interface
9690 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
9692 (ssa_prop_visit_phi): Likewise.
9693 (ssa_propagation_engine::simulate_stmt): Moved into class.
9694 Call visit_phi/visit_stmt from the class rather than via
9695 file scoped static variables.
9696 (ssa_propagation_engine::simulate_block): Moved into class.
9697 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
9698 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
9699 set file scoped statics for the visit_stmt/visit_phi callbacks.
9700 * tree-complex.c (complex_propagate): New class derived from
9701 ssa_propagation_engine.
9702 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
9703 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
9704 (tree_lower_complex): Call ssa_propagate via the complex_propagate
9706 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
9707 ssa_propagation_engine.
9708 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
9709 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
9710 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
9711 * tree-ssa-copy.c (copy_prop): New class derived from
9712 ssa_propagation_engine.
9713 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
9714 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
9715 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
9716 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
9717 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
9718 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
9719 (execute_vrp): Call ssa_propagate from the vrp_prop class.
9721 2017-11-01 Jakub Jelinek <jakub@redhat.com>
9723 PR rtl-optimization/82778
9724 PR rtl-optimization/82597
9725 * compare-elim.c (struct comparison): Add in_a_setter field.
9726 (find_comparison_dom_walker::before_dom_children): Remove killed
9727 bitmap and df_simulate_find_defs call, instead walk the defs.
9728 Compute last_setter and initialize in_a_setter. Merge definitions
9729 with first initialization for a few variables.
9730 (try_validate_parallel): Use insn_invalid_p instead of
9731 recog_memoized. Return insn rather than just the pattern.
9732 (try_merge_compare): Fix up comment. Don't uselessly test if
9733 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
9735 (execute_compare_elim_after_reload): Remove df_chain_add_problem
9738 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9739 Alan Hayward <alan.hayward@arm.com>
9740 David Sherwood <david.sherwood@arm.com>
9742 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
9743 aarch64_hard_regno_nregs to get the number of registers
9746 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9747 Alan Hayward <alan.hayward@arm.com>
9748 David Sherwood <david.sherwood@arm.com>
9750 * config/aarch64/constraints.md (Upl): Rename to...
9752 * config/aarch64/aarch64.md
9753 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
9756 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9757 Alan Hayward <alan.hayward@arm.com>
9758 David Sherwood <david.sherwood@arm.com>
9760 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
9761 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
9764 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9765 Alan Hayward <alan.hayward@arm.com>
9766 David Sherwood <david.sherwood@arm.com>
9768 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
9769 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
9770 (aarch64_evpc_dup): Generate rtl direcly, rather than using
9772 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
9773 of a single element.
9774 * config/aarch64/iterators.md: Add a comment above the permute
9775 unspecs to say that they are generated directly by
9776 aarch64_expand_vec_perm_const.
9777 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
9779 2017-11-01 Nathan Sidwell <nathan@acm.org>
9781 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
9783 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9785 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
9786 and medany code models, and describe what they do.
9788 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9790 Revert accidental duplicate:
9792 * combine.c (can_change_dest_mode): Reject changes in
9793 REGMODE_NATURAL_SIZE.
9795 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9797 PR rtl-optimization/64682
9798 PR rtl-optimization/69567
9799 PR rtl-optimization/69737
9800 PR rtl-optimization/82683
9801 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9802 register mentioned in the note, drop the note, unless it came from I3,
9803 in which case it should go to I3 again.
9805 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9807 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9808 and return false if not.
9809 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9811 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9813 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9814 range is known to be empty.
9816 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9817 Alan Hayward <alan.hayward@arm.com>
9818 David Sherwood <david.sherwood@arm.com>
9820 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9821 and CONST_VECTOR_NUNITS instead of computing the number of units from
9822 the byte sizes of the vector and element.
9823 (simplify_binary_operation_1): Likewise.
9824 (simplify_const_binary_operation): Likewise.
9825 (simplify_ternary_operation): Likewise.
9827 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9828 Alan Hayward <alan.hayward@arm.com>
9829 David Sherwood <david.sherwood@arm.com>
9831 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9832 (int_mem_offset): ...this new function.
9833 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9834 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9835 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
9838 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9839 Alan Hayward <alan.hayward@arm.com>
9840 David Sherwood <david.sherwood@arm.com>
9842 * lower-subreg.c (interesting_mode_p): New function.
9843 (compute_costs, find_decomposable_subregs, decompose_register)
9844 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9845 (resolve_clobber, dump_choices): Use it.
9847 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9848 Alan Hayward <alan.hayward@arm.com>
9849 David Sherwood <david.sherwood@arm.com>
9851 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
9853 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9854 Alan Hayward <alan.hayward@arm.com>
9855 David Sherwood <david.sherwood@arm.com>
9857 * alias.c (find_base_value, find_base_term): Only process integer
9858 truncations. Check the precision rather than the size.
9860 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9861 Alan Hayward <alan.hayward@arm.com>
9862 David Sherwood <david.sherwood@arm.com>
9864 * machmode.h (is_narrower_int_mode): New function
9865 * optabs.c (expand_float, expand_fix): Use it.
9866 * dwarf2out.c (rotate_loc_descriptor): Likewise.
9868 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9869 Alan Hayward <alan.hayward@arm.com>
9870 David Sherwood <david.sherwood@arm.com>
9872 * rtl.h (narrower_subreg_mode): New function.
9873 * ira-color.c (update_costs_from_allocno): Use it.
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 * optabs-query.h (convert_optab_p): New function, split out from...
9880 (convert_optab_handler): ...here.
9881 (widening_optab_handler): Delete.
9882 (find_widening_optab_handler): Remove permit_non_widening parameter.
9883 (find_widening_optab_handler_and_mode): Likewise. Provide an
9884 override that operates on mode class wrappers.
9885 * optabs-query.c (widening_optab_handler): Delete.
9886 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9887 parameter. Assert that the two modes are the same class and that
9888 the "from" mode is narrower than the "to" mode. Use
9889 convert_optab_handler instead of widening_optab_handler.
9890 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9891 instead of widening_optab_handler.
9892 * expr.c (expand_expr_real_2): Update calls to
9893 find_widening_optab_handler.
9894 * optabs.c (expand_widen_pattern_expr): Likewise.
9895 (expand_binop_directly): Take the insn_code as a parameter.
9896 (expand_binop): Only call find_widening_optab_handler for
9897 conversion optabs; use optab_handler otherwise. Update calls
9898 to find_widening_optab_handler and expand_binop_directly.
9899 Use convert_optab_handler instead of widening_optab_handler.
9900 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9901 find_widening_optab_handler and use scalar_mode rather than
9903 (convert_plusminus_to_widen): Likewise.
9905 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9906 Alan Hayward <alan.hayward@arm.com>
9907 David Sherwood <david.sherwood@arm.com>
9909 * machmode.h (fixed_size_mode): New class.
9910 * rtl.h (get_pool_mode): Return fixed_size_mode.
9911 * gengtype.c (main): Add fixed_size_mode.
9912 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9913 (get_raw_arg_mode): Likewise.
9914 * doc/tm.texi: Regenerate.
9915 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9916 * targhooks.c (default_get_reg_raw_mode): Likewise.
9917 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9918 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9919 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9920 (msp430_get_raw_result_mode): Likewise.
9921 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9922 * dbxout.c (dbxout_parms): Require fixed-size modes.
9923 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9924 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9925 * omp-low.c (lower_oacc_reductions): Likewise.
9926 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9927 (simplify_subreg): Update accordingly.
9928 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9929 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9930 that aren't fixed-size.
9931 (get_pool_mode): Return a fixed_size_mode.
9932 (output_constant_pool_2): Take a fixed_size_mode.
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 * doc/rtl.texi (vec_series): Document.
9939 (const): Say that the operand can be a vec_series.
9940 * rtl.def (VEC_SERIES): New rtx code.
9941 * rtl.h (const_vec_series_p_1): Declare.
9942 (const_vec_series_p): New function.
9943 * emit-rtl.h (gen_const_vec_series): Declare.
9944 (gen_vec_series): Likewise.
9945 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9946 (gen_vec_series): Likewise.
9947 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9948 * simplify-rtx.c (simplify_unary_operation): Handle negations
9950 (simplify_binary_operation_series): New function.
9951 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9952 (test_vector_ops_series): New function.
9953 (test_vector_ops): Call it.
9954 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9955 gen_const_vec_series.
9956 (altivec_lvsr): Likewise.
9957 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9959 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9960 Alan Hayward <alan.hayward@arm.com>
9961 David Sherwood <david.sherwood@arm.com>
9963 * doc/rtl.texi (const): Update description of address constants.
9964 Say that vector constants are allowed too.
9965 * common.md (E, F): Use CONSTANT_P instead of checking for
9967 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9968 checking for CONST_VECTOR.
9969 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9971 * expr.c (expand_expr_real_2): Check for vector modes instead
9972 of checking for CONST_VECTOR.
9973 * rtl.h (const_vec_p): New function.
9974 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9975 (unwrap_const_vec_duplicate): Handle them here too.
9977 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9978 David Malcolm <dmalcolm@redhat.com>
9979 Alan Hayward <alan.hayward@arm.com>
9980 David Sherwood <david.sherwood@arm.com>
9982 * rtl.h (vec_duplicate_p): New function.
9983 * selftest-rtl.c (assert_rtx_eq_at): New function.
9984 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9985 (assert_rtx_eq_at): Declare.
9986 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9987 * selftest-run-tests.c (selftest::run_tests): Call it.
9988 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9989 (simplify_unary_operation_1): Recursively handle vector duplicates.
9990 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9992 (simplify_subreg): Handle subregs of vector duplicates.
9993 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9994 (selftest::simplify_rtx_c_tests): New functions.
9996 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9997 Alan Hayward <alan.hayward@arm.com>
9998 David Sherwood <david.sherwood@arm.com>
10000 * emit-rtl.h (gen_const_vec_duplicate): Declare.
10001 (gen_vec_duplicate): Likewise.
10002 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
10004 (gen_const_vector): ...here.
10005 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
10006 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
10007 whose elements are all equal.
10008 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
10009 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
10010 (simplify_relational_operation): Likewise.
10011 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
10013 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
10014 (aarch64_expand_vector_init): Likewise.
10015 * config/arm/arm.c (neon_vdup_constant): Likewise.
10016 (neon_expand_vector_init): Likewise.
10017 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
10018 (arm_block_set_unaligned_vect): Likewise.
10019 (arm_block_set_aligned_vect): Likewise.
10020 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
10021 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
10022 (expand_vec_perm_even_odd_pack): Likewise.
10023 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
10024 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
10025 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
10026 gen_const_vec_duplicate.
10027 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
10028 * config/mips/mips.c (mips_gen_const_int_vector): Use
10029 gen_const_vec_duplicate.
10030 (mips_expand_vector_init): Use CONST0_RTX.
10031 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
10032 (define_split): Use gen_const_vec_duplicate.
10033 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
10034 (define_split): Use gen_const_vec_duplicate.
10035 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
10036 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
10037 * config/spu/spu.c (spu_const): Likewise.
10039 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
10040 Alan Hayward <alan.hayward@arm.com>
10041 David Sherwood <david.sherwood@arm.com>
10043 * combine.c (can_change_dest_mode): Reject changes in
10044 REGMODE_NATURAL_SIZE.
10046 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
10048 * configure.ac (--enable-libssp): New.
10049 (gcc_cv_libc_provides_ssp): Check for explicit setting before
10050 trying to determine target-specific default. Adjust indentation.
10051 * configure: Regenerated.
10052 * doc/install.texi (Configuration): Expand --disable-libssp
10055 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
10057 config/i386/i386.c (ix86_expand_epilogue): Correct stack
10060 2017-10-31 Martin Jambor <mjambor@suse.cz>
10063 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
10065 2017-10-31 David Malcolm <dmalcolm@redhat.com>
10067 * auto-profile.c (autofdo_source_profile::read): Use
10068 UNKNOWN_LOCATION rather than 0.
10069 * diagnostic-core.h (warning_at_rich_loc): Rename to...
10070 (warning_at): ...this overload.
10071 (warning_at_rich_loc_n): Rename to...
10072 (warning_n): ...this overload.
10073 (error_at_rich_loc): Rename to...
10074 (error_at): ...this overload.
10075 (pedwarn_at_rich_loc): Rename to...
10076 (pedwarn): ...this overload.
10077 (permerror_at_rich_loc): Rename to...
10078 (permerror): ...this overload.
10079 (inform_at_rich_loc): Rename to...
10080 (inform): ...this overload.
10081 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
10082 (diagnostic_n_impl_richloc): Rename to...
10083 (diagnostic_n_impl): ...this rich_location *-based decl.
10084 (inform_at_rich_loc): Rename to...
10085 (inform): ...this, and add an assertion.
10086 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
10087 (warning_at_rich_loc): Rename to...
10088 (warning_at): ...this, and add an assertion.
10089 (warning_at_rich_loc_n): Rename to...
10090 (warning_n): ...this, and add an assertion.
10091 (warning_n): Update location_t-based implementation for removal of
10092 location_t-based diagnostic_n_impl.
10093 (pedwarn_at_rich_loc): Rename to...
10094 (pedwarn): ...this, and add an assertion.
10095 (permerror_at_rich_loc): Rename to...
10096 (permerror): ...this, and add an assertion.
10097 (error_n): Update for removal of location_t-based diagnostic_n_impl.
10098 (error_at_rich_loc): Rename to...
10099 (error_at): ...this, and add an assertion.
10100 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
10101 (driver::do_spec_on_infiles): Likewise.
10102 * substring-locations.c (format_warning_va): Update for renaming
10103 of inform_at_rich_loc.
10105 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
10107 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
10108 _Float<N>X built-in functions so that the variant without the
10109 "__builtin_" prefix is only enabled for the GNU C and Objective C
10110 languages when they are in non-strict ANSI/ISO mode.
10111 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
10112 * target.def (floatn_builtin_p): Add a target hook to control
10113 whether _Float<N> and _Float<N>X built-in functions without the
10114 "__builtin_" prefix are enabled, and return true for C and
10115 Objective C in the default hook. Include langhooks.h in
10117 * targhooks.h (default_floatn_builtin_p): Likewise.
10118 * targhooks.c (default_floatn_builtin_p): Likewise.
10119 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
10120 floatn_builtin_p target hook.
10121 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
10123 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
10124 Eric Botcazou <ebotcazou@adacore.com>
10126 PR rtl-optimization/81803
10127 * lra-constraints.c (curr_insn_transform): Also reload the whole
10128 register for a strict subreg no wider than a word if this is for
10129 a WORD_REGISTER_OPERATIONS target.
10131 2017-10-31 Jason Merrill <jason@redhat.com>
10133 * gdbinit.in: Skip over inlines from timevar.h.
10135 2017-10-31 Martin Liska <mliska@suse.cz>
10137 * doc/gcov.texi: Document new option.
10138 * gcov.c (print_usage): Likewise print it.
10139 (process_args): Support the argument.
10140 (format_count): New function.
10141 (format_gcov): Use the function.
10143 2017-10-31 Martin Liska <mliska@suse.cz>
10145 * gcov.c (struct name_map): do not use typedef.
10146 Define operator== and operator<.
10147 (name_search): Remove.
10148 (name_sort): Remove.
10149 (main): Do not allocate names.
10150 (process_file): Add vertical space.
10151 (generate_results): Use std::find.
10152 (release_structures): Do not release memory.
10153 (find_source): Use std::find.
10155 2017-10-31 Martin Liska <mliska@suse.cz>
10157 * gcov.c (struct line_info): Remove it's typedef.
10158 (line_info::line_info): Add proper ctor.
10159 (line_info::has_block): Do not use a typedef.
10160 (struct source_info): Do not use typedef.
10161 (circuit): Likewise.
10162 (get_cycles_count): Likewise.
10163 (output_intermediate_file): Iterate via vector iterator.
10164 (add_line_counts): Use std::vector methods.
10165 (accumulate_line_counts): Likewise.
10166 (output_lines): Likewise.
10168 2017-10-31 Martin Liska <mliska@suse.cz>
10170 * gcov.c (struct source_info): Remove typedef.
10171 (source_info::source_info): Add proper ctor.
10172 (accumulate_line_counts): Use struct, not it's typedef.
10173 (output_gcov_file): Likewise.
10174 (output_lines): Likewise.
10175 (main): Do not allocate an array.
10176 (output_intermediate_file): Use size of vector container.
10177 (process_file): Resize the vector.
10178 (generate_results): Do not preallocate, use newly added vector
10180 (release_structures): Do not release sources.
10181 (find_source): Use vector methods.
10182 (add_line_counts): Do not use typedef.
10184 2017-10-31 Martin Liska <mliska@suse.cz>
10186 * doc/gcov.texi: Document that.
10187 * gcov.c (add_line_counts): Mark lines with a non-executed
10189 (output_line_beginning): Handle such lines.
10190 (output_lines): Pass new argument.
10191 (output_intermediate_file): Print it in intermediate format.
10193 2017-10-31 Martin Liska <mliska@suse.cz>
10195 * color-macros.h: New file.
10196 * diagnostic-color.c: Factor out color related to macros to
10198 * doc/gcov.texi: Document -k option.
10199 * gcov.c (INCLUDE_STRING): Include string.h.
10200 (print_usage): Add -k option.
10201 (process_args): Parse it.
10202 (pad_count_string): New function.
10203 (output_line_beginning): Likewise.
10204 (DEFAULT_LINE_START): New macro.
10205 (output_lines): Support color output.
10207 2017-10-31 Martin Liska <mliska@suse.cz>
10209 PR gcov-profile/82633
10210 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
10211 their interaction with GCOV infrastructure.
10212 * configure.ac: Add -fkeep-{inline,static}-functions to
10214 * configure: Regenerate.
10216 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
10219 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
10221 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
10224 * config/rs6000/rs6000.md (allocate_stack): Force update interval
10225 into a register if it does not fit into an immediate offset field.
10227 2017-10-31 Olivier Hainque <hainque@adacore.com>
10229 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
10231 2017-10-31 Julia Koval <julia.koval@intel.com>
10233 * config.gcc: Add gfniintrin.h.
10234 * config/i386/gfniintrin.h: New.
10235 * config/i386/i386-builtin-types.def
10236 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
10237 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
10238 __builtin_ia32_vgf2p8affineinvqb_v32qi,
10239 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
10240 __builtin_ia32_vgf2p8affineinvqb_v16qi,
10241 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
10242 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
10243 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
10244 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
10245 V64QI_FTYPE_V64QI_V64QI_INT): New types.
10246 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
10247 * config/i386/immintrin.h: Include gfniintrin.h.
10248 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
10250 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
10252 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
10254 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10256 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
10257 (arm_ashldi3_1bit): Remove pattern.
10258 (ashrdi3): Remove shift by 1 expansion.
10259 (arm_ashrdi3_1bit): Remove pattern.
10260 (lshrdi3): Remove shift by 1 expansion.
10261 (arm_lshrdi3_1bit): Remove pattern.
10262 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
10263 cost of ashldi3 by 1.
10264 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
10265 (<shift>di3_neon): Likewise.
10267 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10269 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
10270 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
10271 and (*aarch64_simd_mov<VQ:mode>).
10272 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
10273 pattern alternative.
10274 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
10275 attributes to match pattern alternative.
10277 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
10279 * config.gcc (powerpc*-*-*): Add emmintrin.h.
10280 * config/rs6000/emmintrin.h: New file.
10281 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
10283 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10285 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
10286 * (movdi_vfp_cortexa8): Remove pattern.
10288 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10290 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
10292 (Specific, alpha*-dec-osf5.1): Remove.
10293 (Specific, mips-sgi-irix5): Remove.
10294 (Specific, mips-sgi-irix6): Remove.
10296 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10298 PR middle-end/22141
10299 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
10300 arguments to clear_bit_region_be.
10302 2017-10-30 Jim Wilson <wilson@tuliptree.org>
10304 * gimplify.c: Include memmodel.h.
10306 2017-10-30 Martin Jambor <mjambor@suse.cz>
10308 * omp-grid.c (grid_attempt_target_gridification): Also insert a
10309 condition whether loop should be executed at all.
10311 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
10313 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
10314 gimple folding of vec_madd() intrinsics.
10315 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
10316 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
10317 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
10319 2017-10-30 Richard Biener <rguenther@suse.de>
10321 PR tree-optimization/82762
10323 2017-10-23 Richard Biener <rguenther@suse.de>
10325 PR tree-optimization/82129
10327 2017-08-01 Richard Biener <rguenther@suse.de>
10329 PR tree-optimization/81181
10330 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10331 (compute_antic): ... end of iteration here.
10333 2017-10-30 Joseph Myers <joseph@codesourcery.com>
10335 * doc/invoke.texi (C Dialect Options): Document -std=c17,
10336 -std=iso9899:2017 and -std=gnu17.
10337 * doc/standards.texi (C Language): Document C17 support.
10338 * doc/cpp.texi (Overview): Mention -std=c17.
10339 (Standard Predefined Macros): Document C11 and C17 values of
10340 __STDC_VERSION__. Do not refer to C99 support as incomplete.
10341 * doc/extend.texi (Inline): Do not list individual options for
10342 standards newer than C99.
10343 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
10345 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
10348 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
10350 * asan.c (asan_finish_file): Align asan globals array by shadow
10353 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10355 PR middle-end/22141
10356 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
10357 (struct store_immediate_info): Add bitregion_start and bitregion_end
10359 (store_immediate_info::store_immediate_info): Add brs and bre
10360 arguments and initialize bitregion_{start,end} from those.
10361 (struct merged_store_group): Add bitregion_start, bitregion_end,
10362 align_base and mask fields. Drop unnecessary struct keyword from
10363 struct store_immediate_info. Add do_merge method.
10364 (clear_bit_region_be): Use memset instead of loop storing zeros.
10365 (merged_store_group::do_merge): New method.
10366 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
10367 stores as long as the surrounding bitregions have no gaps.
10368 (merged_store_group::merge_overlapping): Use do_merge.
10369 (merged_store_group::apply_stores): Test that bitregion_{start,end}
10370 is byte aligned, rather than requiring that start and width are
10371 byte aligned. Drop unnecessary struct keyword from
10372 struct store_immediate_info. Allocate and populate also mask array.
10373 Make start of the arrays relative to bitregion_start rather than
10374 start and size them according to bitregion_{end,start} difference.
10375 (struct imm_store_chain_info): Drop unnecessary struct keyword from
10376 struct store_immediate_info.
10377 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
10378 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
10379 struct keyword from struct store_immediate_info.
10380 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
10381 between stores as long as the surrounding bitregions have no gaps.
10383 (struct split_store): Add orig non-static data member.
10384 (split_store::split_store): Initialize orig to false.
10385 (find_constituent_stmts): Return store_immediate_info *, non-NULL
10386 if there is exactly a single original stmt. Change stmts argument
10387 to pointer from reference, if NULL, don't push anything to it. Add
10388 first argument, use it to optimize skipping over orig stmts that
10389 are known to be before bitpos already. Simplify.
10390 (split_group): Return unsigned int count how many stores are or
10391 would be needed rather than a bool. Add allow_unaligned argument.
10392 Change split_stores argument from reference to pointer, if NULL,
10393 only do a dry run computing how many stores would be produced.
10394 Rewritten algorithm to use both alignment and misalign if
10395 !allow_unaligned and handle bitfield stores with gaps.
10396 (imm_store_chain_info::output_merged_store): Set start_byte_pos
10397 from bitregion_start instead of start. Compute allow_unaligned
10398 here, if true, do 2 split_group dry runs to compute which one
10399 produces fewer stores and prefer aligned if equal. Punt if
10400 new count is bigger or equal than original before emitting any
10401 statements, rather than during that. Remove no longer needed
10402 new_ssa_names tracking. Replace num_stmts with
10403 split_stores.length (). Use 32-bit stack allocated entries
10404 in split_stores auto_vec. Try to reuse original store lhs/rhs1
10405 if possible. Handle bitfields with gaps.
10406 (pass_store_merging::execute): Ignore bitsize == 0 stores.
10407 Compute bitregion_{start,end} for the stores and construct
10408 store_immediate_info with that. Formatting fixes.
10410 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
10413 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
10414 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
10416 2017-10-29 Jim Wilson <wilson@tuliptree.org>
10418 * gimplify.c: Include tm_p.h.
10420 * common.opt (gcoff): Re-add as ignored option.
10421 (gcoff1, gcoff2, gcoff3): Likewise.
10423 * Makefile.in (OBJS): Delete sdbout.o.
10424 (GTFILES): Delete $(srcdir)/sdbout.c.
10425 * debug.h: Delete sdb_debug_hooks.
10426 * final.c: Delete sdbout.h include.
10427 (final_scan_insn): Delete SDB_DEBUG check.
10428 (rest_of_clean_state): Likewise.
10429 * output.h: Delete sdb_begin_function_line.
10430 * sdbout.c: Delete.
10431 * sdbout.h: Delete.
10432 * toplev.c: Delete sdbout.h include.
10433 (process_options): Delete SDB_DEBUG check.
10434 * tree-core.h (tree_type_common): Delete pointer field of
10436 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
10437 TYPE_SYMTAB_POINTER.
10438 * tree.h (TYPE_SYMTAB_POINTER): Delete.
10439 (TYPE_SYMTAB_IS_POINTER): Delete.
10440 (TYPE_SYMTAB_IS_DIE): Renumber.
10441 * xcoffout.c: Refer to former sdbout.c file.
10442 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
10444 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
10445 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
10446 (Debugging Options): Delete -gcoff.
10447 (-gstabs): Delete SDB reference.
10449 (-gcoff@var{level}): Delete.
10450 * doc/passes.texi (Debugging information output): Delete SDB and
10451 sdbout.c references.
10452 * doc/tm.texi: Regenerate.
10453 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
10454 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
10456 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
10457 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
10458 (SDB_DEBUGGING_INFO): Delete.
10459 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
10460 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
10461 * target.def (output_source_filename): Delete COFF reference.
10463 * common.opt (gcoff): Delete.
10464 (gxcoff+): Update Negative chain.
10465 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
10467 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
10468 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
10469 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
10470 and SDB references.
10471 (expand_function_start): Change sdb reference to past tense.
10472 (expand_function_end): Change sdb reference to past tense.
10473 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
10474 * opts.c (debug_type_names): Delete coff entry.
10475 (common_handle_option): Delete OPT_gcoff case.
10476 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
10478 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
10479 * config/cris/cris.h: Delete SDB reference in comment.
10480 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
10481 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
10482 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
10483 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
10485 (ix86_expand_prologue): Likewise.
10486 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
10487 * config/ia64/ia64.h: Likewise.
10488 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
10489 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
10491 * config/mmix/mmix.h: Likewise.
10492 * config/nds32/nds32.c: Likewise.
10493 * config/stormy/storym16.h: Likewise.
10494 * config/visium/visium.h: Likewise.
10495 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
10497 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
10499 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
10500 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
10501 FRAME_POINTER_REGNUM point at high end of local var area.
10503 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
10505 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
10506 Move comment around. Do not reset best_edge for a copiable
10507 destination if the copy would cause a partition change.
10508 (better_edge_p): Remove redundant check.
10510 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10512 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
10514 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10516 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
10517 for math functions that have _Float<N> and _Float<N>X variants.
10518 (mathfn_built_in_2): Add support for math functions that have
10519 _Float<N> and _Float<N>X variants.
10520 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
10521 (expand_builtin_mathfn_ternary): Add support for fma with
10522 _Float<N> and _Float<N>X variants.
10523 (expand_builtin): Likewise.
10524 (fold_builtin_3): Likewise.
10525 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
10526 create math function _Float<N> and _Float<N>X variants as external
10528 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
10529 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
10530 the __builtin_ prefix and if not strict ansi, without the prefix.
10531 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
10532 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
10533 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
10534 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
10535 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
10536 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
10537 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
10538 function signatures for fma _Float<N> and _Float<N>X variants.
10539 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
10540 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
10541 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
10542 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
10543 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
10544 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
10545 * gencfn-macros.c (print_case_cfn): Add support for math functions
10546 that have _Float<N> and _Float<N>X variants.
10547 (print_define_operator_list): Likewise.
10548 (fltfn_suffixes): Likewise.
10550 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
10551 for math functions that have _Float<N> and _Float<N>X variants.
10552 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
10553 and _Float<N>X variants.
10554 (COPYSIGN): Likewise.
10557 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
10558 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
10560 (integer_valued_read_call_p): Likewise.
10561 * fold-const-call.c (fold_const_call_ss): Likewise.
10562 (fold_const_call_sss): Add support for copysign, fmin, and fmax
10563 _Float<N> and _Float<N>X variants.
10564 (fold_const_call_ssss): Add support for fma _Float<N> and
10565 _Float<N>X variants.
10566 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
10567 support for copysign and fma _Float<N> and _Float<N>X variants.
10568 (backprop::process_builtin_call_use): Likewise.
10569 * tree-call-cdce.c (can_test_argument_range); Add support for
10570 sqrt _Float<N> and _Float<N>X variants.
10571 (edom_only_function): Likewise.
10572 (get_no_error_domain): Likewise.
10573 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
10574 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
10575 copysign _Float<N> and _Float<N>X variants.
10576 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
10577 handled by machine independent code.
10578 (FMAF128): Likewise.
10579 * doc/cpp.texi (Common Predefined Macros): Document defining
10580 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
10581 fma _Float<N> and _Float<N>X variants.
10583 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10586 * config/i386/i386-modes.def (CCFPU): Remove definition.
10587 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
10588 (ix86_cc_modes_compatible): Ditto.
10589 (ix86_expand_carry_flag_compare): Ditto.
10590 (ix86_expand_int_movcc): Ditto.
10591 (ix86_expand_int_addcc): Ditto.
10592 (ix86_reverse_condition): Ditto.
10593 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
10594 Return true/false for unordered/ordered fp comparisons.
10595 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
10596 (ix86_prepare_fp_compare_args): Update for rename.
10597 (ix86_expand_fp_compare): Update for rename. Generate unordered
10598 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
10599 (ix86_expand_sse_compare_and_jump): Ditto.
10600 * config/i386/predicates.md (fcmov_comparison_operator):
10601 Remove CCFPU mode handling.
10602 (ix86_comparison_operator): Ditto.
10603 (ix86_carry_flag_operator): Ditto.
10604 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
10605 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
10606 (*cmpu<mode>_cc_i387): Ditto.
10607 (FPCMP): Remove mode iterator.
10608 (unord): Remove mode attribute.
10609 (unord_subst): New define_subst transformation
10610 (unord): New define_subst attribute.
10611 (unordered): Ditto.
10612 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
10613 (*cmpi<unord>xf_i387): Ditto.
10614 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
10615 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
10616 using unord_subst transformation.
10617 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
10618 (round_saeonly): Also handle CCFP mode.
10619 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
10620 Remove UNSPEC_SAHF unspec handling.
10622 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10624 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
10626 2017-10-27 Jeff Law <law@redhat.com>
10628 * gimple-ssa-sprintf.c: Include domwalk.h.
10629 (class sprintf_dom_walker): New class, derived from dom_walker.
10630 (sprintf_dom_walker::before_dom_children): New function.
10631 (struct call_info): Moved into sprintf_dom_walker class
10632 (compute_formath_length, handle_gimple_call): Likewise.
10633 (sprintf_length::execute): Call the dominator walker rather
10634 than walking the statements.
10636 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
10637 gimple statement locations.
10638 (check_array_bounds): Corresponding changes. Get the statement's
10639 location directly from wi->stmt.
10641 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
10644 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
10646 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10648 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
10649 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
10651 2017-10-27 Jakub Jelinek <jakub@redhat.com>
10654 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
10655 * config/i386/i386.c (maybe_get_pool_constant): Removed.
10656 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
10657 maybe_get_pool_constant.
10658 * config/i386/predicates.md (zero_extended_scalar_load_operand):
10661 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10663 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
10664 2.26 caveat. Update gas and gld versions.
10665 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
10668 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10670 * cgraph.h (set_malloc_flag): Declare.
10671 * cgraph.c (set_malloc_flag_1): New function.
10672 (set_malloc_flag): Likewise.
10673 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
10674 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
10676 (read_ipa_call_summary): Add support for reading is_return_callee.
10677 (write_ipa_call_summary): Stream is_return_callee.
10678 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
10679 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
10680 ipa-prop.h, ipa-fnsummary.h.
10681 (pure_const_names): Change to static.
10682 (malloc_state_e): Define.
10683 (malloc_state_names): Define.
10684 (funct_state_d): Add field malloc_state.
10685 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
10686 (check_retval_uses): New function.
10687 (malloc_candidate_p): Likewise.
10688 (analyze_function): Add support for malloc attribute.
10689 (pure_const_write_summary): Stream malloc_state.
10690 (pure_const_read_summary): Add support for reading malloc_state.
10691 (dump_malloc_lattice): New function.
10692 (propagate_malloc): New function.
10693 (warn_function_malloc): New function.
10694 (ipa_pure_const::execute): Call propagate_malloc and
10695 ipa_free_fn_summary.
10696 (pass_local_pure_const::execute): Add support for malloc attribute.
10697 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
10698 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
10700 2017-10-27 Martin Liska <mliska@suse.cz>
10702 PR gcov-profile/82457
10703 * doc/invoke.texi: Document that one needs a non-strict ISO mode
10704 for fork-like functions to be properly instrumented.
10706 2017-10-27 Richard Biener <rguenther@suse.de>
10708 PR middle-end/81659
10709 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
10710 info when we redirected EH.
10712 2017-10-26 Michael Collison <michael.collison@arm.com>
10714 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
10716 (<optab>_trunchf<GPI:mode>2: New pattern.
10717 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
10718 * config/aarch64/iterators.md (wv): New mode attribute.
10719 (vf, VF): New mode attributes.
10720 (vgp, VGP): New mode attributes.
10721 (s): Update attribute with SImode and DImode prefixes.
10723 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10725 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
10726 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
10727 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
10728 (nios2_option_overide): Initialize it. Don't allow R0-relative
10729 addressing with PIC.
10730 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
10731 (nios2_symbolic_constant_p): Likewise.
10732 (nios2_legitimate_address_p): Likewise.
10733 (nios2_r0rel_section_name_p): New.
10734 (nios2_symbol_ref_in_r0rel_data_p): New.
10735 (nios2_emit_move_sequence): Handle r0rel_constant_p.
10736 (r0rel_constant_p): New.
10737 (nios2_print_operand_address): Handle r0rel_constant_p.
10738 (nios2_cdx_narrow_form_p): Likewise.
10739 * config/nios2/nios2.opt (mr0rel-sec=): New option.
10740 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
10741 (Nios II Options): Document -mr0rel-sec.
10743 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10745 * config/nios2/nios2.c: Include xregex.h.
10746 (nios2_gprel_sec_regex): New.
10747 (nios2_option_overide): Initialize it. Don't allow GP-relative
10748 addressing with PIC.
10749 (nios2_small_section_name_p): Check for regex match.
10750 * config/nios2/nios2.opt (mgprel-sec=): New option.
10751 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
10752 (Nios II Options): Document -mgprel-sec.
10754 2017-10-26 Jim Wilson <wilson@tuliptree.org>
10756 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
10758 2017-10-26 Tom de Vries <tom@codesourcery.com>
10760 PR tree-optimization/82707
10761 * gimple.c (gimple_copy): Fix unsharing of
10762 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
10764 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
10766 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
10767 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
10768 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
10769 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
10770 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
10771 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
10772 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
10773 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
10774 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
10775 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
10776 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
10777 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
10778 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
10779 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
10780 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
10781 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
10782 _mm512_mask_cmpunord_ps_mask): New intrinsics.
10784 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10786 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
10788 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
10789 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10790 warning to rs6000.c. Remove the Undocumented flag, since it has
10792 (-mabi=ibmlongdouble): Likewise.
10793 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10794 already set, set the default format for long double.
10795 (rs6000_debug_reg_global): Print whether long double is IBM or
10797 (rs6000_option_override_internal): Rework setting long double
10798 format. Only warn if the user is changing the long double default
10799 and they did not use -Wno-psabi.
10800 * doc/invoke.texi (PowerPC options): Update the documentation for
10801 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10803 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10804 Alan Hayward <alan.hayward@arm.com>
10805 David Sherwood <david.sherwood@arm.com>
10807 * rtl.h (wider_subreg_mode): New function.
10808 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10809 rather than an unsigned int *.
10810 * ira-color.c (regno_max_ref_width): Replace with...
10811 (regno_max_ref_mode): ...this new variable.
10812 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10813 Use wider_subreg_mode.
10814 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10815 rather than an unsigned int *.
10816 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10817 (process_alt_operands): Likewise.
10818 (invariant_p): Likewise.
10819 * lra-spills.c (assign_mem_slot): Likewise.
10820 (add_pseudo_to_slot): Likewise.
10821 * lra.c (collect_non_operand_hard_regs): Likewise.
10822 (add_regs_to_insn_regno_info): Likewise.
10823 * reload1.c (regno_max_ref_width): Replace with...
10824 (regno_max_ref_mode): ...this new variable.
10825 (reload): Update accordingly. Update call to
10826 ira_sort_regnos_for_alter_reg.
10827 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10828 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10829 (scan_paradoxical_subregs): Likewise.
10831 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10833 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10834 (aarch64_frame): Add emit_frame_chain boolean.
10835 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10836 Move eh_return case to aarch64_layout_frame.
10837 (aarch64_layout_frame): Initialize emit_frame_chain.
10838 (aarch64_expand_prologue): Use emit_frame_chain.
10840 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10842 * config/aarch64/aarch64.c (aarch64_layout_frame):
10843 Ensure LR is always stored at the bottom of the callee-saves.
10844 Remove rarely used frame layout which saves callee-saves at top of
10845 frame, so the store of LR can be used as a valid probe in all cases.
10847 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10849 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10850 Improve unaligned TImode/TFmode base/offset split.
10852 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10853 Alan Hayward <alan.hayward@arm.com>
10854 David Sherwood <david.sherwood@arm.com>
10856 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10857 * combine.c (find_single_use_1): Likewise.
10858 (expand_field_assignment): Likewise.
10859 (move_deaths): Likewise.
10860 * lra-constraints.c (simplify_operand_subreg): Likewise.
10861 (curr_insn_transform): Likewise.
10862 * lra.c (collect_non_operand_hard_regs): Likewise.
10863 (add_regs_to_insn_regno_info): Likewise.
10864 * rtlanal.c (reg_referenced_p): Likewise.
10865 (covers_regno_no_parallel_p): Likewise.
10867 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10869 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10870 Don't print any bits outside the precision of the value.
10871 * wide-int.cc (test_printing): Add some new tests.
10873 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10875 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10876 supports -xbrace_comment option.
10877 * configure: Regenerate.
10878 * config.in: Regenerate.
10879 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10880 (ASM_CPU_SPEC): Use it.
10882 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10884 * target.def (static_rtx_alignment): New hook.
10885 * targhooks.h (default_static_rtx_alignment): Declare.
10886 * targhooks.c (default_static_rtx_alignment): New function.
10887 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10888 * doc/tm.texi: Regenerate.
10889 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10890 instead of targetm.constant_alignment. Remove call to
10891 set_mem_attributes.
10892 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10893 (cris_preferred_mininum_alignment): New function, split out from...
10894 (cris_constant_alignment): ...here.
10895 (cris_static_rtx_alignment): New function.
10896 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10898 (ix86_constant_alignment): ...here.
10899 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10900 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10901 (mmix_static_rtx_alignment): New function.
10902 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10903 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10905 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10908 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10909 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10911 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10913 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10914 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10915 mark as a sign-extending load.
10916 (local_pic_load_u): Define.
10918 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10920 PR middle-end/82062
10921 * fold-const.c (operand_equal_for_comparison_p): Also return true
10922 if ARG0 is a simple variant of ARG1 with narrower precision.
10923 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10925 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10927 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10929 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10930 scatter_static, scatter_per_elt.
10931 * x86-tune-costs.h: Add new cost entries.
10933 2017-10-25 Richard Biener <rguenther@suse.de>
10935 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10936 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10937 class pass_fre): Move to ...
10938 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10939 class pass_fre): ... here and adjust for statistics.
10941 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10944 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10945 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10946 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10947 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10950 2017-10-25 Richard Biener <rguenther@suse.de>
10952 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10953 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10954 eliminate_push_avail, eliminate_insert): Move inside...
10955 (class eliminate_dom_walker): ... this class in preparation
10957 (fini_eliminate): Remove by merging with ...
10958 (eliminate): ... this function. Adjust for class changes.
10959 (pass_pre::execute): Remove fini_eliminate call.
10960 (pass_fre::execute): Likewise.
10962 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10965 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10966 (VPERMI2, VPERMI2I): New mode iterators.
10967 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10968 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10970 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10971 mode iterator. Remove 3 old define_insn patterns.
10972 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10973 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10974 VPERMI2 mode iterator, remove the other two expanders.
10975 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10976 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10977 instructions, remove the other two patterns.
10978 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10979 mode iterator, remove the other two patterns.
10980 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10981 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10982 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10983 and adjust argument order accordingly.
10984 (ix86_expand_vec_perm): Adjust caller.
10985 (expand_vec_perm_1): Likewise.
10986 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10987 (expand_vec_perm_vpermt2_vpshub2): ... this.
10988 (ix86_expand_vec_perm_const_1): Adjust caller.
10989 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10992 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10993 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10994 (vec_shl_<mode>): Remove unused expander.
10995 (avx512bw_<shift_insn><mode>3): New define_insn.
10996 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10997 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10999 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
11002 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
11005 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
11007 PR rtl-optimization/82396
11008 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
11009 (autopref_multipass_init): Simplify initialization.
11010 (autopref_rank_data): Simplify sort order.
11011 * gcc/sched-int.h (autopref_multipass_data_): Remove
11012 multi_mem_insn_p, min_offset and max_offset.
11014 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
11016 PR middle-end/60580
11017 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
11018 Check special value of flag_omit_frame_pointer.
11019 (aarch64_can_eliminate): Likewise.
11020 (aarch64_override_options_after_change_1): Simplify handling of
11021 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
11023 2017-10-24 Richard Biener <rguenther@suse.de>
11025 PR tree-optimization/82697
11026 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
11027 zero for conditional load and unconditional store.
11029 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
11031 * doc/install.texi: Document bootstrap-cet.
11033 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
11036 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
11037 ENDBR instruction at function entrance if function is only
11040 2017-10-24 Jakub Jelinek <jakub@redhat.com>
11043 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
11044 patterns to better describe from which operation the CF is computed.
11045 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
11046 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
11047 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
11048 is 0, use _0 suffixed expanders instead of emitting a comparison
11051 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
11053 * config/i386/i386.md(*movsf_internal, *movdf_internal):
11054 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
11056 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
11058 PR middle-end/82569
11059 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
11060 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
11061 * loop-iv.c (iv_get_reaching_def): Likewise.
11062 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
11063 variable is promoted and the partition contains undefined values.
11065 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11067 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
11069 (nios2_address_cost): Define.
11070 (nios2_legitimize_address): Recognize (exp + constant) directly.
11071 (TARGET_ADDRESS_COST): Define.
11073 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11075 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
11076 (nios2_symbolic_memory_operand_p): Declare.
11077 (nios2_split_large_constant): Declare.
11078 (nios2_split_symbolic_memory_operand): Declare.
11079 * config/nios2/nios2.c: Adjust includes.
11080 (nios2_symbolic_constant_allowed): New.
11081 (nios2_symbolic_constant_p): New.
11082 (nios2_plus_symbolic_constant_p): New.
11083 (nios2_valid_addr_expr_p): Recognize addresses involving
11084 symbolic constants.
11085 (nios2_legitimate_address_p): Likewise, also LO_SUM.
11086 (nios2_symbolic_memory_operand_p): New.
11087 (nios2_large_constant_p): New.
11088 (nios2_split_large_constant): New.
11089 (nios2_split_plus_large_constant): New.
11090 (nios2_split_symbolic_memory_operand): New.
11091 (nios2_legitimize_address): Code refactoring. Handle addresses
11092 involving symbolic constants.
11093 (nios2_emit_move_sequence): Likewise.
11094 (nios2_print_operand): Improve error output.
11095 (nios2_print_operand_address): Handle LO_SUM.
11096 (nios2_cdx_narrow_form_p): Likewise.
11097 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
11098 operands involving symbolic constants.
11099 (movhi_internal, movsi_internal): Likewise.
11100 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
11101 (extendhisi2, extendqi<mode>2): Likewise.
11103 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11105 * tree-pass.h (PROP_rtl_split_insns): Define.
11106 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
11108 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11110 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
11112 2017-10-23 Jakub Jelinek <jakub@redhat.com>
11115 * target.def (const_not_ok_for_debug_p): Default to
11116 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
11117 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
11118 * targhooks.c (default_const_not_ok_for_debug_p): New function.
11119 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
11120 which targetm.const_not_ok_for_debug_p returned true.
11121 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
11123 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
11125 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
11126 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
11127 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
11129 (ix86_const_not_ok_for_debug_p): New function.
11130 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
11131 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
11133 2017-10-23 David Malcolm <dmalcolm@redhat.com>
11136 * system.h: Conditionally include "unique-ptr.h" if
11137 INCLUDE_UNIQUE_PTR is defined.
11138 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
11139 of defining INCLUDE_UNIQUE_PTR before including "system.h".
11141 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
11143 * config/rl78/rl78.md: New define_expand "subdi3".
11145 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
11148 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
11149 DF_REF_INSN if DF_REF_INSN_INFO is false.
11151 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
11153 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
11154 xmm_move instead of sse_move.
11155 (sse_store_index): New function.
11156 (ix86_register_move_cost): Be more sensible about mismatch stall;
11157 model AVX moves correctly; make difference between sse->integer and
11159 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
11160 moves; make difference between SSE and AVX.
11161 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
11162 and zmm_move. Increase size of sse load and store tables;
11163 add unaligned load and store tables; add ssemmx_to_integer.
11164 * x86-tune-costs.h: Update all entries according to real
11165 move latencies from Agner Fog's manual and chip documentation.
11167 2017-10-23 Jakub Jelinek <jakub@redhat.com>
11170 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
11171 * config/i386/constraints.md (Wf): New constraint.
11172 * config/i386/i386.md (UNSPEC_SBB): New unspec.
11173 (cmp<dwi>_doubleword): Removed.
11174 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
11175 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
11176 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
11177 expand with cmp<dwi>_doubleword. For LTU and GEU use
11178 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
11180 * common.opt (gcolumn-info): Enable by default.
11181 * doc/invoke.texi (gcolumn-info): Document new default.
11183 2017-10-23 Richard Biener <rguenther@suse.de>
11185 PR tree-optimization/82672
11186 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
11187 Fold the stmt if we propagated into it.
11189 2017-10-23 Richard Biener <rguenther@suse.de>
11191 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
11192 (bitmap_remove_expr_from_set): ... this. All callers call this
11193 for non-constant values.
11194 (bitmap_set_subtract): Rename to...
11195 (bitmap_set_subtract_expressions): ... this. Adjust and
11197 (bitmap_set_contains_value): Remove superfluous check.
11198 (bitmap_set_replace_value): Inline into single caller ...
11199 (bitmap_value_replace_in_set): ... here and simplify.
11200 (dependent_clean): Merge into ...
11201 (clean): ... this using an overload. Adjust.
11202 (prune_clobbered_mems): Adjust.
11203 (compute_antic_aux): Likewise.
11204 (compute_partial_antic_aux): Likewise.
11206 2017-10-23 Richard Biener <rguenther@suse.de>
11208 PR tree-optimization/82129
11210 2017-08-01 Richard Biener <rguenther@suse.de>
11212 PR tree-optimization/81181
11213 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
11214 (compute_antic): ... end of iteration here.
11216 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11218 * target.def (starting_frame_offset): New hook.
11219 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
11220 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
11221 * doc/tm.texi.in: Regenerate.
11222 * hooks.h (hook_hwi_void_0): Declare.
11223 * hooks.c (hook_hwi_void_0): New function.
11224 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
11225 STARTING_FRAME_OFFSET.
11226 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
11227 * reload1.c (reload): Likewise.
11228 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
11229 instead of STARTING_FRAME_OFFSET.
11230 * function.c (try_fit_stack_local): Likewise.
11231 (assign_stack_local_1): Likewise
11232 (instantiate_virtual_regs): Likewise.
11233 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
11234 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
11235 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
11236 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
11237 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
11238 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
11239 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
11240 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
11241 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
11242 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
11243 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
11244 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
11245 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
11246 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
11247 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
11248 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
11249 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
11250 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
11251 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
11252 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
11253 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
11254 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
11255 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
11256 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
11257 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
11258 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
11259 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
11260 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
11261 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
11262 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
11263 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
11264 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
11265 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
11266 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
11267 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
11268 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
11269 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
11270 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
11271 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
11272 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
11273 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
11274 * config/avr/avr.c (avr_starting_frame_offset): Make static and
11275 return a HOST_WIDE_INT.
11276 (avr_builtin_setjmp_frame_value): Use it instead of
11277 STARTING_FRAME_OFFSET.
11278 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11279 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
11280 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
11282 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11283 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
11284 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
11285 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11286 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
11287 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
11288 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11289 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
11290 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
11291 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11292 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
11293 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
11295 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11296 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
11297 * config/mips/mips.c (mips_compute_frame_info): Refer to
11298 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11299 (mips_starting_frame_offset): New function.
11300 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11301 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
11302 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
11303 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
11304 and return a HOST_WIDE_INT.
11305 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11306 (mmix_initial_elimination_offset): Refer to
11307 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11308 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
11309 * config/pa/pa.c (pa_starting_frame_offset): New function.
11310 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
11311 (pa_expand_prologue): Likewise.
11312 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11313 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
11314 !FRAME_GROWS_DOWNWARD handling to...
11315 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11316 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
11317 !FRAME_GROWS_DOWNWARD handling to...
11318 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11319 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
11320 !FRAME_GROWS_DOWNWARD handling to...
11321 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11322 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
11324 (rs6000_starting_frame_offset): New function.
11325 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
11326 !FRAME_GROWS_DOWNWARD handling to...
11327 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11328 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
11329 !FRAME_GROWS_DOWNWARD handling to...
11330 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11331 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
11332 !FRAME_GROWS_DOWNWARD handling to...
11333 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11334 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
11335 (rs6000_starting_frame_offset): New function.
11336 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
11337 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
11338 * config/vax/vax.c (vax_starting_frame_offset): New function.
11339 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
11340 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11341 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
11342 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
11343 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11344 * system.h (STARTING_FRAME_OFFSET): Poison.
11346 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11348 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11349 SCALAR_TYPE_MODE instead of TYPE_MODE.
11351 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11352 Alan Hayward <alan.hayward@arm.com>
11353 David Sherwood <david.sherwood@arm.com>
11355 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
11357 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11358 Alan Hayward <alan.hayward@arm.com>
11359 David Sherwood <david.sherwood@arm.com>
11361 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
11363 2017-10-23 Richard Biener <rguenther@suse.de>
11365 PR tree-optimization/82129
11366 * tree-ssa-pre.c (bitmap_set_and): Remove.
11367 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
11368 canonicalizing expressions in the set to those with lowest
11369 ID rather than taking that from the first edge.
11371 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11373 * combine.c (rtx_equal_for_field_assignment_p): Use
11374 byte_lowpart_offset.
11376 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11377 Alan Hayward <alan.hayward@arm.com>
11378 David Sherwood <david.sherwood@arm.com>
11380 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
11381 to a SUBREG_PROMOTED_VAR.
11383 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11384 Alan Hayward <alan.hayward@arm.com>
11385 David Sherwood <david.sherwood@arm.com>
11387 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
11388 (expand_debug_source_expr): Likewise.
11389 * combine.c (combine_simplify_rtx): Likewise.
11390 * cse.c (fold_rtx): Likewise.
11391 * optabs.c (expand_float): Likewise.
11392 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11393 (simplify_binary_operation_1): Likewise.
11395 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11396 Alan Hayward <alan.hayward@arm.com>
11397 David Sherwood <david.sherwood@arm.com>
11399 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
11400 (record_promoted_value): Likewise.
11401 * expr.c (expand_expr_real_2): Likewise.
11402 * ree.c (update_reg_equal_equiv_notes): Likewise.
11403 (combine_set_extension): Likewise.
11404 * rtlanal.c (low_bitmask_len): Likewise.
11405 * simplify-rtx.c (neg_const_int): Likewise.
11406 (simplify_binary_operation_1): Likewise.
11408 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11409 Alan Hayward <alan.hayward@arm.com>
11410 David Sherwood <david.sherwood@arm.com>
11412 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
11413 * regcprop.c (maybe_mode_change): Likewise.
11414 * reload1.c (alter_reg): Likewise.
11416 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11418 * inchash.h (inchash::hash::add_wide_int): New function.
11419 * lto-streamer-out.c (hash_tree): Use it.
11421 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11423 * inchash.h (inchash::hash::add_wide_int): Rename to...
11424 (inchash::hash::add_hwi): ...this.
11425 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
11426 (polymorphic_call_target_hasher::hash): Likewise.
11427 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
11428 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
11429 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
11430 * lto-streamer-out.c (hash_tree): Likewise.
11431 * optc-save-gen.awk: Likewise.
11432 * tree.c (add_expr): Likewise.
11434 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11437 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
11438 for ordered inequality comparisons even with TARGET_IEEE_FP.
11440 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11443 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
11444 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11445 Expand with cmp<dwi>_doubleword.
11447 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11449 * extend.texi: Add x86 specific to 'nocf_check' attribute.
11450 List CET intrinsics.
11451 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
11452 specific to -fcf-protection option.
11454 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11456 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
11457 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
11458 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
11459 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
11460 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
11461 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
11462 (extra_objs): Add cet.o for Linux/x86 targets.
11463 (tmake_file): Add i386/t-cet for Linux/x86 targets.
11464 * config/i386/cet.c: New file.
11465 * config/i386/cetintrin.h: Likewise.
11466 * config/i386/t-cet: Likewise.
11467 * config/i386/cpuid.h (bit_SHSTK): New.
11468 (bit_IBT): Likewise.
11469 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
11470 pass IBT and SHSTK bits.
11471 * config/i386/i386-builtin-types.def
11472 (VOID_FTYPE_UNSIGNED_PVOID): New.
11473 (VOID_FTYPE_UINT64_PVOID): Likewise.
11474 * config/i386/i386-builtin.def: Add CET intrinsics.
11475 * config/i386/i386-c.c (ix86_target_macros_internal): Add
11476 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
11477 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
11478 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
11480 * config/i386/i386.c (rest_of_insert_endbranch): New.
11481 (pass_data_insert_endbranch): Likewise.
11482 (pass_insert_endbranch): Likewise.
11483 (make_pass_insert_endbranch): Likewise.
11484 (ix86_notrack_prefixed_insn_p): Likewise.
11485 (ix86_target_string): Add -mibt, -mshstk flags.
11486 (ix86_option_override_internal): Add flag_cf_protection
11488 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
11489 (ix86_print_operand): Add 'notrack' prefix output.
11490 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
11491 (ix86_expand_builtin): Expand CET intrinsics.
11492 (x86_output_mi_thunk): Add 'endbranch' instruction.
11493 * config/i386/i386.h (TARGET_IBT): New.
11494 (TARGET_IBT_P): Likewise.
11495 (TARGET_SHSTK): Likewise.
11496 (TARGET_SHSTK_P): Likewise.
11497 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
11498 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
11499 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
11500 (builtin_setjmp_setup): New pattern.
11501 (builtin_longjmp): Likewise.
11502 (rdssp<mode>): Likewise.
11503 (incssp<mode>): Likewise.
11504 (saveprevssp): Likewise.
11505 (rstorssp): Likewise.
11506 (wrss<mode>): Likewise.
11507 (wruss<mode>): Likewise.
11508 (setssbsy): Likewise.
11509 (clrssbsy): Likewise.
11510 (nop_endbr): Likewise.
11511 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
11513 * config/i386/immintrin.h: Include <cetintrin.h>.
11514 * config/i386/linux-common.h
11515 (file_end_indicate_exec_stack_and_cet): New prototype.
11516 (TARGET_ASM_FILE_END): New.
11518 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11520 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
11521 latencies instead of having separate table; make difference between
11522 integer and float costs.
11523 * i386.h (processor_costs): Remove scalar_stmt_cost,
11524 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
11525 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
11527 * x86-tune-costs.h: Remove entries which has been removed in
11528 procesor_costs from all tables; make cond_taken_branch_cost
11529 and cond_not_taken_branch_cost COST_N_INSNS based.
11531 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11533 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
11535 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11537 * config/i386/i386.md (isa): Remove fma_avx512f.
11538 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
11539 <avx512>_fmadd_<mode>_mask3<round_name>,
11540 <avx512>_fmsub_<mode>_mask<round_name>,
11541 <avx512>_fmsub_<mode>_mask3<round_name>,
11542 <avx512>_fnmadd_<mode>_mask<round_name>,
11543 <avx512>_fnmadd_<mode>_mask3<round_name>,
11544 <avx512>_fnmsub_<mode>_mask<round_name>,
11545 <avx512>_fnmsub_<mode>_mask3<round_name>,
11546 <avx512>_fmaddsub_<mode>_mask<round_name>,
11547 <avx512>_fmaddsub_<mode>_mask3<round_name>,
11548 <avx512>_fmsubadd_<mode>_mask<round_name>,
11549 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
11550 (*vec_widen_umult_even_v16si<mask_name>,
11551 *vec_widen_smult_even_v16si<mask_name>): Likewise.
11552 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
11554 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11556 * extend.texi: Add 'nocf_check' documentation.
11557 * gimple.texi: Add second parameter to
11558 gimple_build_call_from_tree.
11559 * invoke.texi: Add -fcf-protection documentation.
11560 * rtl.texi: Add REG_CALL_NOTRACK documenation.
11562 2017-10-20 Richard Biener <rguenther@suse.de>
11564 PR tree-optimization/82473
11565 * tree-vect-loop.c (vectorizable_reduction): Properly get at
11566 the largest input type.
11568 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11570 * c-attribs.c (handle_nocf_check_attribute): New function.
11571 (c_common_attribute_table): Add 'nocf_check' handling.
11572 * gimple-parser.c: Add second argument NULL to
11573 gimple_build_call_from_tree.
11574 * attrib.c (comp_type_attributes): Check nocf_check attribute.
11575 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
11577 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
11578 * common.opt: Add fcf-protection flag.
11579 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
11580 * flag-types.h: Add enum cf_protection_level.
11581 * gimple.c (gimple_build_call_from_tree): Add second parameter.
11582 Add 'nocf_check' attribute propagation to gimple call.
11583 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
11584 (gimple_build_call_from_tree): Update prototype.
11585 (gimple_call_nocf_check_p): New function.
11586 (gimple_call_set_nocf_check): Likewise.
11587 * gimplify.c: Add second argument to gimple_build_call_from_tree.
11588 * ipa-icf.c: Add nocf_check attribute in statement hash.
11589 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
11590 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
11591 * toplev.c (process_options): Add flag_cf_protection handling.
11593 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11595 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
11597 2017-10-20 Richard Biener <rguenther@suse.de>
11599 PR tree-optimization/82603
11600 * tree-if-conv.c (predicate_mem_writes): Make sure to only
11601 remove false predicated stores.
11603 2017-10-20 Richard Biener <rguenther@suse.de>
11605 * graphite-isl-ast-to-gimple.c
11606 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
11607 Remove return value and simplify, dump copied stmt after lhs
11609 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
11610 Reduce dump verbosity.
11611 (gsi_insert_earliest): Likewise.
11612 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
11613 * graphite.c (print_global_statistics): Adjust dumping.
11614 (print_graphite_scop_statistics): Likewise.
11615 (print_graphite_statistics): Do not dump loops here.
11616 (graphite_transform_loops): But here.
11618 2017-10-20 Nicolas Roche <roche@adacore.com>
11620 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
11621 * configure: Regenerate.
11623 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11626 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
11627 functions when optimizing replace GIMPLE_RETURN stmts with
11628 calls to __builtin_unreachable ().
11631 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
11632 for -fsanitize=thread link of executables.
11633 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
11634 link of executables.
11637 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
11638 New mode iterators.
11639 (<shift_insn><mode>3<mask_name>): Change the last of the 3
11640 define_insns for logical vector shifts to use VI248_AVX512BW
11641 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
11642 condition, useless isa and prefix attributes. Change the first
11643 2 of these define_insns to ...
11644 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
11645 define_insn for avx512vl.
11646 (<shift_insn><mode>3): ... and this, new define_insn without
11647 masking for non-avx512vl.
11650 * config/i386/sse.md (*andnot<mode>3,
11651 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
11652 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
11653 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
11654 not applied use empty suffix even for TARGET_AVX512VL.
11655 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
11656 is applied, supply evex,evex or evex,evex,evex instead of just
11659 2017-10-20 Julia Koval <julia.koval@intel.com>
11661 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
11662 (OPTION_MASK_ISA_GFNI_UNSET): New.
11663 (ix86_handle_option): Handle OPT_mgfni.
11664 * config/i386/cpuid.h (bit_GFNI): New.
11665 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
11666 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
11667 * config/i386/i386.c (ix86_target_string): Add -mgfni.
11668 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
11669 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
11670 * config/i386/i386.opt: Add mgfni.
11672 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
11674 * config/msp430/msp430.c (msp430_option_override): Disable
11675 -fdelete-null-pointer-checks.
11676 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
11678 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11680 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
11681 of x87 and SSE instructions.
11683 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11685 * asan.c (create_cond_insert_point): Do not update edge count.
11686 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
11687 (afdo_propagate_circuit): Likewise.
11688 (afdo_calculate_branch_prob): Likewise.
11689 (afdo_annotate_cfg): Likewise.
11690 * basic-block.h (struct edge_def): Remove count.
11691 (edge_def::count): New accessor.
11692 * bb-reorder.c (rotate_loop): Update.
11693 (find_traces_1_round): Update.
11694 (connect_traces): Update.
11695 (sanitize_hot_paths): Update.
11696 * cfg.c (unchecked_make_edge): Update.
11697 (make_single_succ_edge): Update.
11698 (check_bb_profile): Update.
11699 (dump_edge_info): Update.
11700 (update_bb_profile_for_threading): Update.
11701 (scale_bbs_frequencies_int): Update.
11702 (scale_bbs_frequencies_gcov_type): Update.
11703 (scale_bbs_frequencies_profile_count): Update.
11704 (scale_bbs_frequencies): Update.
11705 * cfganal.c (connect_infinite_loops_to_exit): Update.
11706 * cfgbuild.c (compute_outgoing_frequencies): Update.
11707 (find_many_sub_basic_blocks): Update.
11708 * cfgcleanup.c (try_forward_edges): Update.
11709 (try_crossjump_to_edge): Update
11710 * cfgexpand.c (expand_gimple_cond): Update
11711 (expand_gimple_tailcall): Update
11712 (construct_exit_block): Update
11713 * cfghooks.c (verify_flow_info): Update
11714 (redirect_edge_succ_nodup): Update
11715 (split_edge): Update
11716 (make_forwarder_block): Update
11717 (duplicate_block): Update
11718 (account_profile_record): Update
11719 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11720 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11721 * cfgloopmanip.c (scale_loop_profile): Update.
11723 (lv_adjust_loop_entry_edge): Update.
11724 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11725 (force_nonfallthru_and_redirect): Update.
11726 (purge_dead_edges): Update.
11727 (rtl_flow_call_edges_add): Update.
11728 * cgraphunit.c (init_lowered_empty_function): Update.
11729 (cgraph_node::expand_thunk): Update.
11730 * gimple-pretty-print.c (dump_probability): Update.
11731 (dump_edge_probability): Update.
11732 * gimple-ssa-isolate-paths.c (isolate_path): Update.
11733 * haifa-sched.c (sched_create_recovery_edges): Update.
11734 * hsa-gen.c (convert_switch_statements): Update.
11735 * ifcvt.c (dead_or_predicable): Update.
11736 * ipa-inline-transform.c (inline_transform): Update.
11737 * ipa-split.c (split_function): Update.
11738 * ipa-utils.c (ipa_merge_profiles): Update.
11739 * loop-doloop.c (add_test): Update.
11740 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
11741 * lto-streamer-in.c (input_cfg): Update.
11742 (input_function): Update.
11743 * lto-streamer-out.c (output_cfg): Update.
11744 * modulo-sched.c (sms_schedule): Update.
11745 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11746 * predict.c (maybe_hot_edge_p): Update.
11747 (unlikely_executed_edge_p): Update.
11748 (probably_never_executed_edge_p): Update.
11749 (dump_prediction): Update.
11750 (drop_profile): Update.
11751 (propagate_unlikely_bbs_forward): Update.
11752 (determine_unlikely_bbs): Update.
11753 (force_edge_cold): Update.
11754 * profile.c (compute_branch_probabilities): Update.
11755 * reg-stack.c (better_edge): Update.
11756 * shrink-wrap.c (handle_simple_exit): Update.
11757 * tracer.c (better_p): Update.
11758 * trans-mem.c (expand_transaction): Update.
11759 (split_bb_make_tm_edge): Update.
11760 * tree-call-cdce.c: Update.
11761 * tree-cfg.c (gimple_find_sub_bbs): Update.
11762 (gimple_split_edge): Update.
11763 (gimple_duplicate_sese_region): Update.
11764 (gimple_duplicate_sese_tail): Update.
11765 (gimple_flow_call_edges_add): Update.
11766 (insert_cond_bb): Update.
11767 (execute_fixup_cfg): Update.
11768 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
11769 * tree-complex.c (expand_complex_div_wide): Update.
11770 * tree-eh.c (lower_resx): Update.
11771 (unsplit_eh): Update.
11772 (cleanup_empty_eh_move_lp): Update.
11773 * tree-inline.c (copy_edges_for_bb): Update.
11774 (freqs_to_counts): Update.
11775 (copy_cfg_body): Update.
11776 * tree-ssa-dce.c (remove_dead_stmt): Update.
11777 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11778 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11779 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11780 (unloop_loops): Update.
11781 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11782 * tree-ssa-loop-split.c (connect_loops): Update.
11783 (split_loop): Update.
11784 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11785 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11786 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11787 * tree-ssa-reassoc.c (branch_fixup): Update.
11788 * tree-ssa-tail-merge.c (replace_block_by): Update.
11789 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11790 (compute_path_counts): Update.
11791 (update_profile): Update.
11792 (recompute_probabilities): Update.
11793 (update_joiner_offpath_counts): Update.
11794 (estimated_freqs_path): Update.
11795 (freqs_to_counts_path): Update.
11796 (clear_counts_path): Update.
11797 (ssa_fix_duplicate_block_edges): Update.
11798 (duplicate_thread_path): Update.
11799 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11800 (case_bit_test_cmp): Update.
11801 (collect_switch_conv_info): Update.
11802 (gen_inbound_check): Update.
11803 (do_jump_if_equal): Update.
11804 (emit_cmp_and_jump_insns): Update.
11805 * tree-tailcall.c (decrease_profile): Update.
11806 (eliminate_tail_call): Update.
11807 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11808 (vect_do_peeling): Update.
11809 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11810 * ubsan.c (ubsan_expand_null_ifn): Update.
11811 (ubsan_expand_ptr_ifn): Update.
11812 * value-prof.c (gimple_divmod_fixed_value): Update.
11813 (gimple_mod_pow2): Update.
11814 (gimple_mod_subtract): Update.
11815 (gimple_ic): Update.
11816 (gimple_stringop_fixed_value): Update.
11818 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11821 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11823 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11825 PR rtl-optimization/82395
11826 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11827 based on non_spilled_static_chain_regno_p.
11829 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11831 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11832 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11833 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11835 2017-10-19 Martin Sebor <msebor@redhat.com>
11837 PR tree-optimization/82596
11838 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11840 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11842 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11843 (get_mem_refs_of_builtin_call): Likewise.
11844 * builtins.c (expand_builtin_apply): Adjust call to
11845 allocate_dynamic_stack_space.
11846 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11847 the third argument to allocate_dynamic_stack_space, otherwise -1.
11848 (expand_builtin): Deal with all alloca variants.
11849 (is_inexpensive_builtin): Likewise.
11850 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11851 * calls.c (special_function_p): Deal with all alloca variants.
11852 (initialize_argument_information): Adjust call to
11853 allocate_dynamic_stack_space.
11854 (expand_call): Likewise.
11855 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11856 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11857 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11858 use it for the stack usage computation.
11859 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11860 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11861 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11862 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11863 (in_loop_p): Remove first argument and useless check.
11864 (pass_walloca::execute): Remove useless test and adjust call to above.
11865 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11866 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11867 (gimplify_call_expr): Deal with all alloca variants.
11868 * hsa-gen.c (gen_hsa_alloca): Likewise.
11869 (gen_hsa_insns_for_call): Likewise.
11870 * ipa-pure-const.c (special_builtin_state): Likewise.
11871 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11872 * tree-object-size.c (alloc_object_size): Likewise.
11873 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11874 (call_may_clobber_ref_p_1): Likewise.
11875 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11876 (ccp_fold_stmt): Likewise.
11877 (optimize_stack_restore): Likewise.
11878 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11879 (mark_all_reaching_defs_necessary_1): Likewise.
11880 (propagate_necessity): Likewise.
11881 (eliminate_unnecessary_stmts): Likewise.
11882 * tree.c (build_common_builtin_nodes): Build
11883 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11884 (build_alloca_call_expr): New function.
11885 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11886 (CASE_BUILT_IN_ALLOCA): Likewise.
11887 (build_alloca_call_expr): Declare.
11888 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11890 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11893 * dwarf2out.c (new_die_raw): New static inline function.
11894 (new_die): Use it to create the DIE.
11895 (add_AT_external_die_ref): Likewise.
11896 (clone_die): Likewise.
11897 (clone_as_declaration): Likewise.
11898 (dwarf2out_vms_debug_main_pointer): Likewise.
11899 (base_type_die): Likewise. Remove early return for corner cases.
11900 Do not call add_pubtype on the DIE here.
11901 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11902 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11903 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11904 native order exists for base types, attach the DIE manually and call
11905 add_pubtype on it. Do not equate a reverse order DIE to the type.
11907 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11909 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11910 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11912 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11914 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11916 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11917 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11918 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11919 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11920 * opts.c (sanitizer_opts): Add builtin.
11921 * ubsan.c (instrument_builtin): New function.
11922 (pass_ubsan::execute): Call it.
11923 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11924 * doc/invoke.texi: Document -fsanitize=builtin.
11926 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11927 builtins, store max (log2 (align), 0) into uchar field instead of
11928 align into uptr field.
11929 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11930 store uchar 0 field instead of uptr 0 field.
11931 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11932 instead of passing one address of struct with 2 locations pass
11933 two addresses of structs with 1 location each.
11934 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11935 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11936 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11937 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11938 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11939 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11940 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11941 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11943 2017-10-19 Martin Liska <mliska@suse.cz>
11946 * file-find.c (remove_prefix): Remove.
11947 * file-find.h (remove_prefix): Likewise.
11948 * gcc-ar.c: Remove smartness of lookup.
11950 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11952 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11953 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11954 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11956 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11959 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11960 (setcc + and to xor + setcc): New peephole2.
11962 2017-10-19 Tom de Vries <tom@codesourcery.com>
11964 * doc/sourcebuild.texi (Test Directives, Variants of
11965 dg-require-support): Add dg-require-stack-size.
11967 2017-10-19 Martin Liska <mliska@suse.cz>
11970 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11971 that have a large alignment.
11972 (gimplify_target_expr): Likewise.
11974 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11976 PR rtl-optimization/82602
11977 * ira.c (rtx_moveable_p): Return false for volatile asm.
11979 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11982 * config/i386/i386-modes.def (CCGZ): New CC mode.
11983 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11984 * config/i386/predicates.md (ix86_comparison_operator):
11986 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11987 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11988 with double-word subtraction.
11989 (put_condition_code): Handle CCGZmode.
11991 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11993 * wide-int.cc (debug (const wide_int &)): New.
11994 (debug (const wide_int *)): New.
11995 (debug (const widest_int &)): New.
11996 (debug (const widest_int *)): New.
11998 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
12000 PR middle-end/82556
12001 * lra-constraints.c (curr_insn_transform): Use non-input operand
12002 instead of output one for matched reload.
12004 2017-10-18 Bin Cheng <bin.cheng@arm.com>
12006 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
12007 (tree-ssa-loop-ivopts.h): New header file.
12008 (struct builtin_info): New fields.
12009 (classify_builtin_1): Compute and record base and offset parts for
12010 memset builtin partition by calling strip_offset.
12011 (offset_cmp, fuse_memset_builtins): New functions.
12012 (finalize_partitions): Fuse adjacent memset partitions by calling
12014 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
12015 Expose the interface.
12016 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
12018 2017-10-18 Bin Cheng <bin.cheng@arm.com>
12020 PR tree-optimization/82574
12021 * tree-loop-distribution.c (find_single_drs): New parameter. Check
12022 that data reference must be executed exactly once per iteration
12023 against the outermost loop in nest.
12024 (classify_partition): Update call to above function.
12026 2017-10-18 Richard Biener <rguenther@suse.de>
12028 PR tree-optimization/82591
12029 * graphite.c (graphite_transform_loops): Move code gen message
12031 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12032 Here. Handle scop_to_isl_ast failing.
12033 (scop_to_isl_ast): Limit the number of ISL operations.
12035 2017-10-18 Richard Biener <rguenther@suse.de>
12037 * graphite-isl-ast-to-gimple.c
12038 (translate_isl_ast_to_gimple::set_rename): Simplify.
12039 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
12040 (graphite_copy_stmts_from_block): ... here.
12041 (copy_bb_and_scalar_dependences): Simplify.
12042 (add_parameters_to_ivs_params): Canonicalize.
12043 (generate_entry_out_of_ssa_copies): Simplify.
12044 * graphite-sese-to-poly.c (extract_affine_name): Simplify
12045 by passing in ISL dimension.
12046 (parameter_index_in_region_1): Rename to ...
12047 (parameter_index_in_region): ... this.
12048 (extract_affine): Adjust assert, pass down parameter index.
12049 (add_param_constraints): Use range-info when available.
12050 (build_scop_context): Adjust.
12051 * sese.c (new_sese_info): Adjust.
12052 (free_sese_info): Likewise.
12053 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
12054 Remove unused typedefs.
12055 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
12057 2017-10-18 Martin Liska <mliska@suse.cz>
12059 * combine.c (simplify_compare_const): Add gcc_fallthrough.
12061 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
12063 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
12064 (s390_sched_init): Do not reset s390_sched_state if we entered the
12065 current basic block via a fallthru edge and all others are unlikely.
12067 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
12069 * config/s390/s390.c (NUM_SIDES): New variable.
12070 (LONGRUNNING_THRESHOLD): New variable.
12071 (LATENCY_FACTOR): New variable.
12072 (s390_sched_score): Decrease score for long-running instructions on
12074 (s390_sched_variable_issue): Perform bookkeeping for long-running
12077 2017-10-18 Richard Biener <rguenther@suse.de>
12079 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
12080 Simplify with removal of the parameter rename map.
12081 (set_rename): Likewise.
12082 (should_copy_to_new_region): Likewise.
12083 (graphite_copy_stmts_from_block): Likewise.
12084 (copy_bb_and_scalar_dependences): Remove initialization of
12085 unused copied_bb_map.
12086 (copy_def): Remove.
12087 (copy_internal_parameters): Likewise.
12088 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
12089 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
12090 Use INTEGRAL_TYPE_P.
12091 (parameter_index_in_region_1): Rename to ...
12092 (assign_parameter_index_in_region): ... this. Assert we have
12093 a parameter we handle.
12094 (scan_tree_for_params): Adjust.
12095 * sese.h (parameter_rename_map_t): Remove.
12096 (struct sese_info_t): Remove unused parameter_rename_map and
12097 copied_bb_map members.
12098 * sese.c (new_sese_info): Adjust.
12099 (free_sese_info): Likewise.
12101 2017-10-18 Martin Liska <mliska@suse.cz>
12104 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
12105 on an abnormal edge.
12107 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
12109 * doc/invoke.texi (ffunction-sections and fdata-sections):
12112 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
12114 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
12115 the use statement can throw internally.
12117 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
12119 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
12120 any RTX present on the RHS of a SET.
12121 * compare-elim.c (try_eliminate_compare): Restore comment.
12123 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12125 * langhooks.h (struct lang_hooks): Document that tree_size langhook
12126 may be also called on tcc_type nodes.
12127 * langhooks.c (lhd_tree_size): Likewise.
12129 2017-10-17 David Malcolm <dmalcolm@redhat.com>
12131 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
12132 format_warning_at_substring.
12133 (maybe_warn): Convert source_range * param to a location_t. Pass
12134 UNKNOWN_LOCATION rather than NULL to fmtwarn.
12135 (format_directive): Remove code to extract source_ranges and
12136 source_range * in favor of just a location_t.
12137 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
12139 * substring-locations.c (format_warning_va): Convert
12140 source_range * param to a location_t.
12141 (format_warning_at_substring): Likewise.
12142 * substring-locations.h (format_warning_va): Likewise.
12143 (format_warning_at_substring): Likewise.
12145 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
12147 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
12149 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
12150 and scatter/gather ops.
12152 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
12153 vec_gather_load and vec_scatter_store.
12154 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
12155 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
12156 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
12157 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
12158 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
12159 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
12161 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
12163 * reg-stack.c (compare_for_stack_reg): Add bool argument.
12164 Detect FTST instruction and handle its register pops. Only pop
12165 second operand if can_pop_second_op is true.
12166 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
12167 set can_pop_second_op to false in the compare_for_stack_reg call.
12169 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
12170 output_fp_compare for stack register operands.
12171 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
12172 instructions here. Do not emit stack register pops here. Assert
12173 that FCOMPP pops next to top stack register. Rewrite function.
12175 2017-10-17 Nathan Sidwell <nathan@acm.org>
12177 PR middle-end/82577
12178 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
12179 use DECL_ASSEMBLER_NAME_RAW.
12181 PR middle-end/82546
12182 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
12185 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
12186 Wilco Dijkstra <wilco.dijkstra@arm.com>
12188 * builtins.c (expand_builtin_update_setjmp_buf): Add a
12189 converstion to Pmode from the buf_addr.
12191 2017-10-17 Richard Biener <rguenther@suse.de>
12193 * graphite-dependences.c (scop_get_reads_and_writes): Change
12194 output parameters to references.
12196 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
12198 PR 71026/tree-optimization
12199 * fold-const.c (distribute_real_division): Removed.
12200 (fold_binary_loc): Remove calls to distribute_real_divison.
12202 2017-10-17 Richard Biener <rguenther@suse.de>
12204 * graphite-scop-detection.c
12205 (scop_detection::stmt_has_simple_data_refs_p): Always use
12206 the full nest as region.
12207 (try_generate_gimple_bb): Likewise.
12208 * sese.c (scalar_evolution_in_region): Simplify now that
12209 SCEV can handle instantiation in regions.
12210 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
12211 in the non-loop part of a function if requested.
12213 2017-10-17 Richard Biener <rguenther@suse.de>
12215 PR tree-optimization/82563
12216 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
12218 (graphite_regenerate_ast_isl): Call it.
12219 * graphite-scop-detection.c (build_scops): Remove entry edge split.
12221 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12223 PR tree-optimization/82549
12224 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
12225 Formatting fixes. Instead of calling make_bit_field_ref with negative
12228 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
12230 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
12231 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
12232 _mm_maskz_reduce_ss): New.
12233 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
12234 __builtin_ia32_reducess_mask): Ditto..
12235 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
12236 * config/i386/sse.md (reduces<mode>): Renamed to ...
12237 (reduces<mode><mask_scalar_name>): ... this.
12238 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
12240 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
12241 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
12243 2017-10-16 David Malcolm <dmalcolm@redhat.com>
12245 * Makefile.in (OBJS): Add unique-ptr-tests.o.
12246 * selftest-run-tests.c (selftest::run_tests): Call
12247 selftest::unique_ptr_tests_cc_tests.
12248 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
12249 * unique-ptr-tests.cc: New file.
12251 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
12254 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12256 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12257 (make_hard_regno_born, make_hard_regno_dead): Update
12258 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
12260 2017-10-16 Jeff Law <law@redhat.com>
12262 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12264 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
12266 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
12268 2017-10-16 Olivier Hainque <hainque@adacore.com>
12270 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
12271 with_cpu if we were configured for an e500v2 target cpu name.
12273 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
12275 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
12276 * doc/invoke.texi: Document +nodsp as a valid extension for
12279 2017-10-16 Martin Liska <mliska@suse.cz>
12281 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
12282 (test_set_range): Likewise.
12283 (test_range_functions): Rename to ...
12284 (test_bit_in_range): ... this.
12285 (sbitmap_c_tests): Add new test.
12287 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12289 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
12291 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
12292 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
12294 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12296 * config/aarch64/aarch64-builtins.c
12297 (aarch64_types_quadopu_lane_qualifiers): New.
12298 (TYPES_QUADOPU_LANE): New.
12299 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
12300 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
12301 (aarch64_<sur>dot_laneq<vsi2qi>): New.
12302 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
12303 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
12304 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
12305 (Vdottype, DOTPROD): New.
12306 (sur): Add SDOT and UDOT.
12308 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12310 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
12311 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
12312 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
12313 Add TARGET_DOTPROD.
12314 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
12315 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
12316 Enable TARGET_DOTPROD.
12317 (cortex-a75.cortex-a55): Likewise.
12318 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
12320 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12322 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
12323 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
12325 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
12327 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
12328 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
12329 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
12330 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
12331 * config/arm/types.md (neon_dot, neon_dot_q): New.
12332 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
12334 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12336 * config/arm/arm.h (TARGET_DOTPROD): New.
12337 * config/arm/arm.c (arm_arch_dotprod): New.
12338 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
12339 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
12340 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
12341 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
12342 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
12343 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
12344 * doc/invoke.texi (armv8.2-a): Document dotprod
12346 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12348 * i386.c (ix86_vec_cost): New function.
12349 (ix86_rtx_costs): Handle vector operations better.
12350 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
12351 * x86-tune-costs.h: Add new costs to all tables.
12353 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12355 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
12357 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
12358 divsd, sqrtss and sqrtsd
12359 * x86-tune-costs.h: Add new entries to all costs.
12360 (znver1_cost): Fix to match real instruction latencies.
12362 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12363 Michael Collison <michael.collison@arm.com>
12365 * compare-elim.c: Include emit-rtl.h.
12366 (can_merge_compare_into_arith): New function.
12367 (try_validate_parallel): Likewise.
12368 (try_merge_compare): Likewise.
12369 (try_eliminate_compare): Call the above when no previous clobber
12371 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
12374 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12376 PR middle-end/62263
12377 PR middle-end/82498
12378 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
12379 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
12381 PR middle-end/62263
12382 PR middle-end/82498
12383 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
12384 to be any operand_equal_p operands. For & (B - 1) require
12385 B to be power of 2. Recognize
12386 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
12388 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12391 * optabs.c (expand_memory_blockage): Fix call of
12392 targetm.have_memory_blockage.
12394 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12397 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
12398 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
12399 objects to extra_objs instead of overwriting it.
12401 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12403 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
12404 Use any_fp_register_operand as operand[3] predicate. Simplify
12405 equality test for operands[2] and operands[4] memory location.
12406 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
12407 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
12408 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
12409 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
12410 any_fp_register_operand as operand[1] predicate. Simplify
12411 equality test for operands[0] and operands[3] memory location.
12412 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
12413 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
12414 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
12416 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12418 * target-insns.def: Add memory_blockage.
12419 * optabs.c (expand_memory_blockage): New function.
12420 (expand_asm_memory_barrier): Rename ...
12421 (expand_asm_memory_blockage): ... to this.
12422 (expand_mem_thread_fence): Call expand_memory_blockage
12423 instead of expand_asm_memory_barrier.
12424 (expand_mem_singnal_fence): Ditto.
12425 (expand_atomic_load): Ditto.
12426 (expand_atomic_store): Ditto.
12427 * doc/md.texi (Standard Pattern Names For Generation):
12428 Document memory_blockage instruction pattern.
12430 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
12432 * config/rl78/rl78.c (rl78_emit_libcall): New function.
12433 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
12434 * config/rl78/rl78.md: New define_expand "adddi3".
12436 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12438 * cfghooks.c (verify_flow_info): Disable check that all probabilities
12441 2017-10-13 Jeff Law <law@redhat.com>
12443 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
12445 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12448 * internal-fn.c (expand_mul_overflow): If both operands have
12449 the same highpart of -1 or 0 and the topmost bit of lowpart
12450 is different, overflow is if res <= 0 rather than res < 0.
12452 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
12454 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
12455 TARGET_P9_VECTOR code for unaligned_load case.
12457 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12459 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
12461 2017-10-13 Nathan Sidwell <nathan@acm.org>
12463 * tree-core.h (tree_contains_struct): Make bool.
12464 * tree.c (tree_contains_struct): Likewise.
12465 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
12466 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
12467 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
12468 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
12469 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
12471 2017-10-13 Richard Biener <rguenther@suse.de>
12473 * graphite-isl-ast-to-gimple.c
12474 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
12475 parameters and dominance check.
12476 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
12477 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
12478 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
12479 Do not update SSA form here or do intermediate IL verification.
12480 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
12481 (graphite_initialize): Remove check on the number of loops in
12482 the function and inline into graphite_transform_loops.
12483 (graphite_finalize): Inline into graphite_transform_loops.
12484 (graphite_transform_loops): Perform SSA update and IL verification
12486 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
12488 2017-10-13 Richard Biener <rguenther@suse.de>
12490 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
12491 graphite_expression_type_precision): Avoid global constructor
12493 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
12494 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
12495 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
12496 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
12497 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
12498 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
12500 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
12503 * config/i386/i386.h (ix86_red_zone_size): New.
12504 * config/i386/i386.md (push peephole2s): Replace
12505 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
12507 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12508 Alan Hayward <alan.hayward@arm.com>
12509 David Sherwood <david.sherwood@arm.com>
12511 * combine.c (can_change_dest_mode): Reject changes in
12512 REGMODE_NATURAL_SIZE.
12514 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12515 Alan Hayward <alan.hayward@arm.com>
12516 David Sherwood <david.sherwood@arm.com>
12518 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
12519 (expand_debug_source_expr): Likewise.
12520 * combine.c (combine_simplify_rtx): Likewise.
12521 * cse.c (fold_rtx): Likewise.
12522 * fwprop.c (canonicalize_address): Likewise.
12523 * targhooks.c (default_shift_truncation_mask): Likewise.
12525 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12526 Alan Hayward <alan.hayward@arm.com>
12527 David Sherwood <david.sherwood@arm.com>
12529 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
12530 (widened_mode): Likewise.
12531 (expand_unop): Likewise.
12532 * ree.c (transform_ifelse): Likewise.
12533 (merge_def_and_ext): Likewise.
12534 (combine_reaching_defs): Likewise.
12535 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12537 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12538 Alan Hayward <alan.hayward@arm.com>
12539 David Sherwood <david.sherwood@arm.com>
12541 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
12542 * combine.c (gen_lowpart_for_combine): Likewise.
12543 * dwarf2out.c (rtl_for_decl_location): Likewise.
12544 * final.c (alter_subreg): Likewise.
12545 * rtlhooks.c (gen_lowpart_general): Likewise.
12546 (gen_lowpart_if_possible): Likewise.
12548 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12549 Alan Hayward <alan.hayward@arm.com>
12550 David Sherwood <david.sherwood@arm.com>
12552 * calls.c (expand_call): Use subreg_lowpart_offset.
12553 * cse.c (cse_insn): Likewise.
12554 * regcprop.c (copy_value): Likewise.
12555 (copyprop_hardreg_forward_1): Likewise.
12557 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12560 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
12561 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
12562 =Q constraints to +Q and into insn condition add check
12563 that operands[0] and operands[1] are equal.
12564 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
12565 =Q constraints to +Q and into insn condition add check
12566 that operands[0] is equal to either operands[1] or operands[2].
12569 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
12570 instead of handling MINUS_EXPR twice (once for each argument),
12571 canonicalize operand order and handle just once, use rtype where
12572 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
12575 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
12576 any values of __C while still being pattern recognizable as a simple
12577 rotate instruction.
12579 2017-10-13 Richard Biener <rguenther@suse.de>
12581 PR tree-optimization/82451
12583 2017-10-02 Richard Biener <rguenther@suse.de>
12585 PR tree-optimization/82355
12586 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12587 a mapping for the enclosing loop but avoid generating one for
12588 the loop tree root.
12589 (copy_bb_and_scalar_dependences): Remove premature codegen
12590 error on PHIs in blocks duplicated into multiple places.
12591 * graphite-scop-detection.c
12592 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12593 in the region use it as loop and nest to analyze the DR in.
12594 (try_generate_gimple_bb): Likewise.
12595 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12596 (add_loop_constraints): For blocks in a loop not in the region
12597 create a dimension with a single iteration.
12598 * sese.h (gbb_loop_at_index): Remove assert.
12600 * cfgloop.c (loop_preheader_edge): For the loop tree root
12601 return the single successor of the entry block.
12602 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12603 Reset the SCEV hashtable and niters.
12604 * graphite-scop-detection.c
12605 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
12606 assert that we only have POLYNOMIAL_CHREC that vary in loops
12607 contained in the region.
12608 (scop_detection::graphite_can_represent_expr): Adjust.
12609 (scop_detection::stmt_has_simple_data_refs_p): For loops
12610 not in the region set loop to NULL. The nest is now the
12611 entry edge to the region.
12612 (try_generate_gimple_bb): Likewise.
12613 * sese.c (scalar_evolution_in_region): Adjust for
12614 instantiate_scev change.
12615 * tree-data-ref.h (graphite_find_data_references_in_stmt):
12616 Make nest parameter the edge into the region.
12617 (create_data_ref): Likewise.
12618 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
12619 entry edge into a region and adjust instantiate_scev calls.
12620 (create_data_ref): Likewise.
12621 (graphite_find_data_references_in_stmt): Likewise.
12622 (find_data_references_in_stmt): Pass the loop preheader edge
12623 from the nest argument.
12624 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
12625 parameter the edge into the region.
12626 (instantiate_parameters): Use the loop preheader edge as entry.
12627 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
12629 (get_instantiated_value_entry): Make instantiate_below parameter
12630 the edge into the region.
12631 (instantiate_scev_name): Likewise. Adjust dominance checks,
12632 when we cannot use loop-based instantiation instantiate by
12633 walking use-def chains.
12634 (instantiate_scev_poly): Adjust.
12635 (instantiate_scev_binary): Likewise.
12636 (instantiate_scev_convert): Likewise.
12637 (instantiate_scev_not): Likewise.
12638 (instantiate_array_ref): Remove.
12639 (instantiate_scev_3): Likewise.
12640 (instantiate_scev_2): Likewise.
12641 (instantiate_scev_1): Likewise.
12642 (instantiate_scev_r): Do not blindly handle N-operand trees.
12643 Do not instantiate array-refs. Handle all constants and invariants.
12644 (instantiate_scev): Make instantiate_below parameter
12645 the edge into the region.
12646 (resolve_mixers): Use the loop preheader edge for the region
12647 parameter to instantiate_scev_r.
12648 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
12650 2017-10-13 Richard Biener <rguenther@suse.de>
12652 PR tree-optimization/82525
12653 * graphite-isl-ast-to-gimple.c
12654 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
12656 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
12657 Fail code generation when we cannot represent the isl integer.
12658 (binary_op_to_tree): Elide modulo operations that are no-ops
12659 in the type we code generate. Remove now superfluous code
12662 2017-10-13 Richard Biener <rguenther@suse.de>
12664 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
12665 (scop_detection::harmful_loop_in_region): Remove premature
12666 IV type restriction.
12667 (scop_detection::graphite_can_represent_scev): We can handle
12668 pointer IVs just fine.
12670 2017-10-13 Alan Modra <amodra@gmail.com>
12672 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
12673 "Clobbers and Scratch Registers". Add paragraph on
12674 alternative to clobbers for scratch registers and OpenBLAS
12677 2017-10-13 Alan Modra <amodra@gmail.com>
12679 * doc/extend.texi (Clobbers): Correct vax example. Delete old
12680 example of a memory input for a string of known length. Move
12681 commentary out of table. Add a number of new examples
12682 covering array memory inputs.
12684 2017-10-12 Martin Liska <mliska@suse.cz>
12686 PR tree-optimization/82493
12687 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
12688 (test_range_functions): New function.
12689 (sbitmap_c_tests): Likewise.
12690 * selftest-run-tests.c (selftest::run_tests): Run new tests.
12691 * selftest.h (sbitmap_c_tests): New function.
12693 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12695 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12697 * config/rs6000/amo.h: Fix spacing issue.
12699 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12702 * config/i386/i386.md (*ashl<mode>3_mask_1,
12703 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
12704 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
12707 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12709 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
12710 (profile_probability): Set max_probability
12711 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
12713 * profile-count.c (profile_probability::differs_from_p): Do not
12714 rely on max_probaiblity == 10000
12716 2017-10-12 Jeff Law <law@redhat.com>
12718 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
12721 2017-10-12 Martin Sebor <msebor@redhat.com>
12725 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
12726 (handle_alias_pairs): Call it.
12727 * common.opt (-Wattribute-alias): New option.
12728 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
12729 * doc/invoke.texi (-Wattribute-alias): Document.
12731 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
12734 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12736 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12738 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12739 (make_hard_regno_born, make_hard_regno_dead): Update
12740 bb_killed_pseudos and bb_gen_pseudos.
12742 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12744 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
12746 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
12748 * config/alpha/alpha.c (alpha_split_conditional_move):
12749 Use std::swap instead of manually swapping.
12750 (alpha_stdarg_optimize_hook): Ditto.
12751 (alpha_canonicalize_comparison): Ditto.
12753 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12755 * tree-loop-distribution.c (struct builtin_info): New struct.
12756 (struct partition): Refactor fields into struct builtin_info.
12757 (partition_free): Free struct builtin_info.
12758 (build_size_arg_loc, build_addr_arg_loc): Delete.
12759 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
12760 information from struct builtin_info.
12761 (find_single_drs): New function refactored from classify_partition.
12762 Also moved builtin validity checks to this function.
12763 (compute_access_range, alloc_builtin): New functions.
12764 (classify_builtin_st, classify_builtin_ldst): New functions.
12765 (classify_partition): Refactor code into functions find_single_drs,
12766 classify_builtin_st and classify_builtin_ldst.
12767 (distribute_loop): Don't do runtime alias check when distributing
12769 (find_seed_stmts_for_distribution): New function.
12770 (pass_loop_distribution::execute): Refactor code finding seed
12771 stmts into above function. Support distribution for the innermost
12772 two-level loop nest. Adjust dump information.
12774 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12776 * tree-loop-distribution.c: Adjust the general comment.
12777 (NUM_PARTITION_THRESHOLD): New macro.
12778 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
12779 (classify_partition): Skip builtin pattern of loop nest's inner loop.
12780 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
12781 in call to build_partition_graph.
12782 (finalize_partitions): New parameter. Make loop distribution more
12783 conservative by fusing more partitions.
12784 (distribute_loop): Don't do runtime alias check in case of loop nest
12786 (find_seed_stmts_for_distribution): New function.
12787 (prepare_perfect_loop_nest): New function.
12788 (pass_loop_distribution::execute): Refactor code finding seed stmts
12789 and loop nest into above functions. Support loop nest distribution.
12790 Adjust dump information accordingly.
12792 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12794 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12795 and set PTYPE_SEQUENTIAL for merged partition.
12797 2017-10-12 Richard Biener <rguenther@suse.de>
12799 PR tree-optimization/69728
12801 2017-09-19 Richard Biener <rguenther@suse.de>
12803 PR tree-optimization/69728
12804 * graphite-sese-to-poly.c (schedule_error): New global.
12805 (add_loop_schedule): Handle empty domain by failing the
12807 (build_original_schedule): Handle schedule_error.
12809 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12810 domain by returning an unchanged schedule.
12812 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12814 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12815 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12817 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12819 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12822 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12825 * expr.c (store_field): Don't optimize away bitsize == 0 store
12826 from CALL_EXPR with addressable return type.
12828 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12830 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12831 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12832 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12833 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12834 TARGET_ISEL instead of TARGET_ISEL<sel>.
12836 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12838 * config/rs6000/rs6000.c
12839 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12841 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12843 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12845 (reg_or_cint_operand): Fix comment.
12846 (reg_or_zero_operand): New predicate.
12847 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12848 * config/rs6000/rs6000.c (output_isel): Delete.
12849 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12850 instead of reg_or_cint_operand. Output instruction directly (not via
12852 (isel_unsigned_<mode>): Ditto.
12853 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12854 gpc_reg_operand. Add an instruction alternative for this. Output
12855 instruction directly.
12856 (*isel_reversed_unsigned_<mode>): Ditto.
12858 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12860 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12861 (TARGET_CANONICALIZE_COMPARISON): Define.
12863 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12866 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12867 Check whether the dest is REG before adding REG_EQUIV note.
12869 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12872 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12874 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12875 (make_hard_regno_born, make_hard_regno_dead): Update
12876 bb_killed_pseudos and bb_gen_pseudos.
12878 2017-10-11 Nathan Sidwell <nathan@acm.org>
12880 * incpath.h (enum incpath_kind): Name enum, prefix values.
12881 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12882 * incpath.c (heads, tails): Use INC_MAX.
12883 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12884 (merge_include_chains, split_quote_chain,
12885 register_include_chains): Update incpath_kind names.
12886 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12887 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12889 (add_framework_path, darwin_register_objc_includes): Likewise.
12890 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12892 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12894 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12895 Do not use float_operator operator predicate.
12896 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12897 * config/i386/predicates.md (float_operator): Remove predicate.
12899 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12901 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12902 (*jccxf_i387): Ditto.
12903 (*jcc<mode>_i387): Ditto.
12904 (*jccu<mode>_i387): Ditto.
12905 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12906 (*jcc_*_i387 splitters): Remove.
12907 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12908 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12909 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12912 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12914 * profile-count.h (slow_safe_scale_64bit): New function.
12915 (safe_scale_64bit): New inline.
12916 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12917 * profile-count.c: Include wide-int.h
12918 (slow_safe_scale_64bit): New.
12920 2017-10-11 Nathan Sidwell <nathan@acm.org>
12922 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12923 HAS_DECL_ASSEMBLER_NAME_P.
12924 * gimple-expr.c (gimple_decl_printable_name: Check
12925 HAS_DECL_ASSEMBLER_NAME_P too.
12926 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12927 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12928 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12929 * passes.c (rest_of_decl_compilation): Check
12930 HAS_DECL_ASSEMBLER_NAME_P too.
12931 * recog.c (verify_changes): Likewise.
12932 * tree-pretty-print.c (dump_decl_name): Likewise.
12933 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12935 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12936 (SET_DECL_ASSEMBLER_NAME): Use it.
12937 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12938 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12939 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12941 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12943 * config.gcc (i386, x86_64): Add extra objects.
12944 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12945 (ix86_min_insn_size): Declare.
12946 (ix86_issue_rate): Declare.
12947 (ix86_adjust_cost): Declare.
12948 (ia32_multipass_dfa_lookahead): Declare.
12949 (ix86_macro_fusion_p): Declare.
12950 (ix86_macro_fusion_pair_p): Declare.
12951 (ix86_bd_has_dispatch): Declare.
12952 (ix86_bd_do_dispatch): Declare.
12953 (ix86_core2i7_init_hooks): Declare.
12954 (ix86_atom_sched_reorder): Declare.
12955 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12956 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12957 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12958 (rip_relative_addr_p): Rename to ...
12959 (ix86_rip_relative_addr_p): ... this one; export.
12960 (memory_address_length): Update.
12961 (ix86_issue_rate): Move to x86-tune-sched.c.
12962 (ix86_flags_dependent): Move to x86-tune-sched.c.
12963 (ix86_agi_dependent): Move to x86-tune-sched.c.
12964 (exact_dependency_1): Move to x86-tune-sched.c.
12965 (exact_store_load_dependency): Move to x86-tune-sched.c.
12966 (ix86_adjust_cost): Move to x86-tune-sched.c.
12967 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12968 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12969 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12970 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12971 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12972 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12973 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12974 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12975 (min_insn_size): Rename to ...
12976 (ix86_min_insn_size): ... this one; export.
12977 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12978 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12979 (core2i7_first_cycle_multipass_backtrack): Move to
12980 x86-tune-sched-core.c.
12981 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12982 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12983 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12984 (ix86_avoid_jump_mispredicts): Update.
12985 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12986 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12987 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12988 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12989 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12990 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12991 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12992 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12993 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12994 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12995 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12996 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12997 (BIG): Move to ix86-tune-sched-bd.c.
12998 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12999 (enum insn_path): Move to ix86-tune-sched-bd.c.
13000 (get_mem_group): Move to ix86-tune-sched-bd.c.
13001 (is_cmp): Move to ix86-tune-sched-bd.c.
13002 (dispatch_violation): Move to ix86-tune-sched-bd.c.
13003 (is_branch): Move to ix86-tune-sched-bd.c.
13004 (is_prefetch): Move to ix86-tune-sched-bd.c.
13005 (init_window): Move to ix86-tune-sched-bd.c.
13006 (allocate_window): Move to ix86-tune-sched-bd.c.
13007 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
13008 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
13009 (process_end_window): Move to ix86-tune-sched-bd.c.
13010 (allocate_next_window): Move to ix86-tune-sched-bd.c.
13011 (find_constant): Move to ix86-tune-sched-bd.c.
13012 (get_num_immediates): Move to ix86-tune-sched-bd.c.
13013 (has_immediate): Move to ix86-tune-sched-bd.c.
13014 (get_insn_path): Move to ix86-tune-sched-bd.c.
13015 (get_insn_group): Move to ix86-tune-sched-bd.c.
13016 (count_num_restricted): Move to ix86-tune-sched-bd.c.
13017 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
13018 (add_insn_window): Move to ix86-tune-sched-bd.c.
13019 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
13020 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
13021 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
13022 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
13023 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
13024 (do_dispatch): Move to ix86-tune-sched-bd.c.
13025 (has_dispatch): Move to ix86-tune-sched-bd.c.
13026 * config/i386/t-i386: Add new object files.
13027 * config/i386/x86-tune-costs.h: New file.
13028 * config/i386/x86-tune-sched-atom.c: New file.
13029 * config/i386/x86-tune-sched-bd.c: New file.
13030 * config/i386/x86-tune-sched-core.c: New file.
13031 * config/i386/x86-tune-sched.c: New file.
13033 2017-10-11 Liu Hao <lh_mouse@126.com>
13035 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
13036 the generic version below instead.
13037 (should_colorize): Recognize Windows consoles as terminals
13039 * pretty-print.c [__MINGW32__] (write_all): New function.
13040 [__MINGW32__] (find_esc_head): Likewise.
13041 [__MINGW32__] (find_esc_terminator): Likewise.
13042 [__MINGW32__] (eat_esc_sequence): Likewise.
13043 [__MINGW32__] (mingw_ansi_fputs): New function that handles
13045 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
13048 2017-10-11 Richard Biener <rguenther@suse.de>
13050 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
13051 Properly call analyze_scalar_evolution with the loop of the stmt.
13053 2017-10-11 Richard Biener <rguenther@suse.de>
13055 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
13056 * tree-core.h (tree_base): Add chrec_var union member.
13057 * tree.h (CHREC_VAR): Remove.
13058 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
13059 * tree-chrec.h (build_polynomial_chrec): Adjust.
13060 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
13061 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
13063 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
13065 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
13066 * match.pd: ... here.
13067 ((T) X == (T) Y): Relax condition.
13069 2017-10-11 Bin Cheng <bin.cheng@arm.com>
13071 PR tree-optimization/82472
13072 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
13074 (break_alias_scc_partitions): Update postorder number.
13076 2017-10-11 Martin Liska <mliska@suse.cz>
13079 * opts.c (parse_no_sanitize_attribute): Do not use error_value
13081 * opts.h (parse_no_sanitize_attribute): Remove last argument.
13083 2017-10-11 Martin Liska <mliska@suse.cz>
13085 * print-rtl.c (print_insn): Move declaration of idbuf
13086 to same scope as name.
13088 2017-10-11 Martin Liska <mliska@suse.cz>
13093 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
13094 volatile arguments.
13096 2017-10-11 Martin Liska <mliska@suse.cz>
13099 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
13100 volatile arguments.
13102 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
13104 * config.gcc (default_gnu_indirect_function): Default to yes for
13105 arm*-*-linux* with glibc.
13107 2017-10-11 Richard Biener <rguenther@suse.de>
13109 * tree-scalar-evolution.c (get_scalar_evolution): Handle
13110 default-defs and types we do not want to analyze.
13111 (interpret_loop_phi): Replace unreachable code with an assert.
13112 (compute_scalar_evolution_in_loop): Remove and inline ...
13113 (analyze_scalar_evolution_1): ... here, replacing condition with
13114 what makes the intent clearer. Remove handling of cases
13115 get_scalar_evolution now handles.
13117 2017-10-10 Jim Wilson <wilson@tuliptree.org>
13119 PR rtl-optimization/81434
13120 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
13121 comment for main loop. In sched_group_found if, also add checks for
13122 pass and min_cost_group.
13124 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
13126 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
13127 (rs6000_insn_cost): New function.
13128 * config/rs6000/rs6000.md (cost): New attribute.
13130 2017-10-10 Jakub Jelinek <jakub@redhat.com>
13131 H.J. Lu <hongjiu.lu@intel.com>
13135 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
13136 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
13137 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
13138 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
13139 to OPTION_MASK_ISA_AVX512VL - builtins that have both
13140 OPTION_MASK_ISA_MMX and some other bit set require both
13141 mmx and the ISAs without the mmx bit.
13142 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
13143 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
13144 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
13145 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
13146 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
13147 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
13148 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
13149 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
13150 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
13151 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
13152 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
13153 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
13154 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
13155 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
13156 Add OPTION_MASK_ISA_MMX.
13158 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
13160 * config.gcc (armv7*-*-freebsd*): New target.
13161 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
13163 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
13165 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
13166 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
13169 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
13171 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
13172 a template parameter.
13173 (WIDE_INT_REF_FOR): Update accordingly.
13174 * tree.h (wi::int_traits <const_tree>): Delete.
13175 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
13176 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
13177 (wi::tree_to_wide_ref): New typedef.
13178 (wi::to_wide): New function.
13179 * calls.c (get_size_range): Use wi::to_wide when operating on
13180 trees as wide_ints.
13181 * cgraph.c (cgraph_node::create_thunk): Likewise.
13182 * config/i386/i386.c (ix86_data_alignment): Likewise.
13183 (ix86_local_alignment): Likewise.
13184 * dbxout.c (stabstr_O): Likewise.
13185 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
13186 * expr.c (const_vector_from_tree): Likewise.
13187 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
13188 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
13189 (fold_negate_expr_1, int_const_binop_1, const_binop)
13190 (fold_convert_const_int_from_real, optimize_bit_field_compare)
13191 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
13192 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
13193 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
13194 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
13195 (fold_not_const, round_up_loc): Likewise.
13196 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
13197 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
13198 (alloca_call_type): Likewise.
13199 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
13200 * godump.c (go_output_typedef): Likewise.
13201 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
13202 * internal-fn.c (get_min_precision): Likewise.
13203 * ipa-cp.c (ipcp_store_vr_results): Likewise.
13204 * ipa-polymorphic-call.c
13205 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
13206 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
13207 (ipa_modify_call_arguments): Likewise.
13208 * match.pd: Likewise.
13209 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
13210 * print-tree.c (print_node_brief, print_node): Likewise.
13211 * stmt.c (expand_case): Likewise.
13212 * stor-layout.c (layout_type): Likewise.
13213 * tree-affine.c (tree_to_aff_combination): Likewise.
13214 * tree-cfg.c (group_case_labels_stmt): Likewise.
13215 * tree-data-ref.c (dr_analyze_indices): Likewise.
13216 (prune_runtime_alias_test_list): Likewise.
13217 * tree-dump.c (dequeue_and_dump): Likewise.
13218 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
13219 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
13220 * tree-pretty-print.c (dump_generic_node): Likewise.
13221 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
13222 (simple_iv_with_niters): Likewise.
13223 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
13224 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
13225 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
13226 * tree-ssa-loop-niter.c (split_to_var_and_offset)
13227 (refine_value_range_using_guard, number_of_iterations_ne_max)
13228 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
13229 (get_cst_init_from_scev, record_nonwrapping_iv)
13230 (scev_var_range_cant_overflow): Likewise.
13231 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
13232 * tree-ssa-pre.c (compute_avail): Likewise.
13233 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
13234 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
13235 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
13236 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
13237 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
13238 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
13239 (dump_case_nodes, try_switch_expansion): Likewise.
13240 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
13241 (vect_do_peeling): Likewise.
13242 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
13243 * tree-vect-stmts.c (vectorizable_load): Likewise.
13244 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
13245 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
13246 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
13247 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
13248 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
13249 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
13250 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
13251 (evrp_dom_walker::before_dom_children): Likewise.
13252 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
13253 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
13254 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
13255 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
13256 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
13257 (get_range_pos_neg): Likewise.
13258 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
13259 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
13260 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
13261 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
13262 * config/avr/avr.c (avr_fold_builtin): Likewise.
13263 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
13264 * config/msp430/msp430.c (msp430_attr): Likewise.
13265 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
13266 * config/powerpcspe/powerpcspe-c.c
13267 (altivec_resolve_overloaded_builtin): Likewise.
13268 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
13269 (rs6000_expand_ternop_builtin): Likewise.
13270 * config/rs6000/rs6000-c.c
13271 (altivec_resolve_overloaded_builtin): Likewise.
13272 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
13273 (rs6000_expand_ternop_builtin): Likewise.
13274 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
13276 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13278 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
13279 when copying loop nest with only one inner loop.
13281 2017-10-10 Richard Biener <rguenther@suse.de>
13283 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
13284 blocks if SCEV is active.
13285 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
13287 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
13288 (scev_initialize): Assert we are not yet initialized.
13290 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13292 * tree-loop-distribution.c (generate_loops_for_partition): Remove
13293 inner loop's exit stmt by making it always exit the loop, otherwise
13294 we would generate an infinite empty loop.
13296 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13298 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
13299 renaming variables in new preheader if it's deleted.
13301 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13303 * tree-loop-distribution.c (struct partition): Remove unused field
13304 loops of the structure.
13305 (partition_alloc, partition_free): Ditto.
13306 (build_rdg_partition_for_vertex): Ditto.
13308 2017-10-09 Jeff Law <law@redhat.com>
13310 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
13311 return type to match prototype and documentation.
13313 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13315 * config/rs6000/rs6000.c (processor_costs): Move to ...
13316 * config/rs6000/rs6000.h: ... here.
13317 (rs6000_cost): Declare.
13319 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
13321 * except.c (setjmp_fn): New global variable.
13322 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
13323 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
13324 if DONT_USE_BUILTIN_SETJMP is defined.
13326 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13328 * target.def (insn_cost): New hook.
13329 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
13330 * doc/tm.texi: Regenerate.
13331 * rtlanal.c (insn_cost): Use the new hook.
13333 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13335 * combine.c (combine_validate_cost): Compute the new insn_cost,
13336 not just pattern_cost.
13337 (try_combine): Adjust comment.
13339 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13341 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
13343 * combine.c (uid_insn_cost): Adjust comment.
13344 (combine_validate_cost): Adjust comment. Use pattern_cost instead
13346 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
13347 * dse.c (find_shift_sequence): Ditto.
13348 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
13349 (bb_valid_for_noce_process_p): Use pattern_cost.
13350 * rtl.h (insn_rtx_cost): Delete.
13351 (pattern_cost): New prototype.
13352 (insn_cost): New prototype.
13353 * rtlanal.c (insn_rtx_cost): Rename to...
13354 (pattern_cost): ... this.
13357 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
13359 * config/i386/i386.md (*jcc_2): Remove insn pattern.
13360 (*jcc<mode>_0_r_i387): Ditto.
13361 (*jccxf_r_i387): Ditto.
13362 (*jcc<mode>_r_i387): Ditto.
13363 (*jccu<mode>_r_i387): Ditto.
13364 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
13365 (*jcc): Rename from *jcc_1.
13367 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13369 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
13370 deferred rescans after the lvx/stvx recombination pre-pass.
13372 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13374 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
13375 memory operation instruction support.
13376 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
13377 (rs6000-ibm-aix[789]*): Likewise.
13378 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
13379 Document new functions.
13381 2017-10-09 Richard Biener <rguenther@suse.de>
13383 PR tree-optimization/82397
13384 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
13385 equality only for semantically equal trees.
13387 2017-10-09 Richard Biener <rguenther@suse.de>
13389 PR tree-optimization/82449
13390 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
13391 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
13392 allow constant addresses.
13393 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
13396 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13398 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
13401 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13404 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
13407 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13410 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
13412 2017-10-09 Jakub Jelinek <jakub@redhat.com>
13415 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
13416 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
13418 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
13420 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
13421 (WI_BINARY_PREDICATE_RESULT): Likewise.
13422 (wi::binary_traits::operator_result): New type.
13423 (wi::binary_traits::predicate_result): Likewise.
13424 (generic_wide_int::operator~, unary generic_wide_int::operator-)
13425 (generic_wide_int::operator==, generic_wide_int::operator!=)
13426 (generic_wide_int::operator&, generic_wide_int::and_not)
13427 (generic_wide_int::operator|, generic_wide_int::or_not)
13428 (generic_wide_int::operator^, generic_wide_int::operator+
13429 (binary generic_wide_int::operator-, generic_wide_int::operator*):
13431 (operator~, unary operator-, operator==, operator!=, operator&)
13432 (operator|, operator^, operator+, binary operator-, operator*): New
13434 * expr.c (get_inner_reference): Use wi::bit_and_not.
13435 * fold-const.c (fold_binary_loc): Likewise.
13436 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
13437 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
13438 (bit_value_binop): Likewise.
13439 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
13440 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
13441 (extract_range_from_binary_expr_1): Likewise.
13442 (masked_increment): Likewise.
13443 (simplify_bit_ops_using_ranges): Likewise.
13445 2017-10-09 Martin Jambor <mjambor@suse.cz>
13448 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
13449 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
13450 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
13452 (hsa_fixup_mov_insn_type): New function.
13453 (hsa_op_with_type::get_in_type): Use it.
13454 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
13455 immediates in an assert.
13456 (hsa_op_with_type::extend_int_to_32bit): New method.
13457 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
13458 types. Convert to dest type if necessary.
13459 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
13460 (reg_for_gimple_ssa): Pass false as min32int to
13461 hsa_type_for_scalar_tree_type.
13462 (gen_hsa_addr): Fixup type when creating addresable temporary.
13463 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
13464 (gen_hsa_unary_operation): Extend operands and convert to dest type if
13465 necessary. Call hsa_fixup_mov_insn_type.
13466 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
13467 extend operands and convert to dest type if necessary.
13468 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
13469 to dest type if necessary.
13470 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
13471 if conversion nt necessary and size matches.
13472 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
13473 to dest type if necessary.
13474 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
13475 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
13477 (gen_hsa_clrsb): Likewise.
13478 (gen_hsa_ffs): Likewise.
13479 (gen_hsa_divmod): Extend operands and convert to dest type if
13481 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
13483 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13485 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
13486 Remove empty default arguments. Use a brace block as output
13488 (conditional return): Ditto.
13490 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
13491 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
13493 (group_ending_nop): Ditto.
13494 (doloop_end): Ditto.
13495 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
13496 (splitters for those): Ditto.
13498 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13500 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
13501 a conditional jump (and the compare for it) so that pc_rtx is the
13503 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
13504 for the deleted and renamed ctr<mode>_internal[234] patterns.
13505 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
13506 Delete second conditional return pattern.
13507 (ctr<mode>_internal2): Delete this second bdnz pattern.
13508 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
13509 (ctr<mode>_internal4): Delete this second bdz pattern.
13511 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13513 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
13514 (always_initialized_rtx_for_ssa_name_p): New predicate.
13515 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
13516 (finish_out_of_ssa): Free new field of SA.
13517 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
13518 * tree-ssa-coalesce.c: Include tree-ssa.h.
13519 (get_parm_default_def_partitions): Remove extern keyword.
13520 (get_undefined_value_partitions): New function.
13521 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
13522 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
13523 uninitialized bits.
13524 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
13526 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13528 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
13530 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13532 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
13533 for targets that preffer 128bit.
13535 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13537 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
13539 2017-10-08 Olivier Hainque <hainque@adacore.com>
13541 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
13542 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
13544 (thumb_set_return_address): Likewise.
13546 2017-10-08 Olivier Hainque <hainque@adacore.com>
13548 * common/config/arm/arm-common.c (arm_except_unwind_info):
13549 Handle DWARF2_UNWIND_INFO.
13551 2017-10-07 Michael Collison <michael.collison@arm.com>
13553 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
13556 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
13558 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
13559 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
13560 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
13561 defined, force the creation of a new block for a dispatch label.
13563 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
13565 * invoke.texi (Wsuggest-attribute=cold): Document.
13566 * common.opt (Wsuggest-attribute=cold): New
13567 * ipa-pure-const.c (warn_function_cold): New function.
13568 * predict.c (compute_function_frequency): Use it.
13569 * predict.h (warn_function_cold): Declare.
13571 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
13573 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
13576 2017-10-06 Martin Liska <mliska@suse.cz>
13578 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
13579 keyword for member functions.
13580 (struct sanopt_tree_couple): New struct.
13581 (struct sanopt_tree_couple_hash): New function.
13582 (struct sanopt_ctx): Add new hash_map.
13583 (has_dominating_ubsan_ptr_check): New function.
13584 (record_ubsan_ptr_check_stmt): Likewise.
13585 (maybe_optimize_ubsan_ptr_ifn): Likewise.
13586 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
13587 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
13589 2017-10-06 Sudakshina Das <sudi.das@arm.com>
13592 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
13593 aarch64_simd_valid_immediate on CONST_VECTORs.
13594 (aarch64_reg_or_bic_imm): Likewise.
13596 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
13598 PR rtl-optimization/82396
13599 * haifa-sched.c (ready_sort_real): Disable qsort checking.
13601 2017-10-06 Sebastian Pop <sebpop@gmail.com>
13603 * graphite-dependences.c (scop_get_reads): Move code to...
13604 (scop_get_must_writes): Move code to...
13605 (scop_get_may_writes): Move code to...
13606 (scop_get_reads_and_writes): ... here.
13607 (scop_get_dependences): Call scop_get_reads_and_writes.
13609 2017-10-06 Jakub Jelinek <jakub@redhat.com>
13611 PR tree-optimization/82434
13612 * fold-const.h (can_native_encode_type_p,
13613 can_native_encode_string_p): Remove.
13614 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
13615 don't encode anything, just return what would be otherwise returned.
13616 (native_encode_fixed, native_encode_complex, native_encode_vector):
13618 (native_encode_string): Likewise. Inline by hand
13619 can_native_encode_string_p.
13620 (can_native_encode_type_p): Remove.
13621 (can_native_encode_string_p): Remove.
13622 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
13623 STRING_CSTs using can_native_encode_string_p, test all
13624 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
13625 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
13626 argument from native_encode_expr.
13627 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
13628 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
13631 2017-10-06 Richard Biener <rguenther@suse.de>
13633 PR tree-optimization/82397
13634 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
13635 operand_equal_p but rely on data_ref_compare_tree for detecting
13637 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
13638 to match up with dr_group_sort_cmp.
13640 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13643 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
13645 * config/s390/s390-builtin-types.def: Regenerate.
13647 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13650 * config/s390/s390-builtin-types.def: Regenerate.
13651 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
13652 Change flag from B_VXE to B_VX.
13653 (s390_vec_min_dbl): Remove B_VXE flag.
13655 2017-10-06 Richard Biener <rguenther@suse.de>
13657 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
13658 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
13659 translate_isl_ast_to_gimple::is_valid_rename,
13660 translate_isl_ast_to_gimple::get_rename,
13661 translate_isl_ast_to_gimple::get_def_bb_for_const,
13662 translate_isl_ast_to_gimple::get_new_name,
13663 translate_isl_ast_to_gimple::collect_all_ssa_names,
13664 translate_isl_ast_to_gimple::copy_loop_phi_args,
13665 translate_isl_ast_to_gimple::collect_all_ssa_names,
13666 translate_isl_ast_to_gimple::copy_loop_phi_args,
13667 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
13668 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
13669 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
13670 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
13671 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
13672 translate_isl_ast_to_gimple::copy_cond_phi_args,
13673 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
13674 translate_isl_ast_to_gimple::edge_for_new_close_phis,
13675 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
13676 translate_isl_ast_to_gimple::rename_uses,
13677 translate_isl_ast_to_gimple::rename_all_uses): Remove.
13678 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
13679 (set_rename_for_each_def): Likewise.
13680 (graphite_copy_stmts_from_block): Handle debug stmt resetting
13681 here. Handle rewriting SCEV analyzable uses here.
13682 (copy_bb_and_scalar_dependences): Generate code for PHI
13684 (graphite_regenerate_ast_isl): Adjust.
13685 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
13686 (add_write, add_read): New functions.
13687 (build_cross_bb_scalars_def): Use it and simplify.
13688 (build_cross_bb_scalars_use): Likewise.
13689 (graphite_find_cross_bb_scalar_vars): Inline into...
13690 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
13691 simulating out-of-SSA. Compute liveout and add dependencies.
13692 (build_scops): Force an empty entry block.
13693 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
13695 (sese_build_liveouts): Declare.
13696 (sese_trivially_empty_bb_p): Likewise.
13697 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
13698 compute liveout and debug_liveout.
13699 (sese_bad_liveouts_use): Remove.
13700 (sese_reset_debug_liveouts_bb): Likewise.
13701 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
13702 (sese_build_liveouts): Build liveout and debug_liveout and store
13704 (new_sese_info): Adjust.
13705 (free_sese_info): Likewise.
13706 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
13707 do not build liveout here.
13708 (move_sese_in_condition): Adjust region entry.
13709 (scev_analyzable_p): Match up with chrec_apply requirements.
13710 (sese_trivially_empty_bb_p): New.
13711 * tree-into-ssa.c (get_reaching_def): Properly support generating
13712 default-defs for incremental rewrite of anonymous names.
13714 2017-10-06 Richard Biener <rguenther@suse.de>
13716 * graphite-sese-to-poly.c (extract_affine): For casts increasing
13717 precision do not perform modulo reduction.
13719 2017-10-06 Richard Biener <rguenther@suse.de>
13721 PR tree-optimization/82436
13722 * tree-vect-slp.c (vect_supported_load_permutation_p): More
13723 conservatively choose the vectorization factor when checking
13724 whether we can perform the required load permutation.
13725 (vect_transform_slp_perm_load): Assert when we may not fail.
13727 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
13729 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
13730 message for incompatible -msdata=* and -mcall-* options.
13732 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13734 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
13735 rate for post-reload scheduling.
13737 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13739 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
13741 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13743 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
13744 to improve monte carlo in scimark.
13746 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13748 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
13749 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
13750 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
13751 pentium4_cost, nocona_cost): Set reassociation width to 1.
13752 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
13753 width to 2 for fp operations and 1 otherwise.
13754 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
13756 (atom_cost): Set reassociation width to 2.
13757 (slm_cost, generic_cost): Set fp reassociation width
13758 to 2 and 1 otherwise.
13759 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
13760 (core_cost): Set fp reassociation width to 4 and vector to 2.
13761 (ix86_reassociation_width): Rewrite using cost table; special case
13762 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
13763 and TARGET_AVX128_OPTIMAL.
13764 * config/i386/i386.h (processor_costs): Add
13765 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
13766 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
13767 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
13768 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
13769 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
13770 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
13772 2017-10-05 Nathan Sidwell <nathan@acm.org>
13774 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
13776 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13778 * config/arm/arm.c (arm_test_fpu_data): New.
13779 (arm_run_selftests): Call arm_test_fpu_data.
13781 2017-10-04 Nathan Sidwell <nathan@acm.org>
13783 * toplev.c (toplev::main): Remove excess parens on pretty_printer
13785 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
13787 2017-10-04 Sudakshina Das <sudi.das@arm.com>
13789 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13790 check type for aarch64_simd_valid_immediate.
13791 (aarch64_output_simd_mov_immediate): Update prototype.
13792 (aarch64_simd_valid_immediate): Update prototype.
13793 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13794 support for ORR-immediate.
13795 (and<mode>3): modified pattern to add support for BIC-immediate.
13796 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13797 now checks for valid immediate for BIC and ORR based on new enum
13799 (aarch64_output_simd_mov_immediate): Function now used to output
13800 BIC/ORR imm as well based on new enum argument.
13801 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13803 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13804 (aarch64_reg_or_bic_imm): Likewise.
13806 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13808 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13809 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13811 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13815 PR rtl-optimization/82396
13816 * haifa-sched.c (autopref_multipass_init): Simplify
13818 (autopref_rank_data): Simplify sort order.
13819 * sched-int.h (autopref_multipass_data_): Remove
13820 multi_mem_insn_p, min_offset and max_offset.
13822 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13824 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13826 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13828 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13829 vect_doubleint_cvt.
13831 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13833 * doc/sourcebuild.texi: Document vect_long_mult.
13835 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13837 PR tree-optimization/82413
13838 * fold-const.c (build_range_check): Use widest_int when comparing
13839 the maximum ETYPE value with HIGH.
13841 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13843 PR rtl-optimization/82396
13844 * haifa-sched.c (autopref_multipass_init): Simplify
13846 (autopref_rank_data): Simplify sort order.
13847 * sched-int.h (autopref_multipass_data_): Remove
13848 multi_mem_insn_p, min_offset and max_offset.
13850 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13852 PR tree-optimization/82381
13853 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13854 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13857 PR tree-optimization/82374
13858 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13859 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13860 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13861 current_function_decl to the new decl.
13863 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13865 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13866 helper macro for IEEE float128 hardware built-in functions.
13867 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13869 (TRUNCF128_ODD): Likewise.
13870 (ADDF128_ODD): Likewise.
13871 (SUBF128_ODD): Likewise.
13872 (MULF128_ODD): Likewise.
13873 (DIVF128_ODD): Likewise.
13874 (FMAF128_ODD): Likewise.
13875 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13876 UNSPEC_TRUNC_ROUND_TO_ODD.
13877 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13878 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13879 floating point round to odd instructions.
13880 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13881 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13882 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13883 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13884 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13885 (trunc<mode>sf2_hw): Change the truncate with round to odd
13886 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13887 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13888 to odd hardware instructions.
13889 (sub<mode>3_odd): Likewise.
13890 (mul<mode>3_odd): Likewise.
13891 (div<mode>3_odd): Likewise.
13892 (sqrt<mode>2_odd): Likewise.
13893 (fma<mode>4_odd): Likewise.
13894 (fms<mode>4_odd): Likewise.
13895 (nfma<mode>4_odd): Likewise.
13896 (nfms<mode>4_odd): Likewise.
13897 (trunc<mode>df2_odd): Change the truncate with round to odd
13898 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13900 * doc/extend.texi (PowerPC built-in functions): Update documentation
13901 for existing IEEE float128-bit built-in functions. Add built-in
13902 functions that generate the IEEE 128-bit floating point round to
13905 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13907 PR rtl-optimization/77729
13908 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13909 to (X&(C1&~C2))|C2 transformations.
13911 2017-10-03 Martin Jambor <mjambor@suse.cz>
13913 PR tree-optimization/82363
13914 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13915 mismatch, mark lacc written regardless of racc.
13917 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13919 PR tree-optimization/82381
13920 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13921 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13922 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13923 fallthrough into reassoc_stmt_dominates_stmt_p.
13926 * combine.c (combine_instructions): Don't combine in unreachable
13929 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13932 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13933 function to not use the have_cpu variable. Do not set cpu_index,
13934 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13935 or the default cpu.
13936 (rs6000_valid_attribute_p): Remove duplicate initializations of
13937 old_optimize and func_optimize.
13938 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13939 (rs6000_activate_target_options): Make global.
13940 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13943 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13945 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13946 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13947 Set *poffset to 0 and *psize and *pmax_size to -1 if
13948 *poffset + *psize overflows in HOST_WIDE_INT.
13950 PR tree-optimization/82387
13951 PR tree-optimization/82388
13952 PR tree-optimization/82389
13953 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13954 instead of live_bytes non-NULL.
13956 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13959 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13962 2017-10-02 Richard Biener <rguenther@suse.de>
13964 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13965 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13966 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13968 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13970 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13971 requested precision matches the type's.
13972 * calls.c (alloc_max_size): Calculate the new candidate size as
13973 a widest_int and use wi::to_widest when comparing it with the
13974 current candidate size.
13975 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13976 zero rather than integer_zero_node.
13977 * match.pd: Check for a no-op conversion before using wi::add
13978 rather than after. Use tree_to_uhwi when summing small shift
13979 counts into an unsigned int.
13981 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13982 Alan Hayward <alan.hayward@arm.com>
13983 David Sherwood <david.sherwood@arm.com>
13986 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13987 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13988 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13989 POINTER_AND_FP_REGS.
13991 2017-10-02 Richard Biener <rguenther@suse.de>
13993 PR tree-optimization/82355
13994 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13995 a mapping for the enclosing loop but avoid generating one for
13996 the loop tree root.
13997 (copy_bb_and_scalar_dependences): Remove premature codegen
13998 error on PHIs in blocks duplicated into multiple places.
13999 * graphite-scop-detection.c
14000 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
14001 in the region use it as loop and nest to analyze the DR in.
14002 (try_generate_gimple_bb): Likewise.
14003 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
14004 (add_loop_constraints): For blocks in a loop not in the region
14005 create a dimension with a single iteration.
14006 * sese.h (gbb_loop_at_index): Remove assert.
14008 2017-10-01 Kevin Buettner <kevinb@redhat.com>
14010 * omp-expand.c (adjust_context_scope): New function.
14011 (expand_parallel_call): Call adjust_context_scope.
14013 2017-10-01 Jeff Law <law@redhat.com>
14015 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
14016 dom_opt_dom_walker class with direct access to private members.
14017 Add comments. Call test_for_singularity.
14018 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
14019 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
14020 m_dummy_cond anymore.
14021 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
14023 (pass_dominator:execute): Build the dummy_cond here and pass it
14024 to the dom_opt_dom_walker ctor.
14025 (test_for_singularity): New function.
14027 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
14028 Maya Rashish <coypu@sdf.org>
14030 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
14031 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
14032 (alpha*-*-netbsd*) Use nbsd_tm_file.
14033 (arm*-*-netbsdelf*) Likewise.
14034 (i[34567]86-*-netbsdelf*) Likewise.
14035 (x86_64-*-netbsd*) Likewise.
14036 (mips*-*-netbsd*) Likewise.
14037 (powerpc-*-netbsd*) Likewise.
14038 (sh*-*-netbsd*) Likewise.
14039 (sparc-*-netbsdelf*) Likewise.
14040 (sparc64-*-netbsd*) Likewise.
14041 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
14043 (vax-*-netbsdelf*) Likewise.
14044 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
14045 (UINT_FAST8_TYPE) Likewise.
14046 (INT_FAST16_TYPE) Check CHAR_FAST16.
14047 (UINT_FAST16_TYPE) Likewise.
14049 2017-09-30 Jakub Jelinek <jakub@redhat.com>
14052 * config/i386/i386.md
14053 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
14054 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
14055 *divmodsi4_zext_2): New define_insn_and_split.
14056 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
14057 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
14058 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
14059 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
14060 New define_insn_and_split.
14061 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
14062 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
14063 operands[1] having DImode when mode is SImode.
14065 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
14066 always SImode for DIV and MOD in REG_EQUAL notes.
14068 2017-09-29 Yury Gribov <tetra2005@gmail.com>
14070 PR middle-end/82319
14071 * match.pd: Fix handling of NaNs in pattern.
14073 2017-09-29 Jeff Law <law@redhat.com>
14075 * sbitmap.c (bitmap_bit_in_range_p): New function.
14076 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
14077 * tree-ssa-dse.c (live_bytes_read): New function.
14078 (dse_classify_store): Ignore reads of dead bytes.
14080 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
14081 typos and whitespace errors.
14082 * config/i386/predicates.md (address_no_seg_operand): Likewise.
14083 * config/s390/s390.c (s390_emit_prologue): Likewise.
14085 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
14088 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
14089 with a symbol for LRA.
14091 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
14093 PR rtl-optimization/82338
14094 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
14096 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
14098 * genmodes.c (calc_wider_mode): Suppress qsort macro.
14099 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
14100 (qsort_chk): Declare.
14101 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
14102 (qsort_chk): New function.
14104 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14106 PR tree-optimization/82337
14107 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
14108 phi definition if the PHI result appears in an abnormal PHI.
14109 (find_basis_for_base_expr): Don't record a basis if the LHS of the
14110 basis appears in an abnormal PHI.
14112 2017-09-29 Richard Biener <rguenther@suse.de>
14114 * graphite-isl-ast-to-gimple.c
14115 (translate_isl_ast_to_gimple::set_codegen_error): New function.
14116 (binary_op_to_tree): Use it.
14117 (get_rename_from_scev): Likewise.
14118 (copy_loop_phi_nodes): Likewise.
14119 (copy_bb_and_scalar_dependences): Likewise.
14120 (translate_pending_phi_nodes): Likewise.
14122 2017-09-29 Jakub Jelinek <jakub@redhat.com>
14125 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
14126 for movabsq $(i32 << shift), r64.
14128 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
14130 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
14131 index when encoding %esp as %rsp to avoid 0x67 prefix.
14133 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
14135 * config/i386/i386.md (*movsf_internal, *movdf_internal):
14136 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
14138 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
14140 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
14141 Extensions with more than 16 double VFP registers.
14142 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
14143 to_clear_mask and all code related to it. Replace the remaining
14144 entry by a sbitmap and adapt code accordingly.
14146 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
14148 * brig-builtins.def: Change pure attributes to const.
14150 2017-09-28 Joseph Myers <joseph@codesourcery.com>
14152 * config.gcc (default_gnu_indirect_function): Default to yes for
14153 sparc*-*-linux* with glibc.
14155 2017-09-28 Joseph Myers <joseph@codesourcery.com>
14157 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
14158 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
14159 when creating .init_array and .fini_array sections with priority
14162 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
14165 * config/aarch64/aarch64.c
14166 (aarch64_builtin_support_vector_misalignment): Always return false
14167 when misalignment is unknown.
14169 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
14171 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
14172 this function to return false if the definition used by the swap
14173 instruction is artificial, or if the memory address from which the
14174 constant value is loaded is not represented by a base address held
14175 in a register or if the base address register is a frame or stack
14176 pointer. Additionally, return false if the base address of the
14177 loaded constant is a SYMBOL_REF but is not considered to be a
14179 (replace_swapped_load_constant): New function.
14180 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
14181 loaded constant vector with a load of a swapped constant vector.
14183 2017-09-27 Carl Love <cel@us.ibm.com>
14185 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
14186 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
14187 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
14190 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
14192 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
14193 first, always call autopref_rank_data otherwise.
14195 2017-09-27 Richard Biener <rguenther@suse.de>
14197 * graphite-scop-detection.c (find_scop_parameters): Move
14198 loop bound handling ...
14199 (gather_bbs::before_dom_children): ... here, avoiding the need
14200 to build scop_info->loop_nest.
14201 (record_loop_in_sese): Remove.
14202 * sese.h (sese_info_t::loop_nest): Remove.
14203 * sese.c (new_sese_info): Do not allocate loop_nest.
14204 (free_sese_info): Do not free loop_nest.
14206 2017-09-27 Jakub Jelinek <jakub@redhat.com>
14209 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
14210 lhs from calls if the lhs has addressable type.
14212 2017-09-27 Richard Biener <rguenther@suse.de>
14214 * graphite.h (scop::max_alias_set): New member.
14215 * graphite-scop-detection.c: Remove references to non-existing
14216 --param in comments.
14217 (build_alias_sets): Record the maximum alias set used for drs.
14218 (build_scops): Support zero as unlimited for
14219 --param graphite-max-arrays-per-scop.
14220 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
14221 and inline into ...
14222 (build_poly_sr_1): ... here. Compute alias set based on the
14223 maximum alias set used for drs rather than
14224 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
14226 2017-09-27 Richard Biener <rguenther@suse.de>
14228 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
14229 --param loop-block-tile-size=0 to disable tiling.
14231 2017-09-27 Richard Biener <rguenther@suse.de>
14233 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
14234 (graphite-max-nb-scop-params): Document special value zero.
14235 * domwalk.h (dom_walker::STOP): New symbolical constant.
14236 (dom_walker::dom_walker): Add optional parameter for bb to
14238 (dom_walker::~dom_walker): Declare.
14239 (dom_walker::before_dom_children): Document STOP return value.
14240 (dom_walker::m_user_bb_to_rpo): New member.
14241 (dom_walker::m_bb_to_rpo): Likewise.
14242 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
14243 mapping here if not provided by the user.
14244 (dom_walker::~dom_walker): Free bb to RPO mapping if not
14245 provided by the user.
14246 (dom_walker::STOP): Define.
14247 (dom_walker::walk): Do not compute bb to RPO mapping here.
14248 Support STOP return value from before_dom_children to stop
14250 * graphite-optimize-isl.c (optimize_isl): If the schedule
14251 is the same still generate code if -fgraphite-identity
14252 or -floop-parallelize-all are given.
14253 * graphite-scop-detection.c: Include cfganal.h.
14254 (gather_bbs::gather_bbs): Get and pass through bb to RPO
14256 (gather_bbs::before_dom_children): Return STOP for BBs
14258 (build_scops): Compute bb to RPO mapping and pass it to
14259 the domwalk. Treat --param graphite-max-nb-scop-params=0
14260 as not limiting the number of params.
14261 * graphite.c (graphite_initialize): Remove limit on the
14262 number of basic-blocks in a function.
14263 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
14264 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
14265 default value of 10.
14267 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14269 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
14270 Adjust code to eliminate needing to do the shift right 32-bits
14271 operation after XSCVDPSPN.
14273 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14275 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
14276 ((X / Y) != 0 -> X >= Y): Likewise.
14278 2017-09-26 Carl Love <cel@us.ibm.com>
14280 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
14281 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
14282 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
14283 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
14284 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
14285 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
14286 definitions and overloading.
14287 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
14288 statement for P9V_BUILTIN_XST_LEN_R.
14289 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
14290 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
14291 define_expand and define_insn for the instructions and builtins.
14292 * doc/extend.texi: Update the built-in documentation file for the new
14293 built-in functions.
14294 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
14295 define_insn for the instructions
14297 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
14300 * gcc/config/netbsd-protos.h: New file.
14301 * gcc/config/netbsd.c: New file.
14302 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
14303 * gcc/config/t-netbsd: New file.
14304 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
14305 (tmake_file) Add t-netbsd.
14306 (extra_objs) Add netbsd.o.
14308 2017-09-26 Janus Weil <janus@gcc.gnu.org>
14312 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
14314 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14316 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
14317 sign extension from a vector register to a GPR by doing a 32-bit
14318 direct move and then an EXTSW.
14319 (extendsi<mode>2 splitter): Likewise.
14320 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
14321 right or vector extract after doing XSCVDPSPN. Use
14322 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
14324 (movdi_from_sf_zero_ext): Likewise.
14325 (reload_gpr_from_vsxsf): Likewise.
14326 (p8_mfvsrd_4_disf): Delete, no longer used.
14327 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
14328 then needing to move the SFmode to a GPR to use the XSCVDPSP
14329 instruction instead of FRSP and XSCVDPSPN.
14330 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
14331 it is adjacent to the other XSCVSPDP insns.
14332 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
14333 SFmode to be in traditional Altivec registers.
14334 (vsx_xscvdpspn): Eliminate useless alternative constraint.
14335 (vsx_xscvspdpn): Likewise.
14336 (vsx_xscvspdpn_scalar): Likewise.
14338 2017-09-26 Martin Jambor <mjambor@suse.cz>
14340 * tree-sra.c (compare_access_positions): Put integral types first,
14341 stabilize sorting of integral types, remove conditions putting
14342 non-full-precision integers last.
14343 (sort_and_splice_var_accesses): Disable scalarization if a
14344 non-integert would be represented by a non-full-precision integer.
14346 2017-09-26 Joseph Myers <joseph@codesourcery.com>
14348 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
14349 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
14350 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
14351 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
14352 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
14353 conditionals inside the function instead of around it. Call
14354 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
14355 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
14357 2017-09-26 Richard Biener <rguenther@suse.de>
14359 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
14361 (scop_detection::build_scop_breadth): ... this. Removed.
14362 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
14363 (scop_detection::harmful_stmt_in_bb): Likewise.
14364 (scop_detection::graphite_can_represent_stmt): Likewise.
14365 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
14366 (scop_detection::can_represent_loop): Remove recursion, fold in ...
14367 (scop_detection::can_represent_loop_1): ... this. Removed.
14368 (scop_detection::harmful_loop_in_region): Simplify after inlining
14369 the above and remove more quadraticness.
14370 (build_scops): Adjust.
14371 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
14374 2017-09-26 Jakub Jelinek <jakub@redhat.com>
14377 * config/i386/i386.c (ix86_print_operand_address_as): Only test
14378 REGNO (base) == SP_REG if base is a REG.
14380 PR middle-end/35691
14381 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
14382 if it is different SSA_NAME.
14383 (optimize_range_tests_cmp_bitwise): New function.
14384 (optimize_range_tests): Call it.
14386 2017-09-26 Richard Biener <rguenther@suse.de>
14388 PR tree-optimization/82321
14389 * graphite.c (canonicalize_loop_closed_ssa): Properly check
14390 for the def being inside the loop.
14392 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14394 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
14396 * config/s390/s390-builtins.def: Fix constraint on op4.
14398 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14400 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
14401 independent expanders.
14402 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
14403 ("vec_ordered", "vec_unordered"): New expanders.
14405 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14407 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
14410 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14412 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
14413 vec_unpacks_lo_v16qi.
14414 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
14416 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14418 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
14419 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
14420 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
14422 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14424 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
14426 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
14428 ("*vec_slb<mode>"): New insn pattern.
14429 ("vec_shr_<mode>"): New expander.
14430 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
14431 and force the shift count operand to V16QImode.
14432 ("vec_srb<mode>"): Set shift count mode to V16QI.
14434 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14436 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
14437 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
14438 ("vec_widen_smult_hi_<mode>"): New expander definitions.
14440 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
14443 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
14445 2017-09-26 Richard Biener <rguenther@suse.de>
14447 PR tree-optimization/82320
14448 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
14451 2017-09-25 Jeff Law <law@redhat.com>
14453 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
14454 prototype for new argument.
14455 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
14456 mostly extracted from rs6000_emit_allocate_stack.
14457 (rs6000_emit_probe_stack_range_stack_clash): New function.
14458 (rs6000_emit_allocate_stack): Call
14459 rs6000_emit_probe_stack_range_stack_clash as needed.
14460 (rs6000_emit_probe_stack_range): Add additional argument
14461 to call to gen_probe_stack_range{si,di}.
14462 (output_probe_stack_range): New.
14463 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
14464 (output_probe_stack_range_stack_clash): New.
14465 (rs6000_emit_prologue): Emit notes into dump file as requested.
14466 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
14467 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
14468 Add additional operand and pass it to output_probe_stack_range.
14470 2017-09-25 Bin Cheng <bin.cheng@arm.com>
14472 PR tree-optimization/82163
14473 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
14474 (checking_verify_loop_closed_ssa): New parameter.
14475 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
14476 (check_loop_closed_ssa_stmt): Delete.
14477 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
14478 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
14479 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
14482 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
14484 * brig-builtins.def: Treat HSAIL barrier builtins as
14485 setjmp/longjump style functions.
14487 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14489 * target.def (constant_alignment): New hook.
14490 * defaults.h (CONSTANT_ALIGNMENT): Delete.
14491 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
14492 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
14493 * doc/tm.texi: Regenerate.
14494 * targhooks.h (default_constant_alignment): Declare.
14495 (constant_alignment_word_strings): Likewise.
14496 * targhooks.c (default_constant_alignment): New function.
14497 (constant_alignment_word_strings): Likewise.
14498 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
14499 instead of CONSTANT_ALIGNMENT.
14500 * varasm.c (align_variable, get_variable_align, build_constant_desc)
14501 (force_const_mem): Likewise.
14502 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
14503 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
14504 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
14505 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14506 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
14508 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
14509 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14510 constant_alignment_word_strings.
14511 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
14512 (CONSTANT_ALIGNMENT): Likewise.
14513 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14514 (arm_constant_alignment): New function.
14515 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
14516 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14517 constant_alignment_word_strings.
14518 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
14519 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14520 constant_alignment_word_strings.
14521 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
14522 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14523 (cris_constant_alignment): New function.
14524 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
14525 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14526 (epiphany_constant_alignment): New function.
14527 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
14528 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14529 constant_alignment_word_strings.
14530 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
14531 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14532 constant_alignment_word_strings.
14533 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
14534 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14535 constant_alignment_word_strings.
14536 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
14537 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
14538 * config/i386/i386.c (ix86_constant_alignment): Make static.
14539 Use the same interface as the target hook.
14540 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14541 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
14542 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14543 constant_alignment_word_strings.
14544 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
14545 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
14546 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14547 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
14548 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14549 constant_alignment_word_strings.
14550 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
14551 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14552 constant_alignment_word_strings.
14553 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
14554 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14555 constant_alignment_word_strings.
14556 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
14557 * config/microblaze/microblaze.c (microblaze_constant_alignment):
14559 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14560 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
14561 * config/mips/mips.c (mips_constant_alignment): New function.
14562 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14563 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
14564 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
14565 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14566 (mmix_constant_alignment): Make static. Use the same interface
14567 as the target hook.
14568 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
14569 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14570 constant_alignment_word_strings.
14571 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
14572 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14573 constant_alignment_word_strings.
14574 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
14575 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14576 constant_alignment_word_strings.
14577 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
14578 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14579 (rs6000_constant_alignment): New function.
14580 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
14581 * config/riscv/riscv.c (riscv_constant_alignment): New function.
14582 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14583 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
14584 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14585 (rs6000_constant_alignment): New function.
14586 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
14587 * config/s390/s390.c (s390_constant_alignment): New function.
14588 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14589 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
14590 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14591 constant_alignment_word_strings.
14592 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
14593 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14594 (sparc_constant_alignment): New function.
14595 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
14596 * config/spu/spu.c (spu_constant_alignment): New function.
14597 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14598 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
14599 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14600 constant_alignment_word_strings.
14601 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
14602 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14603 constant_alignment_word_strings.
14604 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
14605 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14606 constant_alignment_word_strings.
14607 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
14608 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14609 (visium_constant_alignment): New function.
14610 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
14611 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14612 (xtensa_constant_alignment): New function.
14613 * system.h (CONSTANT_ALIGNMENT): Poison.
14615 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
14617 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14618 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
14619 (rs6000_builtin_valid_without_lhs): New helper function.
14620 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14621 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
14623 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14625 * target.h (vec_perm_indices): Use unsigned short rather than
14627 (auto_vec_perm_indices): Likewise.
14628 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
14629 Use unsigned int rather than unsigned char.
14630 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
14632 2017-09-25 Richard Biener <rguenther@suse.de>
14634 * cfgloop.h (sort_sibling_loops): Declare.
14635 * cfgloop.c (sort_sibling_loops_cmp): New helper.
14636 (sort_sibling_loops): New function sorting the sibling loop list
14638 * graphite.c (graphite_transform_loops): Sort sibling loops.
14640 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
14642 * target.def (vec_perm_const_ok): Change sel parameter to
14644 * optabs-query.c (can_vec_perm_p): Update accordingly.
14645 * doc/tm.texi: Regenerate.
14646 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
14647 auto_vec_perm_indices and remove separate nelt field.
14648 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
14649 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
14650 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
14651 (aarch64_expand_vec_perm_const): Update accordingly.
14652 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
14653 to vec_perm_indices.
14654 * config/arm/arm.c (expand_vec_perm_d): Change perm to
14655 auto_vec_perm_indices and remove separate nelt field.
14656 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
14657 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
14658 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
14660 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
14661 to vec_perm_indices.
14662 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
14663 sel to vec_perm_indices.
14664 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
14665 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
14666 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
14668 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
14671 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
14674 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
14675 on the FUNCTION_DECL function context if it has a DIE that is a
14678 2017-09-25 Richard Biener <rguenther@suse.de>
14680 PR tree-optimization/82285
14681 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
14684 2017-09-25 Tom de Vries <tom@codesourcery.com>
14688 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
14691 2017-09-25 Richard Biener <rguenther@suse.de>
14693 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
14694 ISL errors other than isl_error_quota happen. Dump if the
14695 schedule is the same.
14696 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
14697 errors instead of aborting inside ISL.
14699 2017-09-25 Iain Sandoe <iain@codesourcery.com>
14702 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
14703 of libgcc_eh for m64.
14704 * config/i386/darwin64.h: Likewise.
14706 2017-09-25 Richard Biener <rguenther@suse.de>
14708 PR middle-end/82144
14709 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
14710 attribute for incomplete types nor twice for complete ones.
14712 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
14715 * config/i386/i386.c (ix86_print_operand_address_as): Encode
14716 %esp as %rsp to avoid 0x67 prefix if there is no index or base
14719 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
14722 * config/i386/i386.opt (mprefer-avx256): Use
14723 ix86_target_flags variable.
14724 * config/i386/i386.c (ix86_target_string): Move
14725 -mprefer-avx256 to flag2_opts.
14727 2017-09-22 Jakub Jelinek <jakub@redhat.com>
14729 PR middle-end/35691
14730 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
14731 and x != -1 | y != -1 into (x & y) != -1.
14733 2017-09-22 Steve Ellcey <sellcey@cavium.com>
14735 * config.gcc: Add new case statement to set
14736 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
14737 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
14738 s390x-*-linux* case statements. Added aarch64 to the list of
14739 supported architectures.
14741 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14743 PR tree-optimization/82289
14744 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
14745 STMT_VINFO_RELEVANT_P.
14747 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14748 Alan Hayward <alan.hayward@arm.com>
14749 David Sherwood <david.sherwood@arm.com>
14751 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
14752 for VR_RANGE only; don't allow VR_ANTI_RANGE.
14753 (extract_range_from_binary_expr_1): Don't call
14754 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
14756 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14757 Alan Hayward <alan.hayward@arm.com>
14758 David Sherwood <david.sherwood@arm.com>
14760 * target.def (preferred_vector_alignment): New hook.
14761 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
14763 * doc/tm.texi: Regenerate.
14764 * targhooks.h (default_preferred_vector_alignment): Declare.
14765 * targhooks.c (default_preferred_vector_alignment): New function.
14766 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
14768 (DR_TARGET_ALIGNMENT): New macro.
14769 (aligned_access_p): Update commentary.
14770 (vect_known_alignment_in_bytes): New function.
14771 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
14773 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
14774 Calculate the misalignment based on the target alignment rather than
14776 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
14777 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
14778 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
14779 the target alignment, rather than masking the element misalignment
14780 with the number of elements in a vector. Also use the target
14781 alignment when calculating the maximum number of peels.
14782 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
14783 instead of TYPE_ALIGN_UNIT.
14784 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
14785 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
14786 (vect_create_addr_base_for_vector_ref): Update call accordingly.
14787 (vect_create_data_ref_ptr): Likewise.
14788 (vect_setup_realignment): Realign by ANDing with
14789 -DR_TARGET_MISALIGNMENT.
14790 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14791 the number of peels based on DR_TARGET_ALIGNMENT.
14792 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14793 with the guaranteed alignment boundary when deciding whether
14795 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14796 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14797 (ensure_base_align): Remove stmt_info parameter. Get the
14798 target base alignment from DR_TARGET_ALIGNMENT.
14799 (vectorizable_store): Update call accordingly. Interpret
14800 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14802 (vectorizable_load): Likewise.
14804 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14805 Alan Hayward <alan.hayward@arm.com>
14806 David Sherwood <david.sherwood@arm.com>
14808 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14809 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14810 (vect_enhance_data_refs_alignment): Likewise.
14812 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14814 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14815 error. Only quit immediately if parsing is complete.
14816 (BEGIN): Initialize fatal_err and parse_done.
14817 (begin fpu, end fpu): Check number of arguments.
14818 (begin arch, end arch): Likewise.
14819 (begin cpu, end cpu): Likewise.
14820 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14821 (optalias): Likewise.
14823 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14825 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14826 * config/arm/arm-isa.h: Delete. Move definitions to ...
14827 * arm-cpus.in: ... here. Use new feature and fgroup values.
14828 * config/arm/arm.c (arm_option_override): Use lower case for feature
14830 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14831 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14832 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14834 (print_isa_bits_for): New function.
14835 (gen_isa): New function.
14836 (gen_comm_data): Use print_isa_bits_for.
14837 (define feature): New keyword.
14838 (define fgroup): New keyword.
14839 * config/arm/t-arm (TM_H): Remove.
14840 (GTM_H): Add arm-isa.h.
14841 (arm-isa.h): Add rule to generate file.
14842 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14843 case for feature bit names.
14845 2017-09-22 Richard Biener <rguenther@suse.de>
14847 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14849 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14850 print of no dependency loops ...
14851 * graphite.c (graphite_transform_loops): ... here.
14852 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14854 (same_close_phi_node, remove_duplicate_close_phi,
14855 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14856 (canonicalize_loop_closed_ssa): ... here and simplify.
14857 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14858 (optimize_isl): Use dump_printf_loc to tell when we stopped
14859 optimizing because of an ISL timeout.
14861 2017-09-22 Richard Biener <rguenther@suse.de>
14863 PR tree-optimization/82291
14864 * tree-if-conv.c (predicate_mem_writes): Make sure to
14865 remove writes in blocks predicated with false.
14867 2017-09-22 Richard Biener <rguenther@suse.de>
14869 * sese.c: Include cfganal.h.
14870 (if_region_set_false_region): Remove.
14871 (create_if_region_on_edge): Likewise.
14872 (move_sese_in_condition): Re-implement without destroying
14875 2017-09-22 Richard Biener <rguenther@suse.de>
14877 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14878 Verify both BBs contain loop PHI nodes before dispatching to
14879 copy_loop_phi_args.
14880 (graphite_regenerate_ast_isl): Do not recompute dominators,
14881 do not verify three times. Restructure for clarity.
14882 * graphite-scop-detection.c (same_close_phi_node,
14883 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14884 defined_in_loop_p, canonicalize_loop_closed_ssa,
14885 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14886 checking and SSA rewrite, move to ...
14887 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14888 (graphite_initialize): Do not pass in ctx, do not reset the
14889 SCEV cache, compute only dominators.
14890 (graphite_transform_loops): Allocate ISL ctx after
14891 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14892 Maintain post-dominators only around build_scops.
14893 * sese.c (if_region_set_false_region): Make static. Free
14894 and recompute dominators.
14895 (move_sese_in_condition): Assert we don't get called with
14896 post-dominators computed.
14897 * sese.h (if_region_set_false_region): Remove.
14899 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14901 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14902 mode attribute for TARGET_AVX512VL.
14904 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14906 * config/i386/i386.opt (mprefer-avx256): New option.
14907 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14909 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14910 for TARGET_PREFER_AVX256.
14911 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14913 2017-09-21 Jeff Law <law@redhat.com>
14915 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14916 Fix dump output if the only stack space is for pushed registers.
14918 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14920 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14923 2017-09-21 Martin Sebor <msebor@redhat.com>
14926 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14927 code (in C++) or code that triggers warnings.
14929 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14931 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14933 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14935 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14936 * sched-rgn.c: Ditto.
14937 * sel-sched-ir.c: Ditto.
14939 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14941 * toplev.h (set_random_seed): Adjust return type.
14942 * toplev.c (init_local_tick): Move eager initialization of random_seed
14943 to get_random_seed. Adjust comment.
14944 (init_random_seed): Inline to get_random_seed, delete.
14945 (get_random_seed): Initialize random_seed lazily.
14946 (set_random_seed): Do not return previous value.
14947 (print_switch_value): Do not call get_random_seed.
14949 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14951 * cgraph.c (delete_function_version): New, broken out from...
14952 (cgraph_node::delete_function_version): ...here. Rename to
14953 cgraph_node::delete_function_version_by_decl. Update all uses.
14954 (cgraph_node::remove): Call delete_function_version.
14956 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14959 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14960 VAR_DECLs to which addressable non-volatile parameters are mapped
14961 and for id->retvar after the return value assignment. Clear
14962 id->retval and id->retbnd after inlining.
14964 2017-09-21 Richard Biener <rguenther@suse.de>
14966 PR tree-optimization/82276
14967 PR tree-optimization/82244
14968 * tree-vrp.c (build_assert_expr_for): Set
14969 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14971 (remove_range_assertions): Revert earlier change.
14973 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14976 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14978 2017-09-21 Richard Biener <rguenther@suse.de>
14980 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14981 Restore valid IL after code generation errors.
14982 * graphite.c (graphite_transform_loops): Diagnose code
14983 generation issues as MSG_MISSED_OPTIMIZATION and continue
14984 with processing SCOPs.
14986 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14987 Alan Hayward <alan.hayward@arm.com>
14988 David Sherwood <david.sherwood@arm.com>
14990 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14991 rather than choosing between plus_constant and gen_rtx_<CODE>.
14992 * expr.c (emit_push_insn): Likewise.
14993 (expand_expr_real_2): Likewise.
14995 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14996 Alan Hayward <alan.hayward@arm.com>
14997 David Sherwood <david.sherwood@arm.com>
14999 * loop-unroll.c (split_iv): Call copy_rtx on the step.
15001 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
15002 Alan Hayward <alan.hayward@arm.com>
15003 David Sherwood <david.sherwood@arm.com>
15005 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
15006 calling tree_to_uhwi.
15008 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
15009 Alan Hayward <alan.hayward@arm.com>
15010 David Sherwood <david.sherwood@arm.com>
15012 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
15013 INTEGER_CST rather than a negative test for ADDR_EXPR.
15015 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
15016 Alan Hayward <alan.hayward@arm.com>
15017 David Sherwood <david.sherwood@arm.com>
15019 * tree-vrp.c (extract_range_from_binary_expr_1): Check
15020 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
15022 2017-09-21 Richard Biener <rguenther@suse.de>
15024 PR tree-optimization/71351
15025 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
15026 graphite_create_new_loop_guard): Remove, fold remaining parts
15028 (translate_isl_ast_node_for): ... here and simplify.
15030 2017-09-21 Jakub Jelinek <jakub@redhat.com>
15033 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
15034 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
15035 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
15036 alternative always use QI mode, for -Os imov (=R,R) alternative
15037 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
15040 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15041 Jeff Law <law@redhat.com>
15043 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
15044 (allocate_stack_space): New function, partially extracted from
15045 s390_emit_prologue.
15046 (s390_emit_prologue): Track offset to most recent stack probe.
15047 Code to allocate space moved into allocate_stack_space.
15048 Dump actions when no stack is allocated.
15049 (s390_prologue_plus_offset): New function.
15050 (s390_emit_stack_probe): Likewise.
15052 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
15054 * common.opt (Wa, Wl, Wp, g, gz=): Add
15056 (gno-column-info): Remove.
15057 (gcolumn-info): Drop RejectNegative.
15058 (gno-): New prefix.
15059 (gno-record-gcc-switches): Remove.
15060 (grecord-gcc-switches): Drop RejectNegative.
15061 (gno-split-dwarf): Remove.
15062 (gsplit-dwarf): Drop RejectNegative.
15063 (gno-strict-dwarf): Remove.
15064 (gstrict-dwarf): Drop RejectNegative.
15065 * config/darwin.opt (gfull, gused): Add RejectNegative.
15066 * dwarf2out.c (gen_producer_string): Drop
15067 gno-record-gcc-switches handler.
15068 * optc-gen.awk: Add g to prefixes with negative forms.
15069 * opts-common.c (remapping_prefix_p): New.
15070 (find_opt): Check it.
15071 (generate_canonical_option): Test g prefix.
15072 (option_map): Add -gno- mapping.
15073 (add_misspelling_candidates): Check remapping_prefix_p.
15075 2017-09-20 Jeff Law <law@redhat.com>
15077 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
15078 thinko in stack clash protection support.
15080 * explow.c (compute_stack_clash_protection_loop_data): Use
15081 CONST_INT_P instead of explicit test. Verify object is a
15082 CONST_INT_P before looking at INTVAL.
15083 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
15084 instead of explicit test.
15086 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
15089 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
15090 address instead of to r1 and r11.
15092 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
15094 * config.gcc: Support "knm".
15095 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
15096 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
15098 * config/i386/i386.c (m_KNM): Define.
15099 (processor_target_table): Add "knm".
15101 (ix86_option_override_internal): Add "knm".
15102 (ix86_issue_rate): Add PROCESSOR_KNM.
15103 (ix86_adjust_cost): Ditto.
15104 (ia32_multipass_dfa_lookahead): Ditto.
15105 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
15106 (fold_builtin_cpu): Add M_INTEL_KNM.
15107 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
15108 (processor_type): Add PROCESSOR_KNM.
15109 * config/i386/x86-tune.def: Add m_KNM.
15110 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
15112 2017-09-20 Richard Biener <rguenther@suse.de>
15114 PR tree-optimization/80213
15115 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
15116 are allowed in empty BBs as well.
15117 (canonicalize_loop_closed_ssa): Also look for other complex
15119 (scop_detection::get_sese): Include the loop-closed PHI block
15121 (scop_detection::merge_sese): Remove code adding extra blocks.
15122 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
15123 (build_scops): Assert the final returned scop is invalid.
15125 2017-09-20 Richard Biener <rguenther@suse.de>
15127 PR tree-optimization/82264
15128 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
15130 (vn_phi_lookup): Likewise.
15131 (vn_phi_insert): Likewise.
15133 2017-09-20 Jakub Jelinek <jakub@redhat.com>
15135 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
15136 that fits into uhwi or shwi, add DW_AT_const_value regardless
15137 of early_dwarf without going through RTL, using add_AT_unsigned
15140 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
15141 (DEBUG_LTO_ABBREV_SECTION): Likewise.
15142 (DEBUG_LTO_MACINFO_SECTION): Likewise.
15143 (DEBUG_MACRO_SECTION): Likewise.
15144 (DEBUG_LTO_MACRO_SECTION): Likewise.
15145 (DEBUG_STR_DWO_SECTION): Likewise.
15146 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
15147 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
15148 (DEBUG_LTO_DWO_LINE_SECTION): Define.
15149 (DEBUG_LTO_LINE_STR_SECTION): Define.
15150 (init_sections_and_labels): Initialize debug_line_str_section
15151 variable. Initialize debug_loc_section for -gdwarf-5 to
15152 DEBUG_LOCLISTS_SECTION. Formatting fixes.
15154 2017-09-20 Richard Biener <rguenther@suse.de>
15156 * graphite-sese-to-poly.c (extract_affine): Properly handle
15157 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
15159 2017-09-20 Richard Biener <rguenther@suse.de>
15161 PR tree-optimization/81373
15162 * graphite-scop-detection.c (build_cross_bb_scalars_def):
15163 Force SESE live-out defs to be handled even if they are
15166 2017-09-19 Jeff Law <law@redhat.com>
15168 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
15169 nothing for stack adjustments with REG_STACK_CHECK.
15170 * sched-deps.c (parse_add_or_inc): Reject insns with
15171 REG_STACK_CHECK from dependency breaking.
15172 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
15173 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
15174 * reg-notes.def (STACK_CHECK): New note.
15176 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
15177 (ix86_expand_prologue): Dump stack clash info as needed.
15178 Call ix86_adjust_stack_and_probe_stack_clash as needed.
15180 * function.c (dump_stack_clash_frame_info): New function.
15181 * function.h (dump_stack_clash_frame_info): Prototype.
15182 (enum stack_clash_probes): New enum.
15184 * config/alpha/alpha.c (alpha_expand_prologue): Also check
15185 flag_stack_clash_protection.
15186 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
15187 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
15188 (arm_frame_pointer_required): Likewise.
15189 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
15190 (ia64_expand_prologue): Likewise.
15191 * config/mips/mips.c (mips_expand_prologue): Likewise.
15192 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
15193 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15194 (sparc_flat_expand_prologue): Likewise.
15195 * config/spu/spu.c (spu_expand_prologue): Likewise.
15197 * explow.c: Include "params.h".
15198 (anti_adjust_stack_and_probe_stack_clash): New function.
15199 (get_stack_check_protect): Likewise.
15200 (compute_stack_clash_protection_loop_data): Likewise.
15201 (emit_stack_clash_protection_loop_start): Likewise.
15202 (emit_stack_clash_protection_loop_end): Likewise.
15203 (allocate_dynamic_stack_space): Use get_stack_check_protect.
15204 Use anti_adjust_stack_and_probe_stack_clash.
15205 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
15206 (emit_stack_clash_protection_loop_start): Likewise.
15207 (emit_stack_clash_protection_loop_end): Likewise.
15208 * rtl.h (get_stack_check_protect): Prototype.
15209 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
15210 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
15211 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
15213 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
15215 * doc/tm.texi: Rebuilt.
15216 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
15217 get_stack_check_protect.
15218 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
15219 * config/arm/arm.c (arm_expand_prologue): Likewise.
15220 (arm_frame_pointer_required): Likewise.
15221 * config/i386/i386.c (ix86_expand_prologue): Likewise.
15222 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
15223 * config/mips/mips.c (mips_expand_prologue): Likewise.
15224 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
15225 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
15226 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15227 (sparc_flat_expand_prologue): Likewise.
15229 * common.opt (-fstack-clash-protection): New option.
15230 * flag-types.h (enum stack_check_type): Note difference between
15231 -fstack-check= and -fstack-clash-protection.
15232 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
15233 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
15234 * toplev.c (process_options): Issue warnings/errors for cases
15235 not handled with -fstack-clash-protection.
15236 * doc/invoke.texi (-fstack-clash-protection): Document new option.
15237 (-fstack-check): Note additional problem with -fstack-check=generic.
15238 Note that -fstack-check is primarily for Ada and refer users
15239 to -fstack-clash-protection for stack-clash-protection.
15240 Document new params for stack clash protection.
15242 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
15244 * config/i386/i386.c (ix86_split_long_move): Do not handle
15245 address used for LEA in a special way.
15247 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
15249 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
15251 2017-09-19 Martin Sebor <msebor@redhat.com>
15254 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
15255 of incompatible types.
15257 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
15259 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15260 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
15261 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15262 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
15264 2017-09-19 Richard Biener <rguenther@suse.de>
15266 PR tree-optimization/82244
15267 * tree-vrp.c (remove_range_assertions): Do not propagate
15268 a constant to abnormals but replace the assert with a copy.
15270 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
15272 PR rtl-optimization/57878
15273 PR rtl-optimization/68988
15274 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
15275 avoidance test involving non_reload_pseudos. Move frequency test
15276 below the general fragmentation avoidance test.
15278 2017-09-19 Richard Biener <rguenther@suse.de>
15280 PR tree-optimization/69728
15281 * graphite-sese-to-poly.c (schedule_error): New global.
15282 (add_loop_schedule): Handle empty domain by failing the
15284 (build_original_schedule): Handle schedule_error.
15286 2017-09-19 Richard Biener <rguenther@suse.de>
15288 * graphite-scop-detection.c (scop_detection::can_represent_loop):
15289 Do not iterate to sibling loops but only to siblings of inner
15292 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
15295 * config/m68k/m68k.md (moveq feeding equality comparison): Check
15296 that the registers are different.
15298 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
15300 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
15301 to processor_model and "amdfam17h" to arch_names_table.
15302 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
15304 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15307 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
15309 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15310 Alan Hayward <alan.hayward@arm.com>
15311 David Sherwood <david.sherwood@arm.com>
15313 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
15315 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
15317 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
15318 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
15319 vect_schedule_slp_instance.
15320 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
15321 Update call to vect_slp_analyze_node_operations. Simplify return
15323 (vect_slp_analyze_bb_1): Update call accordingly.
15324 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
15325 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
15326 (vect_schedule_slp): Update call accordingly.
15328 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15329 Alan Hayward <alan.hayward@arm.com>
15330 David Sherwood <david.sherwood@arm.com>
15332 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
15333 with types that aren't in fact scalar.
15335 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15337 * tree-vect-slp.c (vect_record_max_nunits): New function,
15339 (vect_build_slp_tree_1): ...here.
15340 (vect_build_slp_tree_2): Call it for phis too.
15342 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15344 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
15345 to vect_get_vec_def_for_operand when getting the mask operand.
15347 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15348 Alan Hayward <alan.hayward@arm.com>
15349 David Sherwood <david.sherwood@arm.com>
15351 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
15354 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15355 Alan Hayward <alan.hayward@arm.com>
15356 David Sherwood <david.sherwood@arm.com>
15358 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
15359 calculation for vector booleans.
15361 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15362 Alan Hayward <alan.hayward@arm.com>
15363 David Sherwood <david.sherwood@arm.com>
15365 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
15367 (vect_transform_stmt): ...here.
15368 (vect_analyze_stmt): Use it instead of calling
15369 vectorizable_live_operation directly.
15371 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
15373 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
15374 non-SIMT targets in acc vector loops.
15376 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
15378 * configure.ac: Add arc and check if assembler supports gdwarf2.
15379 * configure: Regenerate.
15381 2017-09-18 Richard Biener <rguenther@suse.de>
15383 PR tree-optimization/82220
15384 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
15385 epilogue niters from the min_profitable_iters compute.
15387 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15390 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
15391 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
15392 (ix86_init_pic_reg): Revert 2017-09-01 changes.
15394 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
15397 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
15398 switching to a new text section.
15400 2017-09-18 Richard Biener <rguenther@suse.de>
15402 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
15404 (build_alias_set): Reject aliases with no access function.
15406 2017-09-18 Richard Biener <rguenther@suse.de>
15408 PR tree-optimization/79622
15409 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
15411 (build_cross_bb_scalars_use): Likewise.
15413 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
15415 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
15417 2017-09-18 Alan Modra <amodra@gmail.com>
15420 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
15421 stack_pointer_rtx for count 0. Update comments. Break up
15422 large rtl expression.
15424 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
15426 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
15427 Increase to 20 bytes.
15428 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
15429 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
15430 or avx version of the stub.
15432 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
15435 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
15436 compute the minimum stack alignment. Also update preferred stack
15437 boundary for leaf functions.
15439 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
15441 PR tree-optimization/82228
15442 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
15445 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15447 * common/config/nds32/nds32-common.c
15448 (nds32_option_optimization_table): Refine formatting.
15449 (nds32_option_optimization_table): Use -fsched-pressure and
15450 -fomit-frame-pointer for specific optimization level.
15452 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15454 * config/nds32/nds32.c: Refine formatting and comments.
15455 * config/nds32/nds32.h: Likewise.
15456 * config/nds32/nds32.md: Likewise.
15457 * config/nds32/nds32-cost.c: Likewise.
15458 * config/nds32/nds32-isr.c: Likewise.
15459 * config/nds32/nds32-md-auxiliary.c: Likewise.
15460 * config/nds32/nds32-multiple.md: Likewise.
15461 * config/nds32/nds32-predicates.c: Likewise.
15463 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
15464 Jakub Jelinek <jakub@redhat.com>
15466 Add support for -std=c++2a.
15467 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
15469 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
15471 2017-09-15 Steve Ellcey <sellcey@cavium.com>
15474 * doc/extend.texi (Common Function Attributes): Add
15475 references to ARM, AArch64, and S/390 specific attributes.
15476 (Function Specific Option Pragmas): Add AArch64 and S/390
15477 to list of back ends that support the target pragma.
15479 2017-09-15 Nathan Sidwell <nathan@acm.org>
15481 * doc/standards.texi: Fix C++17 description. Update URLs for
15484 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
15486 * common.opt (Wcast-align=strict): New warning option.
15487 * doc/invoke.texi: Document -Wcast-align=strict.
15489 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
15491 * cgraph.h (cgraph_thunk_info): Add comments.
15492 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
15493 assert for VIRTUAL_* arguments stricter.
15495 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
15497 PR tree-optimization/71026
15498 * match.pd: Move RDIV patterns from fold-const.c
15499 * fold-const.c (distribute_real_division): Removed.
15500 (fold_binary_loc): Remove calls to distribute_real_divison.
15502 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15504 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
15505 c++1z and gnu++1z as deprecated. Change other references to
15506 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
15507 Change -Wc++1z-compat to -Wc++17-compat.
15508 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
15509 * dwarf2out.c (highest_c_language): Handle C++17.
15510 (gen_compile_unit_die): Likewise.
15512 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15514 PR rtl-optimization/82192
15515 * combine.c (make_extraction): Don't look through non-paradoxical
15516 SUBREGs or TRUNCATE if pos + len is or might be bigger than
15519 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15520 Alan Hayward <alan.hayward@arm.com>
15521 David Sherwood <david.sherwood@arm.com>
15523 * target.def (function_arg_offset): New hook.
15524 * targhooks.h (default_function_arg_offset): Declare.
15525 * targhooks.c (default_function_arg_offset): New function.
15526 * function.c (locate_and_pad_parm): Use
15527 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
15528 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
15529 (TARGET_FUNCTION_ARG_OFFSET): ...this.
15530 * doc/tm.texi: Regenerate.
15531 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
15532 * config/spu/spu.c (spu_function_arg_offset): New function.
15533 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
15534 * system.h (FUNCTION_ARG_OFFSET): Poison.
15536 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15537 Alan Hayard <alan.hayward@arm.com>
15538 David Sherwood <david.sherwood@arm.com>
15540 * target.def (truly_noop_truncation): New hook.
15541 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
15542 than TRULY_NOOP_TRUNCATION.
15543 * hooks.h (hook_bool_uint_uint_true): Declare.
15544 * hooks.c (hook_bool_uint_uint_true): New function.
15545 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
15546 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
15547 * doc/tm.texi: Regenerate.
15548 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
15549 rather than TRULY_NOOP_TRUNCATION in comments.
15550 (simplify_comparison): Likewise.
15551 (record_truncated_value): Likewise.
15552 * expmed.c (extract_bit_field_1): Likewise.
15553 (extract_split_bit_field): Likewise.
15554 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
15555 instead of TRULY_NOOP_TRUNCATION.
15556 * function.c (assign_parm_setup_block): Likewise.
15557 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
15558 * rtlhooks.c: Include target.h.
15559 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
15560 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
15561 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
15562 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
15563 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
15564 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
15565 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
15566 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
15567 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
15568 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
15569 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
15570 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
15571 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
15572 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
15573 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
15574 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
15575 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
15576 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
15577 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
15578 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
15579 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
15580 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
15581 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
15582 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
15583 * config/mips/mips.c (mips_truly_noop_truncation): New function.
15584 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15585 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15586 rather than TRULY_NOOP_TRUNCATION in comments.
15587 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
15588 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
15589 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
15590 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
15591 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
15592 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
15593 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
15594 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
15595 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
15596 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
15597 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
15598 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15599 rather than TRULY_NOOP_TRUNCATION in comments.
15600 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
15601 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
15602 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
15603 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
15604 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
15605 TRULY_NOOP_TRUNCATION condition.
15606 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
15607 (TRULY_NOOP_TRUNCATION): Delete.
15608 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
15609 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
15610 * config/spu/spu.c (spu_truly_noop_truncation): New function.
15611 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15612 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
15613 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
15614 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
15615 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15616 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15617 rather than TRULY_NOOP_TRUNCATION in comments.
15618 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
15619 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
15620 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
15621 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
15622 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
15623 * system.h (TRULY_NOOP_TRUNCATION): Poison.
15625 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
15628 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
15629 (*cmp_ior): Likewise.
15630 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
15631 (*ior_scc_scc_cmp): Likewise.
15632 (*and_scc_scc): Likewise.
15633 (*and_scc_scc_cmp): Likewise.
15635 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15636 Alan Hayard <alan.hayward@arm.com>
15637 David Sherwood <david.sherwood@arm.com>
15639 * target.def (can_change_mode_class): New hook.
15640 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
15641 (hard_regno_nregs): Likewise.
15642 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
15643 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
15644 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
15645 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
15646 (LOAD_EXTEND_OP): Update accordingly.
15647 * doc/tm.texi: Regenerate.
15648 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
15649 CANNOT_CHANGE_MODE_CLASS.
15650 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
15651 (REG_CAN_CHANGE_MODE_P): ...this new macro.
15652 * combine.c (simplify_set): Update accordingly.
15653 * emit-rtl.c (validate_subreg): Likewise.
15654 * recog.c (general_operand): Likewise.
15655 * regcprop.c (mode_change_ok): Likewise.
15656 * reload1.c (choose_reload_regs): Likewise.
15657 (inherit_piecemeal_p): Likewise.
15658 * rtlanal.c (simplify_subreg_regno): Likewise.
15659 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
15660 instead of CANNOT_CHANGE_MODE_CLASS.
15661 (reload_cse_simplify_operands): Likewise.
15662 * reload.c (push_reload): Use targetm.can_change_mode_class
15663 instead of CANNOT_CHANGE_MODE_CLASS.
15664 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
15665 REG_CANNOT_CHANGE_MODE_P.
15666 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15667 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
15668 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15669 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15670 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15671 (arm_can_change_mode_class): New function.
15672 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15673 than CANNOT_CHANGE_MODE_CLASS in comments.
15674 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15675 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
15676 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
15677 (ix86_can_change_mode_class): ...this new function, inverting the
15678 sense of the return value.
15679 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15680 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15681 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15682 (ia64_can_change_mode_class): New function.
15683 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15684 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
15685 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
15686 (m32c_can_change_mode_class): ...this new function, inverting the
15687 sense of the return value.
15688 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15689 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15690 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
15691 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
15692 (mips_can_change_mode_class): ...this new function, inverting the
15693 sense of the return value.
15694 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15695 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15696 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15697 (msp430_can_change_mode_class): New function.
15698 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15699 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
15700 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15701 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15702 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15703 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
15704 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15705 (pa_cannot_change_mode_class): Replace with...
15706 (pa_can_change_mode_class): ...this new function, inverting the
15707 sense of the return value.
15708 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15709 than CANNOT_CHANGE_MODE_CLASS in comments.
15710 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15711 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
15712 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15713 (pdp11_cannot_change_mode_class): Replace with...
15714 (pdp11_can_change_mode_class): ...this new function, inverting the
15715 sense of the return value.
15716 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15717 * config/powerpcspe/powerpcspe-protos.h
15718 (rs6000_cannot_change_mode_class_ptr): Delete.
15719 * config/powerpcspe/powerpcspe.c
15720 (rs6000_cannot_change_mode_class_ptr): Delete.
15721 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15722 (rs6000_option_override_internal): Assign to
15723 targetm.can_change_mode_class instead of
15724 rs6000_cannot_change_mode_class_ptr.
15725 (rs6000_cannot_change_mode_class): Replace with...
15726 (rs6000_can_change_mode_class): ...this new function, inverting the
15727 sense of the return value.
15728 (rs6000_debug_cannot_change_mode_class): Replace with...
15729 (rs6000_debug_can_change_mode_class): ...this new function.
15730 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15731 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
15732 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15733 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15734 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
15736 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
15737 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15738 (rs6000_option_override_internal): Assign to
15739 targetm.can_change_mode_class instead of
15740 rs6000_cannot_change_mode_class_ptr.
15741 (rs6000_cannot_change_mode_class): Replace with...
15742 (rs6000_can_change_mode_class): ...this new function, inverting the
15743 sense of the return value.
15744 (rs6000_debug_cannot_change_mode_class): Replace with...
15745 (rs6000_debug_can_change_mode_class): ...this new function.
15746 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15747 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
15748 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
15749 (s390_can_change_mode_class): ...this new function, inverting the
15750 sense of the return value.
15751 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15752 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15753 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
15754 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15755 (sh_cannot_change_mode_class): Replace with...
15756 (sh_can_change_mode_class): ...this new function, inverting the
15757 sense of the return value.
15758 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15759 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15760 (sparc_can_change_mode_class): New function.
15761 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15762 * config/spu/spu.c (spu_can_change_mode_class): New function.
15763 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15764 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15765 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15766 (visium_can_change_mode_class): New function.
15767 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
15769 2017-09-15 Richard Biener <rguenther@suse.de>
15771 PR tree-optimization/82217
15772 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
15773 but not undefined case.
15775 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15778 * postreload.c (reload_cse_simplify_operands): Skip
15779 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
15781 2017-09-15 Richard Biener <rguenther@suse.de>
15783 PR tree-optimization/68823
15784 * graphite-scop-detection.c (build_alias_set): If we have a
15785 possible dependence check whether we can handle them by just
15786 looking at the DRs DR_ACCESS_FNs.
15787 (build_scops): If build_alias_set fails, fail the SCOP.
15789 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15791 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15792 to support float128 built-in functions that require the ISA 3.0
15794 (BU_FLOAT128_3_HW): Likewise.
15795 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15796 built-in functions.
15797 (FMAF128): Likewise.
15799 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15800 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15801 floating point instructions.
15802 (rs6000_invalid_builtin): Likewise.
15803 (rs6000_builtin_mask_names): Likewise.
15804 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15805 (RS6000_BTM_FLOAT128_HW): Likewise.
15806 (RS6000_BTM_COMMON): Likewise.
15807 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15809 * doc/extend.texi (RS/6000 built-in functions): Document the
15810 IEEE 128-bit floating point square root and fused multiply-add
15811 built-in functions.
15813 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15815 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15816 reg (r2) isn't in the set of registers defined in the prologue.
15818 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15819 Alan Hayward <alan.hayward@arm.com>
15820 David Sherwood <david.sherwood@arm.com>
15822 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15823 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15824 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15825 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15826 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15828 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15829 max_vectorization_factor.
15830 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15832 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15833 Alan Hayward <alan.hayward@arm.com>
15834 David Sherwood <david.sherwood@arm.com>
15836 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15837 (vect_worthwhile_without_simd_p): Declare.
15838 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15839 (vectorizable_reduction): Use it.
15840 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15841 (vectorizable_operation): Likewise.
15843 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15844 Alan Hayward <alan.hayward@arm.com>
15845 David Sherwood <david.sherwood@arm.com>
15847 * tree-vectorizer.h (vect_get_num_copies): New function.
15848 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15849 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15850 (vectorizable_induction): Likewise.
15851 (vectorizable_live_operation): Likewise.
15852 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15853 (vectorizable_bswap): Likewise.
15854 (vectorizable_call): Likewise.
15855 (vectorizable_conversion): Likewise.
15856 (vectorizable_assignment): Likewise.
15857 (vectorizable_shift): Likewise.
15858 (vectorizable_operation): Likewise.
15859 (vectorizable_store): Likewise.
15860 (vectorizable_load): Likewise.
15861 (vectorizable_condition): Likewise.
15862 (vectorizable_comparison): Likewise.
15863 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15865 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15866 Alan Hayward <alan.hayward@arm.com>
15867 David Sherwood <david.sherwood@arm.com>
15869 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15870 of vect_init_vector.
15872 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15873 Alan Hayward <alan.hayward@arm.com>
15874 David Sherwood <david.sherwood@arm.com>
15876 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15877 an inline wrapper that provides a location.
15878 (gimple_build_vector): Likewise.
15879 * gimple-fold.c (gimple_build_vector_from_val): New function.
15880 (gimple_build_vector): Likewise.
15881 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15882 functions to build the initial value. Always return a gimple value.
15883 (get_initial_defs_for_reduction): Likewise. Only compute
15885 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15886 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15887 (vectorizable_induction): Use gimple_build_vector rather than
15890 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15891 Alan Hayward <alan.hayward@arm.com>
15892 David Sherwood <david.sherwood@arm.com>
15894 * target.h (vec_perm_indices): New typedef.
15895 (auto_vec_perm_indices): Likewise.
15896 * optabs-query.h: Include target.h
15897 (can_vec_perm_p): Take a vec_perm_indices *.
15898 * optabs-query.c (can_vec_perm_p): Likewise.
15899 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15900 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15901 * tree-vect-generic.c (lower_vec_perm): Likewise.
15902 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15903 (vect_grouped_load_supported): Likewise.
15904 (vect_shift_permute_load_chain): Likewise.
15905 (vect_permute_store_chain): Use auto_vec_perm_indices.
15906 (vect_permute_load_chain): Likewise.
15907 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15908 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15909 Update uses of can_vec_perm_p.
15910 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15911 mode with a number of elements. Take a vec_perm_indices *.
15912 (vect_create_epilog_for_reduction): Update accordingly.
15913 Use auto_vec_perm_indices.
15914 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15915 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15916 (vect_transform_slp_perm_load): Likewise.
15917 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15918 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15919 (vect_gen_perm_mask_checked): Likewise.
15920 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15921 (vect_gen_perm_mask_checked): Likewise.
15922 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15923 (vectorizable_store): Likewise.
15924 (vectorizable_load): Likewise.
15925 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15926 (vectorizable_bswap): Likewise.
15928 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15929 Alan Hayward <alan.hayward@arm.com>
15930 David Sherwood <david.sherwood@arm.com>
15932 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15933 * tree.c (build_vector): Likewise.
15934 (build_vector_from_ctor): Update accordingly.
15935 (build_vector_from_val): Likewise.
15936 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15937 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15938 * tree-vect-generic.c (add_rshift): Likewise.
15939 (expand_vector_divmod): Likewise.
15940 (optimize_vector_constructor): Likewise.
15941 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15942 (vect_transform_slp_perm_load): Likewise.
15943 (vect_schedule_slp_instance): Likewise.
15944 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15945 (vectorizable_call): Likewise.
15946 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15947 * expmed.c (make_tree): Likewise.
15948 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15949 a vector passed to build_vector.
15950 (fold_convert_const): Likewise.
15951 (exact_inverse): Likewise.
15952 (fold_ternary_loc): Likewise.
15953 (fold_relational_const): Likewise.
15954 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15955 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15956 (const_unop): Likewise. Store the reduction accumulator in a
15957 variable rather than an array.
15958 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15959 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15960 the new vector, rather than constructing it after the input arrays.
15961 (native_interpret_vector): Use auto_vec<tree> when building
15962 a vector passed to build_vector. Add elements in order.
15963 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15964 auto_vec<tree> when building a vector passed to build_vector.
15965 (vect_create_epilog_for_reduction): Likewise.
15966 (vectorizable_induction): Likewise.
15967 (get_initial_def_for_reduction): Likewise. Fix indentation of
15969 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15971 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15972 passed to build_vector.
15974 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15975 Alan Hayward <alan.hayward@arm.com>
15976 David Sherwood <david.sherwood@arm.com>
15978 * tree-core.h (tree_base::u): Add an "nelts" field.
15979 (tree_vector): Use VECTOR_CST_NELTS as the length.
15980 * tree.c (tree_size): Likewise.
15981 (make_vector): Initialize VECTOR_CST_NELTS.
15982 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15983 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15984 TYPE_VECTOR_SUBPARTS.
15985 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15986 as the number of units, setting it from VECTOR_CST_NELTS.
15987 (const_vector_from_tree): Likewise.
15988 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15989 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15990 (fold_negate_expr_1): Likewise.
15991 (fold_convert_const): Likewise.
15992 (const_binop): Likewise. Differentiate the number of output and
15994 (const_unop): Likewise.
15995 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15996 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15997 in cases that did the opposite.
15999 2017-09-14 Richard Biener <rguenther@suse.de>
16001 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
16004 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
16006 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
16008 2017-09-14 Jakub Jelinek <jakub@redhat.com>
16011 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
16012 if and where to split a bb, except for splitting before debug insn
16013 sequences followed by non-label real insn. Delete debug insns
16014 in between basic blocks.
16016 * combine.c (make_compound_operation_int): Formatting fixes.
16018 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
16019 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16020 * config/netbsd.h (LINK_EH_SPEC): Likewise.
16021 * config/sol2.h (LINK_EH_SPEC): Likewise.
16022 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16023 * config/s390/linux.h (LINK_SPEC): Likewise.
16024 * config/freebsd.h (LINK_EH_SPEC): Likewise.
16025 * config/openbsd.h (LINK_EH_SPEC): Likewise.
16026 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16027 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
16028 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
16029 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16030 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
16031 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
16033 2017-09-13 Jakub Jelinek <jakub@redhat.com>
16035 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
16037 (ENDFILE_LINUX_SPEC): Likewise.
16038 (LINK_EH_SPEC): Likewise.
16039 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
16040 (LINK_OS_LINUX_SPEC32): Likewise.
16041 (LINK_OS_LINUX_SPEC64): Likewise.
16042 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
16043 (LINK_OS_LINUX_SPEC): Likewise.
16045 2017-09-13 Martin Liska <mliska@suse.cz>
16047 PR middle-end/82154
16048 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
16049 CASE_HIGH is NULL_TREE.
16051 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
16052 Alan Hayward <alan.hayward@arm.com>
16053 David Sherwood <david.sherwood@arm.com>
16055 * target.def (secondary_memory_needed): New hook.
16056 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
16057 instead of SECONDARY_MEMORY_NEEDED.
16058 (secondary_memory_needed_mode): Likewise.
16059 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
16060 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
16061 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
16062 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
16063 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
16064 * doc/tm.texi: Regenerate.
16065 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
16066 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
16067 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16068 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
16069 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
16070 * config/i386/i386.c (inline_secondary_memory_needed): Put the
16071 mode argument first and change the reg_class arguments to reg_class_t.
16072 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
16073 Make static. Update the call to inline_secondary_memory_needed.
16074 (ix86_register_move_cost): Update the call to
16075 inline_secondary_memory_needed.
16076 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16077 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
16079 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
16080 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16082 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
16083 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
16084 * config/mips/mips.c (mips_secondary_memory_needed): Make static
16085 and match hook interface. Add comment from mips.h.
16086 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16087 * config/mmix/mmix.md (truncdfsf2): Refer to
16088 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16090 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
16091 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
16092 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16093 (pa_secondary_memory_needed): New function.
16094 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
16095 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
16096 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16097 (pdp11_secondary_memory_needed): Make static and match hook interface.
16098 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
16099 * config/powerpcspe/powerpcspe-protos.h
16100 (rs6000_secondary_memory_needed_ptr): Delete.
16101 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
16103 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16104 (rs6000_option_override_internal): Assign to
16105 targetm.secondary_memory_needed rather than
16106 rs6000_secondary_memory_needed_ptr.
16107 (rs6000_secondary_memory_needed): Match hook interface.
16108 (rs6000_debug_secondary_memory_needed): Likewise.
16109 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
16110 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
16111 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
16112 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16113 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
16114 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
16116 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
16117 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16118 (rs6000_option_override_internal): Assign to
16119 targetm.secondary_memory_needed rather than
16120 rs6000_secondary_memory_needed_ptr.
16121 (rs6000_secondary_memory_needed): Match hook interface.
16122 (rs6000_debug_secondary_memory_needed): Likewise.
16123 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
16124 * config/s390/s390.c (s390_secondary_memory_needed): New function.
16125 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16126 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
16127 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16128 (sparc_secondary_memory_needed): New function.
16129 * lra-constraints.c (check_and_process_move): Refer to
16130 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16132 (curr_insn_transform): Likewise.
16133 (process_alt_operands): Use targetm.secondary_memory_needed
16134 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16135 (check_secondary_memory_needed_p): Likewise.
16136 (choose_split_class): Likewise.
16137 * reload.c: Unconditionally include code that was previously
16138 conditional on SECONDARY_MEMORY_NEEDED.
16139 (push_secondary_reload): Use targetm.secondary_memory_needed
16140 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16141 (push_reload): Likewise.
16142 * reload1.c: Unconditionally include code that was previously
16143 conditional on SECONDARY_MEMORY_NEEDED.
16144 (choose_reload_regs): Use targetm.secondary_memory_needed
16145 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16146 (gen_reload): Likewise.
16147 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
16149 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
16150 Alan Hayward <alan.hayward@arm.com>
16151 David Sherwood <david.sherwood@arm.com>
16153 * target.def (secondary_memory_needed_mode): New hook:
16154 * targhooks.c (default_secondary_memory_needed_mode): Declare.
16155 * targhooks.h (default_secondary_memory_needed_mode): New function.
16156 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
16157 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
16158 * doc/tm.texi: Regenerate.
16159 * lra-constraints.c (check_and_process_move): Use
16160 targetm.secondary_memory_needed_mode instead of
16161 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
16162 (curr_insn_transform): Likewise.
16163 * reload.c (get_secondary_mem): Likewise.
16164 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16165 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
16167 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16168 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16169 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
16170 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16171 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
16173 * config/powerpcspe/powerpcspe-protos.h
16174 (rs6000_secondary_memory_needed_mode): Delete.
16175 * config/powerpcspe/powerpcspe.c
16176 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16177 (rs6000_secondary_memory_needed_mode): Make static.
16178 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16179 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
16181 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16183 (rs6000_secondary_memory_needed_mode): Make static.
16184 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16185 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
16186 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16187 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16188 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16190 (sparc_secondary_memory_needed_mode): New function.
16191 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
16193 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
16195 * config/aarch64/constraints.md (Umq): New constraint.
16196 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
16198 (mov<mode>): Update condition.
16200 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16202 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
16203 when bitposition is the same.
16205 2017-09-13 Richard Biener <rguenther@suse.de>
16207 * dwarf2out.c (output_die_symbol): Remove.
16208 (output_die): Do not output a DIEs symbol.
16210 2017-09-13 Richard Biener <rguenther@suse.de>
16212 PR middle-end/82128
16213 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
16214 default-def to avoid breaking iterator update with the weird
16215 interaction with cgraph_update_edges_for_call_stmt_node.
16217 2017-09-13 Richard Biener <rguenther@suse.de>
16219 * tree-cfg.c (verify_gimple_assign_binary): Add verification
16220 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
16221 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
16222 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
16224 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
16226 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
16227 Disable pc relative literal load irrespective of
16228 TARGET_FIX_ERR_A53_84341 for default.
16230 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
16232 * config/sparc/sparc.c (output_return): Output the source location of
16233 the insn in the delay slot, if any.
16234 (output_sibcall): Likewise.
16236 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
16239 * common.opt (-static-pie): New alias.
16240 (shared): Negate static-pie.
16241 (-no-pie): Update help text.
16243 (static-pie): New option.
16244 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
16245 -static-pie support.
16246 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
16247 (LINK_EH_SPEC): Likewise.
16248 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16249 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16250 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16251 * gcc.c (LINK_COMMAND_SPEC): Likewise.
16252 (init_gcc_specs): Likewise.
16253 (init_spec): Likewise.
16254 (display_help): Update help message for -pie.
16255 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
16258 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
16260 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
16261 (movdi_aarch64): Likewise.
16262 (movti_aarch64): Likewise.
16264 2017-09-12 Simon Wright <simon@pushface.org>
16267 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
16268 calculation of the minor version, always output as 0.
16270 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16273 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16274 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
16275 on it early, rather than manual conversion late. For
16276 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
16277 instead of performing manual conversion.
16279 2017-09-12 Carl Love <cel@us.ibm.com>
16281 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
16282 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
16284 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16285 VMULOSW): Add definitions.
16286 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16287 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16288 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16289 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16290 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16292 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16294 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
16296 (movti_aarch64): Likewise.
16297 (movdf_aarch64): Likewise.
16298 (movtf_aarch64): Likewise.
16299 (load_pairdi): Likewise.
16300 (store_pairdi): Likewise.
16301 (load_pairdf): Likewise.
16302 (store_pairdf): Likewise.
16303 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
16304 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
16305 (ldr_got_small_<mode>): Likewise.
16306 (ldr_got_small_28k_<mode>): Likewise.
16307 (ldr_got_tiny): Likewise.
16308 * config/aarch64/iterators.md (ldst_sz): New.
16309 (ldpstp_sz): Likewise.
16310 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
16312 (thunderx_load): Split load_8 to load_16.
16313 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
16315 (thunderx2t99_storepair_basic): Split store_8 to store_16.
16316 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
16317 (xgene1_store_pair): Split store_8 to store_16.
16318 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
16319 (falkor_st_0_st_sd): Split store_8 to store_16.
16321 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16323 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
16324 and store1/2/3/4 to store_4/8/12/16.
16325 * config/aarch64/aarch64.md: Update for rename.
16326 * config/arm/arm.md: Likewise.: Likewise.
16327 * config/arm/arm.c: Likewise.
16328 * config/arm/thumb1.md: Likewise.
16329 * config/arm/thumb2.md: Likewise.
16330 * config/arm/vfp.md: Likewise.
16331 * config/arm/arm-generic.md: Likewise.
16332 * config/arm/arm1020e.md: Likewise.
16333 * config/arm/arm1026ejs.md: Likewise.
16334 * config/arm/arm1136jfs.md: Likewise.
16335 * config/arm/arm926ejs.md: Likewise.
16336 * config/arm/cortex-a15.md: Likewise.
16337 * config/arm/cortex-a17.md: Likewise.
16338 * config/arm/cortex-a5.md: Likewise.
16339 * config/arm/cortex-a53.md: Likewise.
16340 * config/arm/cortex-a57.md: Likewise.
16341 * config/arm/cortex-a7.md: Likewise.
16342 * config/arm/cortex-a8.md: Likewise.
16343 * config/arm/cortex-a9.md: Likewise.
16344 * config/arm/cortex-m4.md: Likewise.
16345 * config/arm/cortex-m7.md: Likewise.
16346 * config/arm/cortex-r4.md: Likewise.
16347 * config/arm/exynos-m1.md: Likewise.
16348 * config/arm/fa526.md: Likewise.
16349 * config/arm/fa606te.md: Likewise.
16350 * config/arm/fa626te.md: Likewise.
16351 * config/arm/fa726te.md: Likewise.
16352 * config/arm/fmp626.md: Likewise.
16353 * config/arm/iwmmxt.md: Likewise.
16354 * config/arm/ldmstm.md: Likewise.
16355 * config/arm/marvell-pj4.md: Likewise.
16356 * config/arm/xgene1.md: Likewise.
16357 * config/aarch64/thunderx.md: Likewise.
16358 * config/aarch64/thunderx2t99.md: Likewise.
16359 * config/aarch64/falkor.md: Likewise.
16361 2017-09-12 Martin Liska <mliska@suse.cz>
16363 * attribs.c (private_lookup_attribute): New function.
16364 * attribs.h (private_lookup_attribute): Declared here.
16365 (lookup_attribute): Called from this place.
16367 2017-09-12 Richard Biener <rguenther@suse.de>
16369 PR tree-optimization/82157
16370 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
16371 stmts with side-effects.
16373 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16374 Alan Hayward <alan.hayward@arm.com>
16375 David Sherwood <david.sherwood@arm.com>
16377 * target.def (hard_regno_nregs): New hook.
16378 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
16379 * targhooks.h (default_hard_regno_nregs): Declare.
16380 * targhooks.c (default_hard_regno_nregs): New function.
16381 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
16382 (TARGET_HARD_REGNO_NREGS): ...this hook.
16383 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
16384 (CLASS_MAX_NREGS): Likewise.
16385 * doc/tm.texi: Regenerate.
16386 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
16387 instead of HARD_REGNO_NREGS.
16388 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
16389 HARD_REGNO_NREGS in the comment.
16390 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
16391 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
16392 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
16393 Return an unsigned int.
16394 (TARGET_HARD_REGNO_NREGS): Redefine.
16395 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
16396 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
16397 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16398 (arc_hard_regno_nregs): New function.
16399 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
16400 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
16401 (arm_hard_regno_nregs): New function.
16402 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
16403 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
16404 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
16405 (TARGET_HARD_REGNO_NREGS): Redefine.
16406 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
16407 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
16408 (HARD_REGNO_NREGS): Delete.
16409 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
16410 (cr16_hard_regno_nregs): New function.
16411 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
16412 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
16413 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
16414 (cris_hard_regno_nregs): New function.
16415 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
16416 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
16417 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16418 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
16419 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
16420 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
16421 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
16422 (frv_hard_regno_nregs): Make static. Take and return an
16424 (frv_class_max_nregs): Remove outdated copy of documentation.
16425 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
16426 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
16427 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
16428 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
16429 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
16430 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
16431 (TARGET_HARD_REGNO_NREGS): Redefine.
16432 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
16433 (CLASS_MAX_NREGS): Update comment.
16434 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
16435 (ia64_hard_regno_nregs): New function.
16436 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
16437 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
16438 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
16439 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
16440 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
16442 (m32c_hard_regno_nregs): Likewise. Make static.
16443 (TARGET_HARD_REGNO_NREGS): Redefine.
16444 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
16445 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
16446 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
16447 (m68k_hard_regno_nregs): New function.
16448 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
16449 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
16450 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
16451 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
16452 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
16453 Take and return an unsigned int.
16454 (TARGET_HARD_REGNO_NREGS): Redefine.
16455 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
16456 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16457 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
16458 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
16459 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
16460 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
16461 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
16462 (msp430_hard_regno_nregs): Make static. Take and return an
16464 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
16465 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
16466 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
16467 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
16468 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
16469 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
16470 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
16471 (TARGET_HARD_REGNO_NREGS): Redefine.
16472 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
16473 (PA_HARD_REGNO_NREGS): ...this.
16474 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
16475 (PA_HARD_REGNO_NREGS): ...this.
16476 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16477 (pa_hard_regno_nregs): New function.
16478 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
16479 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
16480 (pdp11_hard_regno_nregs): New function.
16481 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
16482 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
16483 (rs6000_hard_regno_nregs_hook): New function.
16484 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
16485 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
16486 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
16487 Take and return an unsigned int. Move earlier in file.
16488 (TARGET_HARD_REGNO_NREGS): Redefine.
16489 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
16490 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
16491 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
16492 (rl78_hard_regno_nregs): Make static. Take and return an
16494 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
16495 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
16496 (rs6000_hard_regno_nregs_hook): New function.
16497 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
16498 * config/rx/rx.c (rx_hard_regno_nregs): New function.
16499 (TARGET_HARD_REGNO_NREGS): Redefine.
16500 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
16501 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
16502 instead of HARD_REGNO_NREGS.
16503 (s390_hard_regno_nregs): New function.
16504 (s390_hard_regno_mode_ok): Add comment from s390.h.
16505 (TARGET_HARD_REGNO_NREGS): Redefine.
16506 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
16507 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
16508 (sh_hard_regno_nregs): New function.
16509 (sh_pass_in_reg_p): Use it.
16510 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
16511 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16512 (sparc_hard_regno_nregs): New function.
16513 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
16514 * config/spu/spu.c (spu_hard_regno_nregs): New function.
16515 (spu_function_arg_advance): Use it, supplying a valid register number.
16516 (TARGET_HARD_REGNO_NREGS): Redefine.
16517 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
16518 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
16519 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
16520 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
16521 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
16522 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
16523 (CLASS_MAX_NREGS): Remove copy of old documentation.
16524 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
16525 (visium_hard_regno_nregs): New function.
16526 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
16527 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
16528 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16529 xtensa_hard_regno_nregs): New function.
16530 * system.h (HARD_REGNO_NREGS): Poison.
16532 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16534 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
16535 hard_regno_nregs instead of HARD_REGNO_NREGS.
16536 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
16537 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
16538 (c6x_expand_epilogue): Likewise.
16539 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
16540 (frv_read_iacc_argument): Likewise.
16541 * config/sh/sh.c: Include regs.h.
16542 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16543 (regs_used): Likewise.
16544 (output_stack_adjust): Likewise.
16545 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
16546 * expmed.c: Include regs.h.
16547 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16548 * ree.c: Include regs.h.
16549 (combine_reaching_defs): Use hard_regno_nregs instead of
16551 (add_removable_extension): Likewise.
16553 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16555 * regs.h (hard_regno_nregs): Turn into a function.
16556 (end_hard_regno): Update accordingly.
16557 * caller-save.c (setup_save_areas): Likewise.
16558 (save_call_clobbered_regs): Likewise.
16559 (replace_reg_with_saved_mem): Likewise.
16560 (insert_restore): Likewise.
16561 (insert_save): Likewise.
16562 * combine.c (can_change_dest_mode): Likewise.
16563 (move_deaths): Likewise.
16564 (distribute_notes): Likewise.
16565 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
16566 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
16567 (rs6000_split_multireg_move): Likewise.
16568 (rs6000_register_move_cost): Likewise.
16569 (rs6000_memory_move_cost): Likewise.
16570 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
16571 (rs6000_split_multireg_move): Likewise.
16572 (rs6000_register_move_cost): Likewise.
16573 (rs6000_memory_move_cost): Likewise.
16574 * cselib.c (cselib_reset_table): Likewise.
16575 (cselib_lookup_1): Likewise.
16576 * emit-rtl.c (set_mode_and_regno): Likewise.
16577 * function.c (aggregate_value_p): Likewise.
16578 * ira-color.c (setup_profitable_hard_regs): Likewise.
16579 (check_hard_reg_p): Likewise.
16580 (calculate_saved_nregs): Likewise.
16581 (assign_hard_reg): Likewise.
16582 (improve_allocation): Likewise.
16583 (calculate_spill_cost): Likewise.
16584 * ira-emit.c (modify_move_list): Likewise.
16585 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
16586 (ira_hard_reg_in_set_p): Likewise.
16587 * ira.c (setup_reg_mode_hard_regset): Likewise.
16588 (clarify_prohibited_class_mode_regs): Likewise.
16589 (check_allocation): Likewise.
16590 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16591 (lra_setup_reg_renumber): Likewise.
16592 (setup_try_hard_regno_pseudos): Likewise.
16593 (spill_for): Likewise.
16594 (assign_hard_regno): Likewise.
16595 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
16596 * lra-constraints.c (in_class_p): Likewise.
16597 (lra_constraint_offset): Likewise.
16598 (simplify_operand_subreg): Likewise.
16599 (lra_constraints): Likewise.
16600 (split_reg): Likewise.
16601 (split_if_necessary): Likewise.
16602 (invariant_p): Likewise.
16603 (inherit_in_ebb): Likewise.
16604 * lra-lives.c (process_bb_lives): Likewise.
16605 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
16606 (get_hard_regs): Likewise.
16607 (do_remat): Likewise.
16608 * lra-spills.c (assign_spill_hard_regs): Likewise.
16609 * mode-switching.c (create_pre_exit): Likewise.
16610 * postreload.c (reload_combine_recognize_pattern): Likewise.
16611 * recog.c (peep2_find_free_register): Likewise.
16612 * regcprop.c (kill_value_regno): Likewise.
16613 (set_value_regno): Likewise.
16614 (copy_value): Likewise.
16615 (maybe_mode_change): Likewise.
16616 (find_oldest_value_reg): Likewise.
16617 (copyprop_hardreg_forward_1): Likewise.
16618 * regrename.c (check_new_reg_p): Likewise.
16619 (regrename_do_replace): Likewise.
16620 * reload.c (push_reload): Likewise.
16621 (combine_reloads): Likewise.
16622 (find_dummy_reload): Likewise.
16623 (operands_match_p): Likewise.
16624 (find_reloads): Likewise.
16625 (find_equiv_reg): Likewise.
16626 (reload_adjust_reg_for_mode): Likewise.
16627 * reload1.c (count_pseudo): Likewise.
16628 (count_spilled_pseudo): Likewise.
16629 (find_reg): Likewise.
16630 (clear_reload_reg_in_use): Likewise.
16631 (free_for_value_p): Likewise.
16632 (allocate_reload_reg): Likewise.
16633 (choose_reload_regs): Likewise.
16634 (reload_adjust_reg_for_temp): Likewise.
16635 (emit_reload_insns): Likewise.
16636 (delete_output_reload): Likewise.
16637 * rtlanal.c (subreg_get_info): Likewise.
16638 * sched-deps.c (sched_analyze_reg): Likewise.
16639 * sel-sched.c (init_regs_for_mode): Likewise.
16640 (mark_unavailable_hard_regs): Likewise.
16641 (choose_best_reg_1): Likewise.
16642 (verify_target_availability): Likewise.
16643 * valtrack.c (dead_debug_insert_temp): Likewise.
16644 * var-tracking.c (track_loc_p): Likewise.
16645 (emit_note_insn_var_location): Likewise.
16646 * varasm.c (make_decl_rtl): Likewise.
16647 * reginfo.c (choose_hard_reg_mode): Likewise.
16648 (init_reg_modes_target): Refer directly to
16649 this_target_regs->x_hard_regno_nregs.
16651 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16653 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
16654 instead of hard_regno_nregs.
16656 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16658 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
16659 end_hard_regno instead of hard_regno_nregs.
16660 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
16661 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
16662 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
16663 * ira-color.c (improve_allocation): Likewise.
16664 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16665 * lra-lives.c (mark_regno_live): Likewise.
16666 (mark_regno_dead): Likewise.
16667 * lra-remat.c (operand_to_remat): Likewise.
16668 * lra.c (collect_non_operand_hard_regs): Likewise.
16669 * postreload.c (reload_combine_note_store): Likewise.
16670 (move2add_valid_value_p): Likewise.
16671 * reload.c (regno_clobbered_p): Likewise.
16673 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16675 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
16677 * config/v850/v850.c (v850_reorg): Likewise.
16678 * reload.c (refers_to_regno_for_reload_p): Likewise.
16679 (find_equiv_reg): Likewise.
16680 * reload1.c (reload_reg_reaches_end_p): Likewise.
16682 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16684 * caller-save.c (add_used_regs): Use REG_NREGS instead of
16686 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
16687 * config/arm/arm.c (output_move_neon): Likewise.
16688 (arm_attr_length_move_neon): Likewise.
16689 (neon_split_vcombine): Likewise.
16690 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
16691 (c6x_mark_reg_written): Likewise.
16692 (c6x_dwarf_register_span): Likewise.
16693 * config/i386/i386.c (ix86_save_reg): Likewise.
16694 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
16695 (rws_access_reg): Likewise.
16696 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
16697 * mode-switching.c (create_pre_exit): Likewise.
16698 * ree.c (combine_reaching_defs): Likewise.
16699 (add_removable_extension): Likewise.
16700 * regcprop.c (find_oldest_value_reg): Likewise.
16701 (copyprop_hardreg_forward_1): Likewise.
16702 * reload.c (reload_inner_reg_of_subreg): Likewise.
16703 (push_reload): Likewise.
16704 (combine_reloads): Likewise.
16705 (find_dummy_reload): Likewise.
16706 (reload_adjust_reg_for_mode): Likewise.
16707 * reload1.c (find_reload_regs): Likewise.
16708 (forget_old_reloads_1): Likewise.
16709 (reload_reg_free_for_value_p): Likewise.
16710 (reload_adjust_reg_for_temp): Likewise.
16711 (emit_reload_insns): Likewise.
16712 (delete_output_reload): Likewise.
16713 * sel-sched.c (choose_best_reg_1): Likewise.
16714 (choose_best_pseudo_reg): Likewise.
16716 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16717 Alan Hayward <alan.hayward@arm.com>
16718 David Sherwood <david.sherwood@arm.com>
16720 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
16721 * target.def (slow_unaligned_access): New hook.
16722 * targhooks.h (default_slow_unaligned_access): Declare.
16723 * targhooks.c (default_slow_unaligned_access): New function.
16724 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
16725 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
16726 * doc/tm.texi: Regenerate.
16727 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
16728 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
16729 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
16731 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
16732 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
16734 (rs6000_slow_unaligned_access): New function.
16735 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16736 (expand_block_compare): Likewise.
16737 (expand_strn_compare): Likewise.
16738 (rs6000_rtx_costs): Likewise.
16739 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
16740 (riscv_slow_unaligned_access): Likewise.
16741 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
16742 (riscv_slow_unaligned_access_p): ...this and make static.
16743 (riscv_option_override): Update accordingly.
16744 (riscv_slow_unaligned_access): New function.
16745 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16746 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
16747 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16748 (rs6000_slow_unaligned_access): New function.
16749 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16750 (rs6000_rtx_costs): Likewise.
16751 * config/rs6000/rs6000-string.c (expand_block_compare)
16752 (expand_strn_compare): Use targetm.slow_unaligned_access instead
16753 of SLOW_UNALIGNED_ACCESS.
16754 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
16755 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
16756 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
16757 of SLOW_UNALIGNED_ACCESS.
16758 * expmed.c (simple_mem_bitfield_p): Likewise.
16759 * expr.c (alignment_for_piecewise_move): Likewise.
16760 (emit_group_load_1): Likewise.
16761 (emit_group_store): Likewise.
16762 (copy_blkmode_from_reg): Likewise.
16763 (emit_push_insn): Likewise.
16764 (expand_assignment): Likewise.
16765 (store_field): Likewise.
16766 (expand_expr_real_1): Likewise.
16767 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16768 * lra-constraints.c (simplify_operand_subreg): Likewise.
16769 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16770 * gimple-ssa-store-merging.c: Likewise in block comment at start
16772 * tree-ssa-strlen.c: Include target.h.
16773 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
16774 of SLOW_UNALIGNED_ACCESS.
16775 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
16777 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16779 PR rtl-optimization/82185
16780 * expmed.c (emit_store_flag_int): Only test tem if it has been
16783 2017-09-12 Richard Biener <rguenther@suse.de>
16785 PR middle-end/82149
16786 * match.pd ((FTYPE) N CMP CST): Fix typo.
16788 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16790 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16792 (mips_near_type_p): Add 'short_call' attribute as a synonym
16794 * doc/extend.texi (short_call): Document new function attribute.
16796 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16799 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16800 assertion check that in the condition.
16801 (get_atomic_generic_size): Likewise. Before testing if parameter
16802 has pointer type, if it has array type, call for C++
16803 default_conversion to perform array-to-pointer conversion.
16805 2017-09-12 Richard Biener <rguenther@suse.de>
16807 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16808 for operations we cannot scalarize.
16810 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16812 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16813 vectors heap vectors. Clean up comments.
16814 Make visited_bbs a reference.
16815 (profitable_jump_thread_path): Make GC
16816 vectors heap vectors. Clean up comments.
16818 (convert_and_register_jump_thread_path): Make GC vectors heap
16820 (check_subpath_and_update_thread_path): Same. Clean up comments.
16821 Make visited_bbs a reference.
16822 (handle_phi): Abstract common code to to
16823 register_jump_thread_path_if_profitable.
16824 Rename VAR_BB to DEF_BB.
16826 Make GC vectors heap vectors.
16827 Make visited_bbs a reference.
16828 (handle_assignment): Same.
16829 (register_jump_thread_path_if_profitable): New.
16830 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16832 Make GC vectors heap vectors. Clean up comments.
16833 Make visited_bbs a reference.
16834 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16835 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16838 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16841 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16842 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16844 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16846 Revert r251800 and r251799.
16848 2017-09-11 Martin Jambor <mjambor@suse.cz>
16851 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16852 arguments in advance.
16853 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16854 use it to find predecessor edges.
16855 (naive_outof_ssa): Collect vector of predecessors.
16857 2017-09-08 Jason Merrill <jason@redhat.com>
16859 PR c++/70029 - ICE with ref-qualifier and -flto
16860 * langhooks.h (struct lang_hooks_for_types): Add
16861 copy_lang_qualifiers.
16862 * attribs.c (build_type_attribute_qual_variant): Use it.
16863 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16865 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16866 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16868 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16871 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16872 (*mulsi3_sp64): New instruction.
16873 (mulsi3): New expander.
16875 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16877 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16879 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16881 * sancov.c: Include memmodel.h.
16883 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16886 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16889 2017-09-07 Carl Love <cel@us.ibm.com>
16891 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16892 the sldi instruction.
16894 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16896 * sancov.c: Include tm_p.h.
16898 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16901 * output.h (switch_to_other_text_partition): New declaration.
16902 * varasm.c (switch_to_other_text_partition): New function.
16903 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16904 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16905 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16906 to the other text partition before emitting LCL label and switch back
16907 after emitting the word after it.
16909 2017-09-07 Richard Biener <rguenther@suse.de>
16911 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16912 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16913 critical edges here, after loop init.
16914 (pass_data_pre): Remove PROP_no_crit_edges flags.
16915 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16916 for valueization of call args to avoid leaking VN_TOP.
16917 (visit_use): Assert we do not visit default defs.
16918 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16919 Use error_mark_node to more easily detect leaking VN_TOP.
16920 All default-defs are varying, not VN_TOP. Mark them visited.
16921 (run_scc_vn): Make code match comment.
16923 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16925 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16926 OPTION_MASK_FLOAT128_KEYWORD.
16927 (POWERPC_MASKS): Likewise.
16928 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16929 support for the -mfloat128-type option, and make -mfloat128
16930 default on PowerPC Linux systems. Define or undefine
16931 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16932 Define __float128 to be __ieee128 if IEEE 128-bit support is
16933 enabled, or undefine it.
16934 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16935 Delete defining __FLOAT128_TYPE__.
16936 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16937 -mfloat128-type option and make -mfloat128 default on PowerPC
16939 (TARGET_FLOAT128_TYPE): Likewise.
16940 (-mfloat128-type): Likewise.
16941 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16942 Delete the -mfloat128-type option and make -mfloat128 default on
16943 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16944 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16945 128-bit floating point is enabled. Change tests from using
16946 -mfloat128-type to -mfloat128.
16947 (rs6000_mangle_type): Use the correct mangling for the __float128
16948 type even if normal long double is restricted to 64-bits.
16949 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16951 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16952 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16953 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16955 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16956 documentation for -mfloat128.
16958 2017-09-06 Olivier Hainque <hainque@adacore.com>
16960 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16962 2017-09-06 Wish Wu <wishwu007@gmail.com>
16963 Jakub Jelinek <jakub@redhat.com>
16965 * asan.c (initialize_sanitizer_builtins): Add
16966 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16967 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16968 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16969 BT_FN_VOID_UINT64_PTR variables.
16970 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16971 (BT_FN_VOID_UINT16_UINT16): Likewise.
16972 (BT_FN_VOID_UINT32_UINT32): Likewise.
16973 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16974 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16975 (BT_FN_VOID_UINT64_PTR): Likewise.
16976 * common.opt (flag_sanitize_coverage): New variable.
16977 (fsanitize-coverage=trace-pc): Remove.
16978 (fsanitize-coverage=): Add.
16979 * flag-types.h (enum sanitize_coverage_code): New enum.
16980 * fold-const.c (fold_range_test): Disable non-short-circuit
16981 optimization if flag_sanitize_coverage.
16982 (fold_truth_andor): Likewise.
16983 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16984 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16985 (coverage_sanitizer_opts): New array.
16986 (get_closest_sanitizer_option): Add OPTS argument, handle also
16987 OPT_fsanitize_coverage_.
16988 (parse_sanitizer_options): Adjusted to also handle
16989 OPT_fsanitize_coverage_.
16990 (common_handle_option): Add OPT_fsanitize_coverage_.
16991 * sancov.c (instrument_comparison, instrument_switch): New function.
16992 (sancov_pass): Add trace-cmp support.
16993 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16994 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16995 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16996 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16997 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16998 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16999 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
17000 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
17001 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
17002 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
17004 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
17006 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
17007 error. Only quit immediately if parsing is complete.
17008 (BEGIN): Initialize fatal_err and parse_done.
17009 (begin fpu, end fpu): Check number of arguments.
17010 (begin arch, end arch): Likewise.
17011 (begin cpu, end cpu): Likewise.
17012 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
17013 (optalias): Likewise.
17015 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
17017 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
17018 * config/arm/arm-isa.h: Delete. Move definitions to ...
17019 * arm-cpus.in: ... here. Use new feature and fgroup values.
17020 * config/arm/arm.c (arm_option_override): Use lower case for feature
17022 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
17023 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
17024 * config/arm/parsecpu.awk (END): Add new command 'isa'.
17026 (print_isa_bits_for): New function.
17027 (gen_isa): New function.
17028 (gen_comm_data): Use print_isa_bits_for.
17029 (define feature): New keyword.
17030 (define fgroup): New keyword.
17031 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
17032 (arm-isa.h): Add rule to generate file.
17033 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
17034 case for feature bit names.
17036 2017-09-06 Richard Biener <rguenther@suse.de>
17038 * tree-ssa-pre.c (NECESSARY): Remove.
17039 (create_expression_by_pieces): Do not touch pass-local flags.
17040 (insert_into_preds_of_block): Likewise.
17041 (do_pre_regular_insertion): Likewise.
17042 (eliminate_insert): Likewise.
17043 (eliminate_dom_walker::before_dom_children): Likewise.
17044 (fini_eliminate): Do not look at inserted_exprs.
17045 (mark_operand_necessary): Remove.
17046 (remove_dead_inserted_code): Replace with simple work-list
17047 algorithm based on inserted_exprs and SSA uses.
17048 (pass_pre::execute): Re-order fini_eliminate and
17049 remove_dead_inserted_code.
17051 2017-09-06 Olivier Hainque <hainque@adacore.com>
17053 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
17054 for VxWorks 7. Adjust surrounding comments.
17056 2017-09-06 Richard Biener <rguenther@suse.de>
17058 * gimple-ssa-strength-reduction.c
17059 (find_candidates_dom_walker::before_dom_children): Also allow
17062 2017-09-06 Richard Biener <rguenther@suse.de>
17064 PR tree-optimization/82108
17065 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
17066 for gap in the non-permutation SLP case.
17068 2017-09-06 Martin Jambor <mjambor@suse.cz>
17070 PR tree-optimization/82078
17071 * tree-sra.c (sort_and_splice_var_accesses): Move call to
17072 add_access_to_work_queue...
17073 (build_accesses_from_assign): ...here.
17074 (propagate_all_subaccesses): Make sure racc is the group
17075 representative, if there is one.
17077 2017-09-06 Jakub Jelinek <jakub@redhat.com>
17079 PR middle-end/82095
17080 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
17083 2017-09-06 Richard Biener <rguenther@suse.de>
17085 * gimple-ssa-strength-reduction.c
17086 (find_candidates_dom_walker::before_doom_children): Use a
17087 type and not a mode check.
17089 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
17092 * config/arm/predicates.md (arm_general_adddi_operand): Create new
17094 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
17096 2017-09-05 Jeff Law <law@redhat.com>
17098 PR tree-optimization/64910
17099 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
17100 cases where we have 3 or more operands.
17102 2017-09-05 Jakub Jelinek <jakub@redhat.com>
17104 PR middle-end/81768
17105 * omp-low.c (lower_omp_for): Recompute tree invariant if
17106 gimple_omp_for_initial/final is ADDR_EXPR.
17108 PR middle-end/81768
17109 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
17110 into gimple val before gimplification fo the COND_EXPR.
17112 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
17114 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
17115 REGION_COPY argument.
17116 (thread_through_all_blocks): Remove unused argument to
17117 duplicate_thread_path.
17119 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17120 Alan Hayward <alan.hayward@arm.com>
17121 David Sherwood <david.sherwood@arm.com>
17123 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
17124 Take a scalar_mode rather than a machine_mode.
17125 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17126 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
17127 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17128 (aarch64_gen_adjusted_ldpstp): Likewise.
17129 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
17131 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17132 Alan Hayward <alan.hayward@arm.com>
17133 David Sherwood <david.sherwood@arm.com>
17135 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
17136 Take a scalar_int_mode instead of a machine_mode.
17137 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
17138 (aarch64_output_scalar_simd_mov_immediate): Likewise.
17139 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
17140 (aarch64_simd_attr_length_rglist): Delete.
17141 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
17142 a scalar_int_mode instead of a machine_mode.
17143 (aarch64_add_offset): Likewise.
17144 (aarch64_internal_mov_immediate): Likewise
17145 (aarch64_add_constant_internal): Likewise.
17146 (aarch64_add_constant): Likewise.
17147 (aarch64_movw_imm): Likewise.
17148 (aarch64_rtx_arith_op_extract_p): Likewise.
17149 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
17150 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
17151 Remove assert that the mode isn't a vector.
17152 (aarch64_output_scalar_simd_mov_immediate): Likewise.
17153 (aarch64_expand_mov_immediate): Update calls after above changes.
17154 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
17155 (aarch64_and_bitmask_imm): Check for scalar integer modes.
17156 (aarch64_move_imm): Likewise.
17157 (aarch64_can_const_movi_rtx_p): Likewise.
17158 (aarch64_strip_extend): Likewise.
17159 (aarch64_extr_rtx_p): Likewise.
17160 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
17161 a CONST_INT when the mode parameter is VOIDmode.
17162 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
17164 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17166 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
17167 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17168 (bitwise_type_for_mode): Update accordingly.
17170 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17172 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
17173 * stor-layout.c (mode_for_size_tree): Likewise.
17174 (mode_for_array): Update accordingly.
17175 (layout_decl): Likewise.
17176 (compute_record_mode): Likewise. Only set the mode once.
17178 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17180 * target.def (get_mask_mode): Change return type to opt_mode.
17182 * doc/tm.texi: Regenerate.
17183 * targhooks.h (default_get_mask_mode): Return an opt_mode.
17184 * targhooks.c (default_get_mask_mode): Likewise.
17185 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17186 * optabs-query.c (can_vec_mask_load_store_p): Update use of
17187 targetm.get_mask_mode.
17188 * tree.c (build_truth_vector_type): Likewise.
17190 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17192 * machmode.h (mode_for_vector): Return an opt_mode.
17193 * stor-layout.c (mode_for_vector): Likewise.
17194 (mode_for_int_vector): Update accordingly.
17195 (layout_type): Likewise.
17196 * config/i386/i386.c (emit_memmov): Likewise.
17197 (ix86_expand_set_or_movmem): Likewise.
17198 (ix86_expand_vector_init): Likewise.
17199 (ix86_get_mask_mode): Likewise.
17200 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
17202 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
17203 * expmed.c (extract_bit_field_1): Likewise.
17204 * expr.c (expand_expr_real_2): Likewise.
17205 * optabs-query.c (can_vec_perm_p): Likewise.
17206 (can_vec_mask_load_store_p): Likewise.
17207 * optabs.c (expand_vec_perm): Likewise.
17208 * targhooks.c (default_get_mask_mode): Likewise.
17209 * tree-vect-stmts.c (vectorizable_store): Likewise.
17210 (vectorizable_load): Likewise.
17211 (get_vectype_for_scalar_type_and_size): Likewise.
17213 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17215 * machmode.h (mode_for_int_vector): New function.
17216 * stor-layout.c (mode_for_int_vector): Likewise.
17217 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
17218 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
17219 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17220 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17221 (s390_expand_vcond): Likewise.
17223 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17225 * machmode.h (opt_machine_mode): New type.
17226 (opt_mode<T>): Allow construction from anything that can be
17228 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
17229 (mode_for_size): Return an opt_machine_mode.
17230 * stor-layout.c (mode_for_size): Likewise.
17231 (mode_for_size_tree): Update call accordingly.
17232 (bitwise_mode_for_mode): Likewise.
17233 (make_fract_type): Likewise.
17234 (make_accum_type): Likewise.
17235 * caller-save.c (replace_reg_with_saved_mem): Update call
17237 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17238 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17239 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17240 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17241 * expmed.c (extract_bit_field_1): Likewise.
17242 * reload.c (get_secondary_mem): Likewise.
17243 * varasm.c (assemble_integer): Likewise.
17244 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
17247 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17249 * machmode.h (decimal_float_mode_for_size): New function.
17250 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
17251 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
17252 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
17253 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
17254 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
17255 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
17257 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17259 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
17260 (get_builtin_sync_mode): Likewise.
17261 (expand_ifn_atomic_compare_exchange): Likewise.
17262 (expand_builtin_atomic_clear): Likewise.
17263 (expand_builtin_atomic_test_and_set): Likewise.
17264 (fold_builtin_atomic_always_lock_free): Likewise.
17265 * calls.c (compute_argument_addresses): Likewise.
17266 (emit_library_call_value_1): Likewise.
17267 (store_one_arg): Likewise.
17268 * combine.c (combine_instructions): Likewise.
17269 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
17270 * config/arm/arm.c (arm_function_value): Likewise.
17271 (aapcs_allocate_return_reg): Likewise.
17272 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
17273 * config/i386/i386.c (construct_container): Likewise.
17274 (ix86_gimplify_va_arg): Likewise.
17275 (ix86_expand_sse_cmp): Likewise.
17276 (emit_memmov): Likewise.
17277 (emit_memset): Likewise.
17278 (expand_small_movmem_or_setmem): Likewise.
17279 (ix86_expand_pextr): Likewise.
17280 (ix86_expand_pinsr): Likewise.
17281 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
17282 * config/microblaze/microblaze.c (microblaze_block_move_straight):
17284 * config/mips/mips.c (mips_function_value_1) Likewise.
17285 (mips_block_move_straight): Likewise.
17286 (mips_expand_ins_as_unaligned_store): Likewise.
17287 * config/powerpcspe/powerpcspe.c
17288 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17289 (rs6000_darwin64_record_arg_flush): Likewise.
17290 * config/rs6000/rs6000.c
17291 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17292 (rs6000_darwin64_record_arg_flush): Likewise.
17293 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
17294 (sparc_function_value_1): Likewise.
17295 * config/spu/spu.c (adjust_operand): Likewise.
17296 (spu_emit_branch_or_set): Likewise.
17297 (arith_immediate_p): Likewise.
17298 * emit-rtl.c (gen_lowpart_common): Likewise.
17299 * expr.c (expand_expr_real_1): Likewise.
17300 * function.c (assign_parm_setup_block): Likewise.
17301 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
17302 * reload1.c (alter_reg): Likewise.
17303 * stor-layout.c (mode_for_vector): Likewise.
17304 (layout_type): Likewise.
17306 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17308 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
17309 (spu_convert_move): Likewise.
17310 * lower-subreg.c (resolve_simple_move): Likewise.
17312 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17315 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
17316 define_insn to a define_expand.
17317 (altivec_vsum2sws_direct): New define_insn.
17318 (altivec_vsumsws): Convert from a define_insn to a define_expand.
17320 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17322 * config/arm/arm.c (arm_option_params_internal): Improve setting of
17325 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
17330 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
17331 realign stack if stack alignment needed is less than incoming
17334 2017-09-05 Marek Polacek <polacek@redhat.com>
17337 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
17340 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17342 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
17344 2017-09-05 Richard Biener <rguenther@suse.de>
17346 PR tree-optimization/82084
17347 * fold-const.c (can_native_encode_string_p): Handle wide characters.
17349 2017-09-05 Richard Biener <rguenther@suse.de>
17351 PR tree-optimization/82102
17352 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
17354 2017-09-05 Martin Liska <mliska@suse.cz>
17356 PR tree-optimization/82032
17357 * tree-cfg.c (generate_range_test): New function.
17358 * tree-cfg.h (generate_range_test): Declared here.
17359 * tree-cfgcleanup.c (convert_single_case_switch): New function.
17360 (cleanup_control_expr_graph): Use it.
17361 * tree-switch-conversion.c (try_switch_expansion): Remove
17363 (emit_case_nodes): Use generate_range_test.
17365 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17368 * config/i386/i386.md (*<btsc><mode>_mask): Add
17369 TARGET_USE_BT to insn constraint.
17370 (*btr<mode>_mask): Ditto.
17372 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
17374 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
17375 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
17377 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
17380 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
17381 TARGET_NEON and TARGET_IWMMXT.
17382 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
17383 TARGET_NEON and TARGET_IWMMXT.
17384 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
17386 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17388 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
17389 (ix86_rewrite_tls_address): Ditto.
17390 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
17391 (ix86_rewrite_tls_address_1): Ditto.
17392 (ix86_rewrite_tls_address): Ditto.
17393 * config/i386/predicates.md (tls_address_pattern): New predicate.
17394 * config/i386/i386.md (TLS address splitter): New splitter.
17396 2017-09-04 Richard Biener <rguenther@suse.de>
17398 PR tree-optimization/82084
17399 * fold-const.h (can_native_encode_string_p): Declare.
17400 * fold-const.c (can_native_encode_string_p): Factor out from ...
17401 (native_encode_string): ... here.
17402 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
17403 vectorizing stores from constants we later cannot handle.
17405 2017-09-04 Marek Polacek <polacek@redhat.com>
17408 * doc/invoke.texi: Update -Wtautological-compare documentation.
17410 2017-09-04 Jeff Law <law@redhat.com>
17412 PR tree-optimization/64910
17413 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
17414 swap the first and last operand if the last is a constant.
17416 2017-09-04 Marek Polacek <polacek@redhat.com>
17419 * convert.c (do_narrow): When sanitizing signed integer overflows,
17420 bail out for signed types.
17421 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
17423 2017-09-04 Richard Biener <rguenther@suse.de>
17425 PR tree-optimization/82060
17426 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17427 Move devirtualization after stmt folding and before EH/AB/noreturn
17428 cleanup to get the stmt refs canonicalized. Use a bool instead
17429 of gimple_modified_p since that doesn't work for NOPs. Schedule
17430 NOPs generated by folding for removal.
17432 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17433 Alan Hayward <alan.hayward@arm.com>
17434 David Sherwood <david.sherwood@arm.com>
17436 * coretypes.h (pad_direction): New enum.
17437 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
17438 (FUNCTION_ARG_PADDING): Likewise.
17439 * target.def (function_arg_padding): New hook.
17440 * targhooks.h (default_function_arg_padding): Declare.
17441 * targhooks.c (default_function_arg_padding): New function.
17442 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
17443 (TARGET_FUNCTION_ARG_PADDING): ...this.
17444 * doc/tm.texi: Regenerate.
17445 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
17446 instead of direction.
17447 (compute_argument_addresses): Likewise.
17448 (load_register_parameters): Likewise.
17449 (emit_library_call_value_1): Likewise.
17450 (store_one_arg): Use targetm.calls.function_arg_padding instead
17451 of FUNCTION_ARG_PADDING.
17452 (must_pass_in_stack_var_size_or_pad): Likewise.
17453 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
17454 (emit_group_store): Likewise.
17455 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
17456 instead of FUNCTION_ARG_PADDING.
17457 (emit_push_insn): Likewise, and propagate enum change throughout
17459 * function.h (direction): Delete.
17460 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
17462 * function.c (assign_parm_find_stack_rtl): Likewise.
17463 (assign_parm_setup_block_p): Likewise.
17464 (assign_parm_setup_block): Likewise.
17465 (gimplify_parameters): Likewise.
17466 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
17467 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
17469 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
17470 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17471 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
17472 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
17473 (aarch64_function_arg_padding): ...this new function.
17474 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
17475 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17476 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
17477 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17478 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
17479 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17480 (arm_pad_arg_upward): Replace with...
17481 (arm_function_arg_padding): ...this new function.
17482 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
17484 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
17485 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
17486 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17487 (ia64_hpux_function_arg_padding): Replace with...
17488 (ia64_function_arg_padding): ...this new function. Use pad_direction
17489 instead of direction. Check for TARGET_HPUX.
17490 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
17491 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17492 (iq2000_function_arg_padding): New function.
17493 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
17494 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
17495 (mips_function_arg_padding): ...this new function.
17496 (mips_pad_reg_upward): Update accordingly.
17497 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17498 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
17499 targetm.calls.function_arg_padding.
17500 (FUNCTION_ARG_PADDING): Delete.
17501 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17502 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
17503 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17504 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
17505 (nios2_block_reg_padding): Return pad_direction instead of direction.
17506 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
17507 instead of direction.
17508 (nios2_function_arg_padding): Likewise. Make static.
17509 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17510 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
17511 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
17512 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
17513 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17514 (pa_function_arg_padding): Make static. Return pad_direction instead
17516 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
17517 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17518 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
17519 instead of direction. Use targetm.calls.function_arg_padding.
17520 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
17521 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17522 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
17523 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
17524 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
17526 (function_arg_padding): Rename to...
17527 (rs6000_function_arg_padding): ...this. Make static. Return
17528 pad_direction instead of direction.
17529 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17530 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
17531 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17532 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
17533 instead of direction. Use targetm.calls.function_arg_padding.
17534 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
17535 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17536 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
17537 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
17538 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17539 (function_arg_padding): Rename to...
17540 (rs6000_function_arg_padding): ...this. Make static. Return
17541 pad_direction instead of direction.
17542 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17543 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
17544 * config/s390/s390.c (s390_function_arg_padding): New function.
17545 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17546 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
17547 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
17548 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17549 (function_arg_padding): Rename to...
17550 (sparc_function_arg_padding): ...this. Make static. Return
17551 pad_direction instead of direction.
17552 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
17553 * config/spu/spu.c (spu_function_arg_padding): New function.
17554 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17555 * system.h (FUNCTION_ARG_PADDING): Poison.
17557 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17558 Alan Hayward <alan.hayward@arm.com>
17559 David Sherwood <david.sherwood@arm.com>
17561 * target.def (modes_tieable_p): New hook.
17562 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
17563 (TARGET_MODES_TIEABLE_P): ...this.
17564 * doc/tm.texi.in: Regenerate.
17565 * hooks.h (hook_bool_mode_mode_true): Declare.
17566 * hooks.c (hook_bool_mode_mode_true): New function.
17567 * combine.c (subst): Use targetm.modes_tieable_p instead of
17569 * dse.c (find_shift_sequence): Likewise.
17570 * expmed.c (extract_low_bits): Likewise.
17571 * lower-subreg.c: Include target.h.
17572 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
17574 * rtlanal.c (rtx_cost): Likewise.
17575 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
17576 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
17577 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
17578 (TARGET_MODES_TIEABLE_P): Redefine.
17579 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
17580 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
17581 (TARGET_MODES_TIEABLE_P): Redefine.
17582 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
17583 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
17584 (arc_modes_tieable_p): New function.
17585 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
17586 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
17587 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
17588 (arm_modes_tieable_p): Make static.
17589 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
17590 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
17591 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
17592 (TARGET_MODES_TIEABLE_P): Redefine.
17593 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
17594 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
17595 (TARGET_MODES_TIEABLE_P): Redefine.
17596 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
17597 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
17598 (cr16_modes_tieable_p): New function.
17599 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
17600 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
17601 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
17602 (TRULY_NOOP_TRUNCATION): Update comment.
17603 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
17604 (TRULY_NOOP_TRUNCATION): Update comment.
17605 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
17606 (frv_modes_tieable_p): New function.
17607 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
17608 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
17609 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
17610 (TARGET_MODES_TIEABLE_P): Redefine.
17611 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
17612 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
17613 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
17614 (TARGET_MODES_TIEABLE_P): Redefine.
17615 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
17616 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
17617 (ia64_modes_tieable_p): New function.
17618 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
17619 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
17620 (iq2000_modes_tieable_p): New function.
17621 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
17622 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
17623 (lm32_modes_tieable_p): New function.
17624 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
17625 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
17626 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
17627 (TARGET_MODES_TIEABLE_P): Redefine.
17628 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
17629 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
17630 (m32r_modes_tieable_p): New function.
17631 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
17632 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
17633 (m68k_modes_tieable_p): New function.
17634 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
17635 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
17636 (mcore_modes_tieable_p): New function.
17637 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
17638 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
17640 (TARGET_MODES_TIEABLE_P): Redefine.
17641 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
17642 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
17643 * config/mips/mips.c (mips_modes_tieable_p): Make static.
17644 (TARGET_MODES_TIEABLE_P): Redefine.
17645 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
17646 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
17647 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
17648 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
17649 (mn10300_modes_tieable_p): ...this and make static.
17650 (TARGET_MODES_TIEABLE_P): Redefine.
17651 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
17652 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
17653 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
17654 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
17655 (msp430_modes_tieable_p): Make static.
17656 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
17657 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
17658 (TARGET_MODES_TIEABLE_P): Redefine.
17659 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
17660 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
17661 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
17662 (TARGET_MODES_TIEABLE_P): Redefine.
17663 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
17664 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
17665 * config/pa/pa.c (pa_modes_tieable_p): Make static.
17666 (TARGET_MODES_TIEABLE_P): Redefine.
17667 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
17668 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
17669 (pdp11_modes_tieable_p): New function.
17670 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
17671 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
17672 (rs6000_modes_tieable_p): New function.
17673 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17674 * config/powerpcspe/powerpcspe.md: Update comment.
17675 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
17676 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
17677 (TARGET_MODES_TIEABLE_P): Redefine.
17678 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
17679 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
17680 (rl78_modes_tieable_p): New function.
17681 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
17682 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
17683 (rs6000_modes_tieable_p): New function.
17684 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17685 * config/rs6000/rs6000.md: Update comment.
17686 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
17687 * config/rx/rx.c (rx_modes_tieable_p): New function.
17688 (TARGET_MODES_TIEABLE_P): Redefine.
17689 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
17690 * config/s390/s390.c (s390_modes_tieable_p): New function.
17691 (TARGET_MODES_TIEABLE_P): Redefine.
17692 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
17693 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
17694 (sh_modes_tieable_p): New function.
17695 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
17696 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
17697 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
17698 (sparc_modes_tieable_p): Make static.
17699 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
17700 * config/spu/spu.c (spu_modes_tieable_p): New function.
17701 (TARGET_MODES_TIEABLE_P): Redefine.
17702 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
17703 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
17704 (TARGET_MODES_TIEABLE_P): Redefine.
17705 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
17706 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
17707 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
17708 * config/v850/v850.c (v850_modes_tieable_p): New function.
17709 (TARGET_MODES_TIEABLE_P): Redefine.
17710 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
17711 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
17712 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
17713 (visium_modes_tieable_p): New function.
17714 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
17715 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
17716 (xtensa_modes_tieable_p): New function.
17717 * system.h (MODES_TIEABLE_P): Poison.
17719 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17720 Alan Hayward <alan.hayward@arm.com>
17721 David Sherwood <david.sherwood@arm.com>
17723 * target.def (hard_regno_mode_ok): New hook.
17724 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
17725 (TARGET_HARD_REGNO_MODE_OK): ...this.
17726 * doc/tm.texi.in: Regenerate.
17727 * hooks.h (hook_bool_uint_mode_true): Declare.
17728 * hooks.c (hook_bool_uint_mode_true): New function.
17729 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
17730 HARD_REGNO_MODE_OK.
17731 * genpreds.c (write_insn_preds_c): Add an include of target.h.
17732 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
17733 instead of HARD_REGNO_MODE_OK.
17734 * caller-save.c: Include target.h.
17735 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
17736 HARD_REGNO_MODE_OK.
17737 * combine.c (can_combine_p): Likewise.
17738 (combinable_i3pat): Likewise.
17739 (can_change_dest_mode): Likewise.
17740 * expr.c (init_expr_target): Likewise.
17741 (convert_move): Likewise.
17742 (convert_modes): Likewise.
17743 * ira.c (setup_prohibited_class_mode_regs): Likewise.
17744 (setup_prohibited_mode_move_regs): Likewise.
17745 * ira.h (target_ira): Likewise.
17746 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17747 * lra-constraints.c (process_alt_operands): Likewise.
17748 (split_reg): Likewise.
17749 * recog.c (peep2_find_free_register): Likewise.
17750 * ree.c (combine_reaching_defs): Likewise.
17751 * regcprop.c (maybe_mode_change): Likewise.
17752 * reginfo.c (init_reg_sets_1): Likewise.
17753 (choose_hard_reg_mode): Likewise.
17754 (simplifiable_subregs): Likewise.
17755 * regrename.c (check_new_reg_p): Likewise.
17756 * reload.c (find_valid_class): Likewise.
17757 (find_valid_class_1): Likewise.
17758 (reload_inner_reg_of_subreg): Likewise.
17759 (push_reload): Likewise.
17760 (combine_reloads): Likewise.
17761 (find_dummy_reload): Likewise.
17762 (find_reloads): Likewise.
17763 * reload1.c (find_reg): Likewise.
17764 (set_reload_reg): Likewise.
17765 (allocate_reload_reg): Likewise.
17766 (choose_reload_regs): Likewise.
17767 (reload_adjust_reg_for_temp): Likewise.
17768 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
17769 (simplify_subreg_regno): Likewise.
17770 * sel-sched.c (init_regs_for_mode): Likewise.
17771 * varasm.c (make_decl_rtl): Likewise.
17772 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
17773 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
17774 HARD_REGNO_MODE_OK.
17775 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
17776 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
17777 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17778 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
17779 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
17780 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17781 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
17782 (arc_mode_class): Delete.
17783 (HARD_REGNO_MODE_OK): Delete.
17784 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17785 (arc_hard_regno_mode_ok): Rename old array to...
17786 (arc_hard_regno_mode_ok_modes): ...this.
17787 (arc_conditional_register_usage): Update accordingly.
17788 (arc_mode_class): Make static.
17789 (arc_hard_regno_mode_ok): New function.
17790 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17791 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17792 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17793 (arm_hard_regno_mode_ok): Make static.
17794 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17795 HARD_REGNO_MODE_OK.
17796 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17797 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17798 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17800 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17801 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17802 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17803 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17804 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17805 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17806 * config/bfin/predicates.md (valid_reg_operand): Use
17807 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17808 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17809 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17810 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17811 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17812 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17813 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17814 (cr16_hard_regno_mode_ok): Make static and return a bool.
17815 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17816 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17817 (cris_hard_regno_mode_ok): New function.
17818 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17819 (epiphany_mode_class): Delete.
17820 (HARD_REGNO_MODE_OK): Delete.
17821 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17822 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17823 (hard_regno_mode_ok): Rename to...
17824 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17825 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17826 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17827 HARD_REGNO_MODE_OK.
17828 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17829 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17830 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17831 (frv_hard_regno_mode_ok): Make static and return a bool.
17832 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17833 HARD_REGNO_MODE_OK.
17834 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17835 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17836 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17837 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17839 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17840 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17841 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17842 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17844 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17845 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17846 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17847 (ia64_hard_regno_mode_ok): New function.
17848 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17849 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17850 (iq2000_hard_regno_mode_ok): New function.
17851 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17852 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17853 (lm32_hard_regno_mode_ok): New function.
17854 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17855 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17856 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17857 instead of HARD_REGNO_MODE_OK.
17858 (m32c_hard_regno_ok): Rename to...
17859 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17860 (m32c_cannot_change_mode_class): Update accordingly.
17861 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17862 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17863 (m32r_mode_class): Delete.
17864 (HARD_REGNO_MODE_OK): Delete.
17865 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17866 (m32r_hard_regno_mode_ok): Rename to...
17867 (m32r_hard_regno_modes): ...this.
17868 (m32r_mode_class): Make static.
17869 (m32r_hard_regno_mode_ok): New function.
17870 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17871 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17872 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17873 (m68k_hard_regno_mode_ok): Make static.
17874 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17875 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17876 (mcore_hard_regno_mode_ok): New function.
17877 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17878 (HARD_REGNO_MODE_OK): Delete.
17879 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17881 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17882 (microblaze_hard_regno_mode_ok): New function.
17883 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17884 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17885 (mips_hard_regno_mode_ok): Delete.
17886 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17887 (mips_hard_regno_mode_ok_p): ...this and make static.
17888 (mips_hard_regno_mode_ok_p): Rename to...
17889 (mips_hard_regno_mode_ok_uncached): ...this.
17890 (mips_hard_regno_mode_ok): New function.
17891 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17892 of HARD_REGNO_MODE_OK.
17893 (mips_option_override): Update after above name changes.
17894 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17895 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17896 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17897 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17898 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17899 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17900 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17901 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17902 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17903 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17904 (msp430_hard_regno_mode_ok): Make static and return a bool.
17905 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17906 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17907 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17909 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17910 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17911 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17912 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17913 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17914 (PA_HARD_REGNO_MODE_OK): ...this
17915 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17916 (PA_HARD_REGNO_MODE_OK): ...this.
17917 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17918 (pa_hard_regno_mode_ok): New function.
17919 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17920 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17921 (pdp11_hard_regno_mode_ok): New function.
17922 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17923 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17925 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17927 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17928 (rs6000_hard_regno_mode_ok): Rename to...
17929 (rs6000_hard_regno_mode_ok_uncached): ...this.
17930 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17931 (rs6000_hard_regno_mode_ok): New function.
17932 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17933 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17934 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17935 (riscv_hard_regno_mode_ok): ...this and make static.
17936 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17937 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17938 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17939 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17940 (rl78_hard_regno_mode_ok): Make static and return bool.
17941 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17942 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17944 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17945 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17946 (rs6000_hard_regno_mode_ok): Rename to...
17947 (rs6000_hard_regno_mode_ok_uncached): ...this.
17948 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17949 (rs6000_hard_regno_mode_ok): New function.
17950 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17951 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17952 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17953 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17954 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17955 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17956 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17957 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17958 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17959 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17960 (sh_hard_regno_mode_ok): Make static.
17961 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17962 instead of HARD_REGNO_MODE_OK.
17963 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17964 (sparc_mode_class): Delete.
17965 (HARD_REGNO_MODE_OK): Delete.
17966 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17967 (hard_regno_mode_classes): Make static.
17968 (sparc_mode_class): Likewise.
17969 (sparc_hard_regno_mode_ok): New function.
17970 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17971 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17972 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17974 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17975 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17976 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17977 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17978 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17979 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17980 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17981 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17982 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17983 (visium_hard_regno_mode_ok): New function.
17984 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17985 instead of HARD_REGNO_MODE_OK.
17986 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17987 (HARD_REGNO_MODE_OK): Delete.
17988 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17989 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17990 (xtensa_option_override): Update accordingly.
17991 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17992 (xtensa_hard_regno_mode_ok): New function.
17993 * system.h (HARD_REGNO_MODE_OK): Poison.
17995 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17996 Alan Hayward <alan.hayward@arm.com>
17997 David Sherwood <david.sherwood@arm.com>
17999 * target.def (hard_regno_call_part_clobbered): New hook.
18000 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
18001 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
18002 * doc/tm.texi: Regenerate.
18003 * hooks.h (hook_bool_uint_mode_false): Declare.
18004 * hooks.c (hook_bool_uint_mode_false): New function.
18005 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18006 * cselib.c (cselib_process_insn): Use
18007 targetm.hard_regno_call_part_clobbered instead of
18008 HARD_REGNO_CALL_PART_CLOBBERED.
18009 * ira-conflicts.c (ira_build_conflicts): Likewise.
18010 * ira-costs.c (ira_tune_allocno_costs): Likewise.
18011 * lra-constraints.c (need_for_call_save_p): Likewise.
18012 * lra-lives.c: Include target.h.
18013 (check_pseudos_live_through_calls): Use
18014 targetm.hard_regno_call_part_clobbered instead of
18015 HARD_REGNO_CALL_PART_CLOBBERED.
18016 * regcprop.c: Include target.h.
18017 (copyprop_hardreg_forward_1): Use
18018 targetm.hard_regno_call_part_clobbered instead of
18019 HARD_REGNO_CALL_PART_CLOBBERED.
18020 * reginfo.c (choose_hard_reg_mode): Likewise.
18021 * regrename.c (check_new_reg_p): Likewise.
18022 * reload.c (find_equiv_reg): Likewise.
18023 * reload1.c (emit_reload_insns): Likewise.
18024 * sched-deps.c (deps_analyze_insn): Likewise.
18025 * sel-sched.c (init_regs_for_mode): Likewise.
18026 (mark_unavailable_hard_regs): Likewise.
18027 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
18028 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18029 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
18031 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18032 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18033 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
18035 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
18037 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18038 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18039 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
18041 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18042 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18043 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
18045 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18046 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
18048 * config/powerpcspe/powerpcspe.c
18049 (rs6000_hard_regno_call_part_clobbered): New function.
18050 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18051 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18052 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
18054 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18055 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18056 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
18058 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18059 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18060 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
18062 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
18063 Alan Hayward <alan.hayward@arm.com>
18064 David Sherwood <david.sherwood@arm.com>
18066 * rtl.h (subreg_memory_offset): Declare.
18067 * emit-rtl.c (subreg_memory_offset): New function.
18068 * expmed.c (store_bit_field_1): Use it.
18069 * expr.c (undefined_operand_subword_p): Likewise.
18070 * simplify-rtx.c (simplify_subreg): Likewise.
18072 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
18074 PR rtl-optimization/57448
18077 * optabs.c (expand_atomic_load): Place compiler memory barriers if
18078 using atomic_load pattern.
18079 (expand_atomic_store): Likewise.
18081 2017-09-04 Jakub Jelinek <jakub@redhat.com>
18084 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
18085 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
18086 handling. Use replace_call_with_value with NULL instead of
18087 gsi_replace, unlink_stmt_vdef and release_defs.
18089 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
18092 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
18094 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
18097 * rtl.h (emit_library_call_value_1): Declare.
18098 (emit_library_call): Replace declaration with a series of overloads.
18099 Remove the parameter count argument.
18100 (emit_library_call_value): Likewise.
18101 * calls.c (emit_library_call_value_1): Make global. Replace varargs
18102 with an "rtx_mode_t *".
18103 (emit_library_call_value): Delete.
18104 (emit_library_call): Likewise.
18105 * asan.c (asan_emit_stack_protection): Update calls accordingly.
18106 (asan_emit_allocas_unpoison): Likewise.
18107 * builtins.c (expand_builtin_powi): Likewise.
18108 (expand_asan_emit_allocas_unpoison): Likewise.
18109 * cfgexpand.c (expand_main_function): Likewise.
18110 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
18111 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
18112 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
18113 * config/arm/arm.c (arm_trampoline_init): Likewise.
18114 (arm_call_tls_get_addr): Likewise.
18115 (arm_expand_divmod_libfunc): Likewise.
18116 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
18117 (smulsi3_highpart): Likewise.
18118 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
18119 (c6x_expand_compare): Likewise.
18120 (c6x_expand_movmem): Likewise.
18121 * config/frv/frv.c (frv_trampoline_init): Likewise.
18122 * config/i386/i386.c (ix86_trampoline_init): Likewise.
18123 (ix86_expand_divmod_libfunc): Likewise.
18124 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
18125 (ia64_expand_compare): Likewise.
18126 (ia64_profile_hook): Likewise.
18127 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
18128 (nonlocal_goto): Likewise.
18129 (restore_stack_nonlocal): Likewise.
18130 * config/m32r/m32r.c (block_move_call): Likewise.
18131 (m32r_trampoline_init): Likewise.
18132 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
18133 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
18134 (m68k_call_m68k_read_tp): Likewise.
18135 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
18136 (microblaze_expand_divide): Likewise.
18137 * config/mips/mips.h (mips_args): Likewise.
18138 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
18139 (MIPS_ICACHE_SYNC): Likewise.
18140 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
18141 (nios2_trampoline_init): Likewise.
18142 * config/pa/pa.c (hppa_tls_call): Likewise.
18143 (pa_trampoline_init): Likewise.
18144 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
18145 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
18146 (expand_strn_compare): Likewise.
18147 (rs6000_generate_compare): Likewise.
18148 (rs6000_expand_float128_convert): Likewise.
18149 (output_profile_hook): Likewise.
18150 (rs6000_trampoline_init): Likewise.
18151 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
18152 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
18153 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
18154 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
18155 (rs6000_generate_compare): Likewise.
18156 (rs6000_expand_float128_convert): Likewise.
18157 (output_profile_hook): Likewise.
18158 (rs6000_trampoline_init): Likewise.
18159 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
18160 * config/sh/sh.c (sh_trampoline_init): Likewise.
18161 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
18162 (sparc_emit_float_lib_cmp): Likewise.
18163 (sparc32_initialize_trampoline): Likewise.
18164 (sparc64_initialize_trampoline): Likewise.
18165 (sparc_profile_hook): Likewise.
18166 * config/spu/spu.c (ea_load_store): Likewise.
18167 * config/spu/spu.md (floatunssidf2): Likewise.
18168 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
18169 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
18170 * config/visium/visium.c (expand_block_move_4): Likewise.
18171 (expand_block_move_2): Likewise.
18172 (expand_block_move_1): Likewise.
18173 (expand_block_set_4): Likewise.
18174 (expand_block_set_2): Likewise.
18175 (expand_block_set_1): Likewise.
18176 (visium_trampoline_init): Likewise.
18177 (visium_profile_hook): Likewise.
18178 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
18179 (xtensa_setup_frame_addresses): Likewise.
18180 (xtensa_trampoline_init): Likewise.
18181 * except.c (sjlj_emit_function_enter): Likewise.
18182 (sjlj_emit_function_exit): Likewise.
18183 * explow.c (allocate_dynamic_stack_space): Likewise.
18184 (probe_stack_range): Likewise.
18185 * expr.c (convert_mode_scalar): Likewise.
18186 * optabs.c (expand_binop): Likewise.
18187 (expand_twoval_binop_libfunc): Likewise.
18188 (expand_unop): Likewise.
18189 (prepare_cmp_insn): Likewise.
18190 (prepare_float_lib_cmp): Likewise.
18191 (expand_float): Likewise.
18192 (expand_fix): Likewise.
18193 (expand_fixed_convert): Likewise.
18194 (maybe_emit_sync_lock_test_and_set): Likewise.
18195 (expand_atomic_compare_and_swap): Likewise.
18196 (expand_mem_thread_fence): Likewise.
18197 (expand_atomic_fetch_op): Likewise.
18199 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
18201 * doc/generic.texi (OpenACC): Adjust URL.
18202 * doc/invoke.texi (C Dialect Options): Ditto.
18204 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
18206 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
18207 predicate for operand 1. Add (m,<S>) constraint.
18208 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
18209 Prevent memory operand 1 with register operand 2.
18211 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
18213 PR rtl-optimization/82024
18214 * combine.c (try_combine): If the combination result is a PARALLEL,
18215 and we only need to retain the SET in there that would be placed
18216 at I2, check that we can place that at I3 instead, before doing so.
18218 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18221 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
18223 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
18224 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
18227 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
18228 Jeff Law <law@redhat.com>
18230 * varasm.c (bss_initializer_p): Do not put constants into .bss
18231 (categorize_decl_for_section): Handle bss_initializer_p returning
18232 false when DECL_INITIAL is NULL.
18234 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18237 * config/s390/s390.c (s390_can_inline_p): New function.
18239 2017-09-01 Jeff Law <law@redhat.com>
18241 PR tree-optimization/82052
18242 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18243 Always initialize the returned slot after a hash table miss
18244 when INSERT is true.
18246 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
18248 * config/s390/s390.md (mem_signal_fence): Remove.
18249 * doc/md.texi (mem_signal_fence): Remove.
18250 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
18252 * target-insns.def (mem_signal_fence): Remove.
18254 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18257 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
18260 * asan.c (create_odr_indicator): Strip name encoding from assembler
18261 name before appending it after __odr_asan_.
18263 2017-09-01 Martin Liska <mliska@suse.cz>
18265 PR tree-optimization/82059
18266 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
18267 frequency only when an edge is redirected.
18269 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18271 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
18272 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
18273 (arc_conditional_register_usage): Remove ARC600 lp_count
18275 (arc_file_start): Emit Tag_ARC_CPU_variation.
18276 (arc_can_use_doloop_p): New conditions to use ZOLs.
18277 (hwloop_fail): New function.
18278 (hwloop_optimize): Likewise.
18279 (hwloop_pattern_reg): Likewise.
18280 (arc_doloop_hooks): New struct, to be used with reorg_loops.
18281 (arc_reorg_loops): New function, calls reorg_loops.
18282 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
18283 (arc600_corereg_hazard): Remove ZOL checking, case handled by
18285 (arc_loop_hazard): Remove function, functionality moved into
18287 (arc_hazard): Remove arc_loop_hazard call.
18288 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
18289 into hwloop_optimize.
18290 (arc_label_align): Remove ZOL handling.
18291 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
18292 * config/arc/arc.md (doloop_begin): Remove pattern.
18293 (doloop_begin_i): Likewise.
18294 (doloop_end_i): Likewise.
18295 (doloop_fallback): Likewise.
18296 (doloop_fallback_m): Likewise.
18297 (doloop_end): Reimplement expand.
18298 (arc_lp): New pattern for LP instruction.
18299 (loop_end): New pattern.
18300 (loop_fail): Likewise.
18301 (decrement_and_branch_until_zero): Likewise.
18302 * config/arc/arc.opt (mlpc-width): New option.
18303 * doc/invoke.texi (mlpc-width): Document option.
18305 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18307 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
18308 (arc_ccfsm_advance): Fix checking for delay slots.
18309 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
18311 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18313 * config/arc/arc.md (movqi_insn): Add stores to save constant long
18315 (movhi_insn): Update store instruction constraint which are saving
18316 6-bit short immediates.
18317 (movsi_insn): Consider also short scaled load operations.
18318 (zero_extendhisi2_i): Use Usd constraint instead of T.
18319 (extendhisi2_i): Add q constraint.
18320 (arc_clzsi2): Add type and length attributes.
18321 (arc_ctzsi2): Likewise.
18322 * config/arc/constraints.md (Usc): Update constraint, the
18323 assembler can parse two relocations for a single instruction.
18325 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18327 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
18328 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
18330 2017-08-31 Olivier Hainque <hainque@adacore.com>
18332 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
18333 match as powerpc-wrs-vxworks*.
18335 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
18337 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
18338 register constraint for by-element operand.
18339 (aarch64_mls_elt_merge<mode>): Likewise.
18341 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18343 * config/arc/arc.c (arc_can_follow_jump): Check for short
18346 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18348 * config.gcc: Use g.opt for arc.
18349 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
18350 functionality moved to ...
18351 (legitimate_scaled_address_p): New function, ...here.
18352 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
18353 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
18354 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
18356 (arc_override_options): Handle G option.
18357 (arc_output_pic_addr_const): Correct function definition.
18358 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
18359 (arc_decl_anon_ns_mem_p): Delete.
18360 (arc_in_small_data_p): Overhaul this function to take into
18361 consideration the value given via G option.
18362 (arc_rewrite_small_data_1): Renamed and corrected old
18363 arc_rewrite_small_data function.
18364 (arc_rewrite_small_data): New function.
18365 (small_data_pattern): Don't use pic_offset_table_rtx.
18366 * config/arc/arc.h (CC1_SPEC): Recognize G option.
18367 * config/arc/simdext.md (movmisalignv2hi): Use
18368 prepare_move_operands function.
18370 (movmisalign*): Likewise.
18371 * doc/invoke.texi (ARC options): Document -G option.
18373 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18375 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
18377 * config/arc/arc.c (arc_print_operand): Output scalled address for
18378 sdata whenever is possible.
18379 (arc_in_small_data_p): Allow sdata for 64bit datum when double
18380 load/stores are available.
18381 (compact_sda_memory_operand): Check for the alignment required by
18382 code density instructions.
18383 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
18385 * config/arc/constraints.md (Usd): Update constraint.
18386 (Us0): New constraint.
18387 (Usc): Update constraint.
18389 2017-08-31 Richard Biener <rguenther@suse.de>
18391 PR middle-end/82054
18392 * dwarf2out.c (dwarf2out_early_global_decl): Process each
18393 function only once.
18395 2017-08-31 Tamar Christina <tamar.christina@arm.com>
18397 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
18398 Resize type_signature.
18400 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18401 Alan Hayward <alan.hayward@arm.com>
18402 David Sherwood <david.sherwood@arm.com>
18404 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
18405 subregs whose inner modes can be stored in GPRs.
18406 (aarch64_classify_index): Likewise.
18408 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18409 Alan Hayward <alan.hayward@arm.com>
18410 David Sherwood <david.sherwood@arm.com>
18412 * config/aarch64/iterators.md (V_cmp_result): Rename to...
18413 (V_INT_EQUIV): ...this.
18414 (v_cmp_result): Rename to...
18415 (v_int_equiv): ...this.
18416 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
18417 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
18418 (copysign<mode>3): Likewise.
18419 (aarch64_simd_bsl<mode>_internal): Likewise.
18420 (aarch64_simd_bsl<mode>): Likewise.
18421 (vec_cmp<mode><mode>): Likewise.
18422 (vcond<mode><mode>): Likewise.
18423 (vcond<v_cmp_mixed><mode>): Likewise.
18424 (vcondu<mode><v_cmp_mixed>): Likewise.
18425 (aarch64_cm<optab><mode>): Likewise.
18426 (aarch64_cmtst<mode>): Likewise.
18427 (aarch64_fac<optab><mode>): Likewise.
18428 (vec_perm_const<mode>): Likewise.
18429 (vcond_mask_<mode><v_cmp_result>): Rename to...
18430 (vcond_mask_<mode><v_int_equiv>): ...this.
18431 (vec_cmp<mode><v_cmp_result>): Rename to...
18432 (vec_cmp<mode><v_int_equiv>): ...this.
18434 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18435 Alan Hayward <alan.hayward@arm.com>
18436 David Sherwood <david.sherwood@arm.com>
18438 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
18440 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
18441 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
18442 (UNSPEC_LD4_DREG): New unspecs.
18443 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
18444 (aarch64_ld2<mode>_dreg_be): Replace with...
18445 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
18447 (aarch64_ld3<mode>_dreg_le)
18448 (aarch64_ld3<mode>_dreg_be): Replace with...
18449 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
18451 (aarch64_ld4<mode>_dreg_le)
18452 (aarch64_ld4<mode>_dreg_be): Replace with...
18453 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
18456 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18458 PR tree-optimization/81987
18459 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
18460 insert an initializer in a location not dominated by the stride
18463 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18465 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
18466 on the entire header of the finally block in the fallthru case.
18468 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18470 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
18472 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
18474 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
18475 rs6000_emit_move_from_cr and call renamed function.
18476 (rs6000_emit_prologue): Call renamed functions.
18477 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
18478 movesi_from_cr, remove volatile CRs.
18480 2017-08-30 Jon Beniston <jon@beniston.com>
18481 Richard Biener <rguenther@suse.de>
18483 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
18484 of VECTOR_MODE_P check.
18485 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
18486 element vector types.
18488 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18490 * df.h (df_read_modify_subreg_p): Remove in favor of...
18491 * rtl.h (read_modify_subreg_p): ...this new function. Take a
18492 const_rtx instead of an rtx.
18493 * cprop.c (local_cprop_find_used_regs): Update accordingly.
18494 * df-problems.c (df_word_lr_mark_ref): Likewise.
18495 * ira-lives.c (mark_pseudo_reg_live): Likewise.
18496 (mark_pseudo_reg_dead): Likewise.
18497 (mark_ref_dead): Likewise.
18498 * reginfo.c (init_subregs_of_mode): Likewise.
18499 * sched-deps.c (sched_analyze_1): Likewise.
18500 * df-scan.c (df_def_record_1): Likewise.
18501 (df_uses_record): Likewise.
18502 (df_read_modify_subreg_p): Remove in favor of...
18503 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
18504 const_rtx instead of an rtx.
18506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18507 Alan Hayward <alan.hayward@arm.com>
18508 David Sherwood <david.sherwood@arm.com>
18510 * rtl.h (partial_subreg_p): New function.
18511 * caller-save.c (save_call_clobbered_regs): Use it.
18512 * calls.c (expand_call): Likewise.
18513 * combine.c (combinable_i3pat): Likewise.
18514 (simplify_set): Likewise.
18515 (make_extraction): Likewise.
18516 (make_compound_operation_int): Likewise.
18517 (gen_lowpart_or_truncate): Likewise.
18518 (force_to_mode): Likewise.
18519 (make_field_assignment): Likewise.
18520 (reg_truncated_to_mode): Likewise.
18521 (record_truncated_value): Likewise.
18522 (move_deaths): Likewise.
18523 * cse.c (record_jump_cond): Likewise.
18524 (cse_insn): Likewise.
18525 * cselib.c (cselib_lookup_1): Likewise.
18526 * expmed.c (extract_bit_field_using_extv): Likewise.
18527 * function.c (assign_parm_setup_reg): Likewise.
18528 * ifcvt.c (noce_convert_multiple_sets): Likewise.
18529 * ira-build.c (create_insn_allocnos): Likewise.
18530 * lra-coalesce.c (merge_pseudos): Likewise.
18531 * lra-constraints.c (match_reload): Likewise.
18532 (simplify_operand_subreg): Likewise.
18533 (curr_insn_transform): Likewise.
18534 * lra-lives.c (process_bb_lives): Likewise.
18535 * lra.c (new_insn_reg): Likewise.
18536 (lra_substitute_pseudo): Likewise.
18537 * regcprop.c (mode_change_ok): Likewise.
18538 (maybe_mode_change): Likewise.
18539 (copyprop_hardreg_forward_1): Likewise.
18540 * reload.c (push_reload): Likewise.
18541 (find_reloads): Likewise.
18542 (find_reloads_subreg_address): Likewise.
18543 * reload1.c (alter_reg): Likewise.
18544 (eliminate_regs_1): Likewise.
18545 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18547 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
18549 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
18550 back to if statements, including unpack.
18552 2017-08-30 Martin Liska <mliska@suse.cz>
18554 PR inline-asm/82001
18555 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
18557 (func_checker::compare_asm_inputs_outputs): ... this function.
18558 (func_checker::compare_gimple_asm): Use the function to compare
18559 also ASM constrains.
18560 * ipa-icf-gimple.h: Rename the function.
18562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18563 Alan Hayward <alan.hayward@arm.com>
18564 David Sherwood <david.sherwood@arm.com>
18566 * coretypes.h (complex_mode): New type.
18567 * gdbhooks.py (build_pretty_printer): Handle it.
18568 * machmode.h (complex_mode): New class.
18569 (complex_mode::includes_p): New function.
18570 (is_complex_int_mode): Likewise.
18571 (is_complex_float_mode): Likewise.
18572 * genmodes.c (get_mode_class): Handle complex mode classes.
18573 * function.c (expand_function_end): Use is_complex_int_mode.
18575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18576 Alan Hayward <alan.hayward@arm.com>
18577 David Sherwood <david.sherwood@arm.com>
18579 * coretypes.h (scalar_mode_pod): New typedef.
18580 * gdbhooks.py (build_pretty_printer): Handle it.
18581 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
18582 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
18583 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
18584 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
18585 as_a <scalar_mode>.
18587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18588 Alan Hayward <alan.hayward@arm.com>
18589 David Sherwood <david.sherwood@arm.com>
18591 * machmode.h (mode_for_vector): Take a scalar_mode instead
18593 * stor-layout.c (mode_for_vector): Likewise.
18594 * explow.c (promote_mode): Use as_a <scalar_mode>.
18595 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
18597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18598 Alan Hayward <alan.hayward@arm.com>
18599 David Sherwood <david.sherwood@arm.com>
18601 * target.def (preferred_simd_mode): Take a scalar_mode
18602 instead of a machine_mode.
18603 * targhooks.h (default_preferred_simd_mode): Likewise.
18604 * targhooks.c (default_preferred_simd_mode): Likewise.
18605 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
18606 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
18607 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
18608 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
18609 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
18610 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
18611 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
18612 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
18614 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
18615 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
18616 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
18617 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
18618 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
18619 * doc/tm.texi: Regenerate.
18620 * optabs-query.c (can_vec_mask_load_store_p): Return false for
18623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18624 Alan Hayward <alan.hayward@arm.com>
18625 David Sherwood <david.sherwood@arm.com>
18627 * target.def (scalar_mode_supported_p): Take a scalar_mode
18628 instead of a machine_mode.
18629 * targhooks.h (default_scalar_mode_supported_p): Likewise.
18630 * targhooks.c (default_scalar_mode_supported_p): Likewise.
18631 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
18632 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18633 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
18634 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
18635 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
18636 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
18637 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
18638 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
18639 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
18640 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
18641 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
18642 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
18644 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
18645 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
18646 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18647 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18648 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
18650 * doc/tm.texi: Regenerate.
18652 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18653 Alan Hayward <alan.hayward@arm.com>
18654 David Sherwood <david.sherwood@arm.com>
18656 * coretypes.h (opt_scalar_mode): New typedef.
18657 * gdbhooks.py (build_pretty_printers): Handle it.
18658 * machmode.h (mode_iterator::get_2xwider): Add overload for
18660 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
18662 * expr.c (convert_mode_scalar): Likewise.
18663 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18664 * optabs.c (expand_float): Likewise.
18665 (expand_fix): Likewise.
18666 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18668 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18669 Alan Hayward <alan.hayward@arm.com>
18670 David Sherwood <david.sherwood@arm.com>
18672 * optabs.c (expand_float): Explicitly check for scalars before
18673 using a branching expansion.
18674 (expand_fix): Likewise.
18676 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18677 Alan Hayward <alan.hayward@arm.com>
18678 David Sherwood <david.sherwood@arm.com>
18680 * expr.c (convert_mode): Split scalar handling out into...
18681 (convert_mode_scalar): ...this new function. Treat the modes
18684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18685 Alan Hayward <alan.hayward@arm.com>
18686 David Sherwood <david.sherwood@arm.com>
18688 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
18690 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
18692 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18693 Alan Hayward <alan.hayward@arm.com>
18694 David Sherwood <david.sherwood@arm.com>
18696 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
18697 rather than a machine_mode.
18698 (fixed_from_string): Likewise.
18699 (fixed_convert): Likewise.
18700 (fixed_convert_from_int): Likewise.
18701 (fixed_convert_from_real): Likewise.
18702 (real_convert_from_fixed): Likewise.
18703 * fixed-value.c (fixed_from_double_int): Likewise.
18704 (fixed_from_string): Likewise.
18705 (fixed_convert): Likewise.
18706 (fixed_convert_from_int): Likewise.
18707 (fixed_convert_from_real): Likewise.
18708 (real_convert_from_fixed): Likewise.
18709 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
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 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
18716 of separate mode class checks. Do not allow vector modes here.
18717 (immed_wide_int_const): Use as_a <scalar_mode>.
18718 * explow.c (trunc_int_for_mode): Likewise.
18719 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
18720 (wi::shwi): Likewise.
18721 (wi::min_value): Likewise.
18722 (wi::max_value): Likewise.
18723 * dwarf2out.c (loc_descriptor): Likewise.
18724 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
18725 for CONST_WIDE_INT.
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 * tree.h (SCALAR_TYPE_MODE): New macro.
18732 * expr.c (expand_expr_addr_expr_1): Use it.
18733 (expand_expr_real_2): Likewise.
18734 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
18735 (fold_convert_const_fixed_from_int): Likewise.
18736 (fold_convert_const_fixed_from_real): Likewise.
18737 (native_encode_fixed): Likewise
18738 (native_encode_complex): Likewise
18739 (native_encode_vector): Likewise.
18740 (native_interpret_fixed): Likewise.
18741 (native_interpret_real): Likewise.
18742 (native_interpret_complex): Likewise.
18743 (native_interpret_vector): Likewise.
18744 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
18745 (simd_clone_adjust_argument_types): Likewise.
18746 (simd_clone_init_simd_arrays): Likewise.
18747 (simd_clone_adjust): Likewise.
18748 * stor-layout.c (layout_type): Likewise.
18749 * tree.c (build_minus_one_cst): Likewise.
18750 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18751 * tree-inline.c (estimate_move_cost): Likewise.
18752 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
18753 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
18754 (vectorizable_reduction): Likewise.
18755 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
18756 (vect_recog_mixed_size_cond_pattern): Likewise.
18757 (check_bool_pattern): Likewise.
18758 (adjust_bool_pattern): Likewise.
18759 (search_type_for_mask_1): Likewise.
18760 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
18761 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18762 (vectorizable_load): Likewise.
18763 (vectorizable_store): Likewise.
18764 * ubsan.c (ubsan_encode_value): Likewise.
18765 * varasm.c (output_constant): Likewise.
18767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18768 Alan Hayward <alan.hayward@arm.com>
18769 David Sherwood <david.sherwood@arm.com>
18771 * coretypes.h (scalar_mode): New class.
18772 * machmode.h (scalar_mode): Likewise.
18773 (scalar_mode::includes_p): New function.
18774 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
18775 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
18776 * genmodes.c (get_mode_class): Handle remaining scalar modes.
18777 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
18778 * expmed.c (store_bit_field_1): Likewise.
18779 (extract_bit_field_1): Likewise.
18780 * expr.c (write_complex_part): Likewise.
18781 (read_complex_part): Likewise.
18782 (emit_move_complex_push): Likewise.
18783 (expand_expr_real_2): Likewise.
18784 * function.c (assign_parm_setup_reg): Likewise.
18785 (assign_parms_unsplit_complex): Likewise.
18786 * optabs.c (expand_binop): Likewise.
18787 * rtlanal.c (subreg_get_info): Likewise.
18788 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18789 * varasm.c (output_constant_pool_2): Likewise.
18791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18792 Alan Hayward <alan.hayward@arm.com>
18793 David Sherwood <david.sherwood@arm.com>
18795 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18797 (expmed_mult_highpart_optab): Likewise.
18798 (expmed_mult_highpart): Likewise.
18800 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18801 Alan Hayward <alan.hayward@arm.com>
18802 David Sherwood <david.sherwood@arm.com>
18804 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18805 instead of a machine_mode.
18806 (builtin_memset_read_str): Likewise.
18807 * builtins.c (c_readstr): Likewise.
18808 (builtin_memcpy_read_str): Likewise.
18809 (builtin_strncpy_read_str): Likewise.
18810 (builtin_memset_read_str): Likewise.
18811 (builtin_memset_gen_str): Likewise.
18812 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18813 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18814 instead of a machine_mode.
18815 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18817 (make_extraction): Likewise.
18818 (try_widen_shift_mode): Take and return scalar_int_modes instead
18820 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18821 a scalar_int_mode instead of a machine_mode.
18822 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18823 (avr_addr_space_pointer_mode): Likewise.
18824 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18825 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18826 (msp430_unwind_word_mode): Likewise.
18827 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18828 (spu_addr_space_pointer_mode): Likewise.
18829 (spu_addr_space_address_mode): Likewise.
18830 (spu_libgcc_cmp_return_mode): Likewise.
18831 (spu_libgcc_shift_count_mode): Likewise.
18832 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18833 (rl78_addr_space_pointer_mode): Likewise.
18834 (fl78_unwind_word_mode): Likewise.
18835 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18837 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18838 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18839 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18840 (mips_valid_pointer_mode): Likewise.
18841 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18842 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18843 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18845 (ft32_addr_space_address_mode): Likewise.
18846 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18847 scalar_int_mode instead of a machine_mode.
18848 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18850 (m32c_addr_space_address_mode): Likewise.
18851 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18852 (rs6000_eh_return_filter_mode): Likewise.
18853 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18854 (rs6000_eh_return_filter_mode): Likewise.
18855 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18856 (s390_libgcc_shift_count_mode): Likewise.
18857 (s390_unwind_word_mode): Likewise.
18858 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18860 * target.def (mode_rep_extended): Likewise.
18861 (valid_pointer_mode): Likewise.
18862 (addr_space.valid_pointer_mode): Likewise.
18863 (eh_return_filter_mode): Return a scalar_int_mode rather than
18865 (libgcc_cmp_return_mode): Likewise.
18866 (libgcc_shift_count_mode): Likewise.
18867 (unwind_word_mode): Likewise.
18868 (addr_space.pointer_mode): Likewise.
18869 (addr_space.address_mode): Likewise.
18870 * doc/tm.texi: Regenerate.
18871 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18873 (do_jump): Use scalar_int_mode for local variables.
18874 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18875 rather than a machine_mode.
18876 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18877 (scompare_loc_descriptor_wide): Likewise.
18878 (scompare_loc_descriptor_narrow): Likewise.
18879 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18881 * except.c (sjlj_emit_dispatch_table): Likewise.
18882 (expand_builtin_eh_copy_values): Likewise.
18883 * explow.c (convert_memory_address_addr_space_1): Likewise.
18884 Take a scalar_int_mode rather than a machine_mode.
18885 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18886 than a machine_mode.
18887 (memory_address_addr_space): Use scalar_int_mode for local variables.
18888 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18889 rather than a machine_mode.
18890 * expmed.c (mask_rtx): Likewise.
18891 (init_expmed_one_conv): Likewise.
18892 (expand_mult_highpart_adjust): Likewise.
18893 (extract_high_half): Likewise.
18894 (expmed_mult_highpart_optab): Likewise.
18895 (expmed_mult_highpart): Likewise.
18896 (expand_smod_pow2): Likewise.
18897 (expand_sdiv_pow2): Likewise.
18898 (emit_store_flag_int): Likewise.
18899 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18901 (extract_low_bits): Likewise.
18902 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18904 * expr.c (pieces_addr::adjust): Likewise.
18905 (can_store_by_pieces): Likewise.
18906 (store_by_pieces): Likewise.
18907 (clear_by_pieces_1): Likewise.
18908 (expand_expr_addr_expr_1): Likewise.
18909 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18910 (expand_expr_real_1): Likewise.
18911 (try_casesi): Likewise.
18912 * final.c (shorten_branches): Likewise.
18913 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18914 type of "mode" to machine_mode.
18915 * internal-fn.c (expand_arith_overflow_result_store): Take a
18916 scalar_int_mode rather than a machine_mode.
18917 (expand_mul_overflow): Use scalar_int_mode for local variables.
18918 * loop-doloop.c (doloop_modify): Likewise.
18919 (doloop_optimize): Likewise.
18920 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18921 than a machine_mode.
18922 (expand_doubleword_shift_condmove): Likewise.
18923 (expand_doubleword_shift): Likewise.
18924 (expand_doubleword_clz): Likewise.
18925 (expand_doubleword_popcount): Likewise.
18926 (expand_doubleword_parity): Likewise.
18927 (expand_absneg_bit): Use scalar_int_mode for local variables.
18928 (prepare_float_lib_cmp): Likewise.
18929 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18930 rather than a machine_mode.
18931 (convert_memory_address_addr_space): Likewise.
18932 (get_mode_bounds): Likewise.
18933 (get_address_mode): Return a scalar_int_mode rather than a
18935 * rtlanal.c (get_address_mode): Likewise.
18936 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18937 than a machine_mode.
18938 * targhooks.c (default_mode_rep_extended): Likewise.
18939 (default_valid_pointer_mode): Likewise.
18940 (default_addr_space_valid_pointer_mode): Likewise.
18941 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18942 than a machine_mode.
18943 (default_libgcc_cmp_return_mode): Likewise.
18944 (default_libgcc_shift_count_mode): Likewise.
18945 (default_unwind_word_mode): Likewise.
18946 (default_addr_space_pointer_mode): Likewise.
18947 (default_addr_space_address_mode): Likewise.
18948 * targhooks.h (default_eh_return_filter_mode): Likewise.
18949 (default_libgcc_cmp_return_mode): Likewise.
18950 (default_libgcc_shift_count_mode): Likewise.
18951 (default_unwind_word_mode): Likewise.
18952 (default_addr_space_pointer_mode): Likewise.
18953 (default_addr_space_address_mode): Likewise.
18954 (default_mode_rep_extended): Take a scalar_int_mode rather than
18956 (default_valid_pointer_mode): Likewise.
18957 (default_addr_space_valid_pointer_mode): Likewise.
18958 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18960 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18961 rather than a machine_mode.
18962 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18963 for local variables.
18964 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18965 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18966 than a machine_mode.
18968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18969 Alan Hayward <alan.hayward@arm.com>
18970 David Sherwood <david.sherwood@arm.com>
18972 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18973 the mode argument to scalar_int_mode.
18974 (do_jump_by_parts_zero_rtx): Likewise.
18975 (do_jump_by_parts_equality_rtx): Likewise.
18976 (do_jump_by_parts_greater): Take a mode argument.
18977 (do_jump_by_parts_equality): Likewise.
18978 (do_jump_1): Update calls accordingly.
18980 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18981 Alan Hayward <alan.hayward@arm.com>
18982 David Sherwood <david.sherwood@arm.com>
18984 * is-a.h (safe_dyn_cast): New function.
18985 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18986 (jump_table_for_label): Likewise.
18987 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18988 instead of an rtx_insn *.
18989 (shorten_branches): Use dyn_cast instead of LABEL_P and
18990 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18991 rtx_jump_table_data::get_data_mode.
18992 (final_scan_insn): Likewise.
18994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18995 Alan Hayward <alan.hayward@arm.com>
18996 David Sherwood <david.sherwood@arm.com>
18998 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18999 trying to combine a full-register integer set with a subreg
19002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19003 Alan Hayward <alan.hayward@arm.com>
19004 David Sherwood <david.sherwood@arm.com>
19006 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
19007 that is always either address_mode or pointer_mode.
19009 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19010 Alan Hayward <alan.hayward@arm.com>
19011 David Sherwood <david.sherwood@arm.com>
19013 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
19014 when the two are known to be equal.
19016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19017 Alan Hayward <alan.hayward@arm.com>
19018 David Sherwood <david.sherwood@arm.com>
19020 * simplify-rtx.c (simplify_const_unary_operation): Use
19021 is_a <scalar_int_mode> instead of checking for a nonzero
19022 precision. Forcibly convert op_mode to a scalar_int_mode
19023 in that case. More clearly differentiate the operand and
19024 result modes and use the former when deciding what the value
19025 of a count-bits operation should be. Use is_int_mode instead
19026 of checking for a MODE_INT. Remove redundant check for whether
19027 this mode has a zero precision.
19029 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19030 Alan Hayward <alan.hayward@arm.com>
19031 David Sherwood <david.sherwood@arm.com>
19033 * optabs.c (widen_leading): Change the type of the mode argument
19034 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
19035 (widen_bswap): Likewise.
19036 (expand_parity): Likewise.
19037 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
19038 (expand_ffs): Likewise.
19039 (epand_unop): Check for scalar integer modes before calling the
19042 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19043 Alan Hayward <alan.hayward@arm.com>
19044 David Sherwood <david.sherwood@arm.com>
19046 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
19048 (expand_expr_real_1): Update call accordingly.
19050 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19051 Alan Hayward <alan.hayward@arm.com>
19052 David Sherwood <david.sherwood@arm.com>
19054 * expmed.c (store_bit_field_using_insv): Add op0_mode and
19055 value_mode arguments. Use scalar_int_mode internally.
19056 (store_bit_field_1): Rename the new integer mode from imode
19057 to op0_mode and use it instead of GET_MODE (op0). Update calls
19058 to store_split_bit_field, store_bit_field_using_insv and
19059 store_fixed_bit_field.
19060 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
19061 Use scalar_int_mode internally. Use a bit count rather than a mode
19062 when calculating the largest bit size for get_best_mode.
19063 Update calls to store_split_bit_field and store_fixed_bit_field_1.
19064 (store_fixed_bit_field_1): Add mode and value_mode arguments.
19065 Remove assertion that OP0 has a scalar integer mode.
19066 (store_split_bit_field): Add op0_mode and value_mode arguments.
19067 Update calls to extract_fixed_bit_field.
19068 (extract_bit_field_using_extv): Add an op0_mode argument.
19069 Use scalar_int_mode internally.
19070 (extract_bit_field_1): Rename the new integer mode from imode to
19071 op0_mode and use it instead of GET_MODE (op0). Update calls to
19072 extract_split_bit_field, extract_bit_field_using_extv and
19073 extract_fixed_bit_field.
19074 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
19075 to extract_split_bit_field and extract_fixed_bit_field_1.
19076 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
19077 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
19078 on the target mode.
19079 (extract_split_bit_field): Add an op0_mode argument. Update call
19080 to extract_fixed_bit_field.
19082 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19083 Alan Hayward <alan.hayward@arm.com>
19084 David Sherwood <david.sherwood@arm.com>
19086 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
19087 * explow.c (hard_function_value): Likewise.
19088 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
19089 convert_to_mode call outside the loop.
19090 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
19091 for the mode iterator. Require the mode specified by max_pieces
19093 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
19095 (copy_blkmode_to_reg): Likewise.
19096 (set_storage_via_setmem): Likewise.
19097 * optabs.c (prepare_cmp_insn): Likewise.
19098 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19099 * stor-layout.c (finish_bitfield_representative): Likewise.
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 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
19106 * expr.c (convert_move): Use them.
19107 (convert_modes): Likewise.
19108 (store_expr_with_bounds): Likewise.
19110 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19111 Alan Hayward <alan.hayward@arm.com>
19112 David Sherwood <david.sherwood@arm.com>
19114 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
19115 parameter for the mode of "x". Remove the "known_x", "known_mode"
19116 and "known_ret" arguments. Change the type of the mode argument
19117 to scalar_int_mode.
19118 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
19119 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
19120 (reg_num_sign_bit_copies_for_combine): Likewise.
19121 * rtlanal.c (nonzero_bits1): Likewise.
19122 (num_sign_bit_copies1): Likewise.
19123 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
19124 (reg_num_sign_bit_copies_general): Likewise.
19125 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
19126 (reg_nonzero_bits_general): Likewise.
19128 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19129 Alan Hayward <alan.hayward@arm.com>
19130 David Sherwood <david.sherwood@arm.com>
19132 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
19133 than in subroutines. Return 1 for non-integer modes.
19134 (cached_num_sign_bit_copies): Change the type of the mode parameter
19135 to scalar_int_mode.
19136 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
19137 classes. Handle CONST_INT_P first and then check whether X also
19138 has a scalar integer mode. Check the same thing for inner registers
19139 of a SUBREG and for values that are being extended or truncated.
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 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
19146 in subroutines. Return the mode mask for non-integer modes.
19147 (cached_nonzero_bits): Change the type of the mode parameter
19148 to scalar_int_mode.
19149 (nonzero_bits1): Likewise. Remove early exit for other mode
19150 classes. Handle CONST_INT_P first and then check whether X
19151 also has a scalar integer mode.
19153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19154 Alan Hayward <alan.hayward@arm.com>
19155 David Sherwood <david.sherwood@arm.com>
19157 * expr.c (widest_int_mode_for_size): Make the comment match the code.
19158 Return a scalar_int_mode and assert that the size is greater than
19160 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
19161 (op_by_pieces_d::op_by_pieces_d): Likewise.
19162 (op_by_pieces_d::run): Likewise.
19163 (can_store_by_pieces): Likewise.
19165 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19166 Alan Hayward <alan.hayward@arm.com>
19167 David Sherwood <david.sherwood@arm.com>
19169 * combine.c (extract_left_shift): Add a mode argument and update
19171 (make_compound_operation_int): Change the type of the mode parameter
19172 to scalar_int_mode and update the call to extract_left_shift.
19174 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19175 Alan Hayward <alan.hayward@arm.com>
19176 David Sherwood <david.sherwood@arm.com>
19178 * combine.c (simplify_and_const_int): Change the type of the mode
19179 parameter to scalar_int_mode.
19180 (simplify_and_const_int_1): Likewise. Update recursive call.
19182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19183 Alan Hayward <alan.hayward@arm.com>
19184 David Sherwood <david.sherwood@arm.com>
19186 * combine.c (simplify_compare_const): Check that the mode is a
19187 scalar_int_mode (rather than VOIDmode) before testing its
19189 (simplify_comparison): Move COMPARISON_P handling out of the
19190 loop and restrict the latter part of the loop to scalar_int_modes.
19191 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
19192 and when considering SUBREG_REGs. Use is_int_mode instead of
19193 checking GET_MODE_CLASS against MODE_INT.
19195 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19196 Alan Hayward <alan.hayward@arm.com>
19197 David Sherwood <david.sherwood@arm.com>
19199 * combine.c (try_widen_shift_mode): Move check for equal modes to...
19200 (simplify_shift_const_1): ...here. Use scalar_int_mode for
19201 shift_unit_mode and for modes involved in scalar shifts.
19203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19204 Alan Hayward <alan.hayward@arm.com>
19205 David Sherwood <david.sherwood@arm.com>
19207 * combine.c (force_int_to_mode): New function, split out from...
19208 (force_to_mode): ...here. Keep xmode up-to-date and use it
19209 instead of GET_MODE (x).
19211 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19212 Alan Hayward <alan.hayward@arm.com>
19213 David Sherwood <david.sherwood@arm.com>
19215 * optabs-query.h (extraction_insn::struct_mode): Change type to
19216 opt_scalar_int_mode and update comment.
19217 (extraction_insn::field_mode): Change type to scalar_int_mode.
19218 (extraction_insn::pos_mode): Likewise.
19219 * combine.c (make_extraction): Update accordingly.
19220 * optabs-query.c (get_traditional_extraction_insn): Likewise.
19221 (get_optab_extraction_insn): Likewise.
19222 * recog.c (simplify_while_replacing): Likewise.
19223 * expmed.c (narrow_bit_field_mem): Change the type of the mode
19224 parameter to opt_scalar_int_mode.
19226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19227 Alan Hayward <alan.hayward@arm.com>
19228 David Sherwood <david.sherwood@arm.com>
19230 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
19231 to a scalar_int_mode instead of a machine_mode.
19232 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
19233 (get_best_mode): Return a boolean and use a pointer argument to store
19234 the selected mode. Replace the limit mode parameter with a bit limit.
19235 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
19236 for the values returned by bit_field_mode_iterator::next_mode.
19237 (store_bit_field): Update call to get_best_mode.
19238 (store_fixed_bit_field): Likewise.
19239 (extract_fixed_bit_field): Likewise.
19240 * expr.c (optimize_bitfield_assignment_op): Likewise.
19241 * fold-const.c (optimize_bit_field_compare): Likewise.
19242 (fold_truth_andor_1): Likewise.
19243 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
19244 Update for new type of m_mode.
19245 (get_best_mode): As above.
19247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19248 Alan Hayward <alan.hayward@arm.com>
19249 David Sherwood <david.sherwood@arm.com>
19251 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
19252 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
19253 (store_bit_field): Check is_a <scalar_int_mode> before calling
19254 strict_volatile_bitfield_p.
19255 (extract_bit_field): Likewise.
19257 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19258 Alan Hayward <alan.hayward@arm.com>
19259 David Sherwood <david.sherwood@arm.com>
19261 * target.def (cstore_mode): Return a scalar_int_mode.
19262 * doc/tm.texi: Regenerate.
19263 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
19264 * targhooks.h (default_cstore_mode): Likewise.
19265 * targhooks.c (default_cstore_mode): Likewise, using a forced
19267 * expmed.c (emit_cstore): Expect the target of the cstore to be
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 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
19276 (niter_desc): Likewise mode.
19277 (iv_analyze): Add a mode parameter.
19279 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
19280 and change its type to scalar_int_mode.
19281 * loop-iv.c: Update commentary at head of file.
19282 (iv_constant): Pass the mode paraeter before the rtx it describes
19283 and change its type to scalar_int_mode. Remove VOIDmode handling.
19284 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
19285 (iv_extend): Likewise.
19286 (shorten_into_mode): Likewise.
19287 (iv_add): Use scalar_int_mode.
19288 (iv_mult): Likewise.
19289 (iv_shift): Likewise.
19290 (canonicalize_iv_subregs): Likewise.
19291 (get_biv_step_1): Pass the outer_mode parameter before the rtx
19292 it describes and change its mode to scalar_int_mode. Also change
19293 the type of the returned inner_mode to scalar_int_mode.
19294 (get_biv_step): Likewise, turning outer_mode from a pointer
19295 into a direct parameter. Update call to get_biv_step_1.
19296 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
19297 iv_constant and get_biv_step.
19298 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
19299 and change its type to scalar_int_mode. Don't initialise iv->mode
19300 to VOIDmode and remove later checks for its still being VOIDmode.
19301 Update calls to iv_analyze_op and iv_analyze_expr. Check
19302 is_a <scalar_int_mode> when changing the mode under consideration.
19303 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
19304 Update call to iv_analyze_expr.
19305 (iv_analyze_op): Add a mode parameter. Reject subregs whose
19306 inner register is not also a scalar_int_mode. Update call to
19308 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
19309 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
19310 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
19311 separate mode class checks. Update calls to iv_analyze. Remove
19312 fix-up of VOIDmodes after iv_analyze_biv.
19313 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
19314 don't have a scalar_int_mode. Update call to biv_p.
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 * cfgexpand.c (convert_debug_memory_address): Use
19321 as_a <scalar_int_mode>.
19322 * combine.c (expand_compound_operation): Likewise.
19323 (make_extraction): Likewise.
19324 (change_zero_ext): Likewise.
19325 (simplify_comparison): Likewise.
19326 * cse.c (cse_insn): Likewise.
19327 * dwarf2out.c (minmax_loc_descriptor): Likewise.
19328 (mem_loc_descriptor): Likewise.
19329 (loc_descriptor): Likewise.
19330 * expmed.c (init_expmed_one_mode): Likewise.
19331 (synth_mult): Likewise.
19332 (emit_store_flag_1): Likewise.
19333 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
19334 of a comparison with size.
19335 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
19336 (reduce_to_bit_field_precision): Likewise.
19337 * function.c (expand_function_end): Likewise.
19338 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
19339 * loop-doloop.c (doloop_modify): Likewise.
19340 * optabs.c (expand_binop): Likewise.
19341 (expand_unop): Likewise.
19342 (expand_copysign_absneg): Likewise.
19343 (prepare_cmp_insn): Likewise.
19344 (maybe_legitimize_operand): Likewise.
19345 * recog.c (const_scalar_int_operand): Likewise.
19346 * rtlanal.c (get_address_mode): Likewise.
19347 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19348 (simplify_cond_clz_ctz): Likewise.
19349 * tree-nested.c (get_nl_goto_field): Likewise.
19350 * tree.c (build_vector_type_for_mode): Likewise.
19351 * var-tracking.c (use_narrower_mode): Likewise.
19353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19354 Alan Hayward <alan.hayward@arm.com>
19355 David Sherwood <david.sherwood@arm.com>
19357 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
19358 * builtins.c (expand_builtin_signbit): Use it.
19359 * cfgexpand.c (expand_debug_expr): Likewise.
19360 * dojump.c (do_jump): Likewise.
19361 (do_compare_and_jump): Likewise.
19362 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
19363 * expmed.c (make_tree): Likewise.
19364 * expr.c (expand_expr_real_2): Likewise.
19365 (expand_expr_real_1): Likewise.
19366 (try_casesi): Likewise.
19367 * fold-const-call.c (fold_const_call_ss): Likewise.
19368 * fold-const.c (unextend): Likewise.
19369 (extract_muldiv_1): Likewise.
19370 (fold_single_bit_test): Likewise.
19371 (native_encode_int): Likewise.
19372 (native_encode_string): Likewise.
19373 (native_interpret_int): Likewise.
19374 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
19375 * internal-fn.c (expand_addsub_overflow): Likewise.
19376 (expand_neg_overflow): Likewise.
19377 (expand_mul_overflow): Likewise.
19378 (expand_arith_overflow): Likewise.
19379 * match.pd: Likewise.
19380 * stor-layout.c (layout_type): Likewise.
19381 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
19382 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
19383 * tree-ssanames.c (get_range_info): Likewise.
19384 * tree-switch-conversion.c (array_value_type) Likewise.
19385 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
19386 (vect_recog_divmod_pattern): Likewise.
19387 (vect_recog_mixed_size_cond_pattern): Likewise.
19388 * tree-vrp.c (extract_range_basic): Likewise.
19389 (simplify_float_conversion_using_ranges): Likewise.
19390 * tree.c (int_fits_type_p): Likewise.
19391 * ubsan.c (instrument_bool_enum_load): Likewise.
19392 * varasm.c (mergeable_string_section): Likewise.
19393 (narrowing_initializer_constant_valid_p): Likewise.
19394 (output_constant): Likewise.
19396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19397 Alan Hayward <alan.hayward@arm.com>
19398 David Sherwood <david.sherwood@arm.com>
19400 * machmode.h (NARROWEST_INT_MODE): New macro.
19401 * expr.c (alignment_for_piecewise_move): Use it instead of
19402 GET_CLASS_NARROWEST_MODE (MODE_INT).
19403 (push_block): Likewise.
19404 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
19406 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19409 Alan Hayward <alan.hayward@arm.com>
19410 David Sherwood <david.sherwood@arm.com>
19412 * postreload.c (move2add_valid_value_p): Change the type of the
19413 mode parameter to scalar_int_mode.
19414 (move2add_use_add2_insn): Add a mode parameter and use it instead
19416 (move2add_use_add3_insn): Likewise.
19417 (reload_cse_move2add): Update accordingly.
19419 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19420 Alan Hayward <alan.hayward@arm.com>
19421 David Sherwood <david.sherwood@arm.com>
19423 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
19425 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
19426 * optabs.c (expand_unop): Likewise.
19428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19429 Alan Hayward <alan.hayward@arm.com>
19430 David Sherwood <david.sherwood@arm.com>
19432 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
19433 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
19434 (popcount_loc_descriptor): Likewise.
19435 (bswap_loc_descriptor): Likewise.
19436 (rotate_loc_descriptor): Likewise.
19437 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
19438 calling the functions above.
19440 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19441 Alan Hayward <alan.hayward@arm.com>
19442 David Sherwood <david.sherwood@arm.com>
19444 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
19446 (try_combine): Likewise.
19447 (simplify_if_then_else): Likewise.
19448 * cse.c (cse_insn): Likewise.
19449 * dwarf2out.c (mem_loc_descriptor): Likewise.
19450 * emit-rtl.c (gen_lowpart_common): Likewise.
19451 * simplify-rtx.c (simplify_truncation): Likewise.
19452 (simplify_binary_operation_1): Likewise.
19453 (simplify_const_relational_operation): Likewise.
19454 (simplify_ternary_operation): Likewise.
19455 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
19457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19458 Alan Hayward <alan.hayward@arm.com>
19459 David Sherwood <david.sherwood@arm.com>
19461 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
19462 * reload.c (push_reload): Likewise.
19463 (find_reloads): Likewise.
19465 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19466 Alan Hayward <alan.hayward@arm.com>
19467 David Sherwood <david.sherwood@arm.com>
19469 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
19470 (make_compound_operation_int): Likewise.
19471 (change_zero_ext): Likewise.
19472 * expr.c (convert_move): Likewise.
19473 (convert_modes): Likewise.
19474 * fwprop.c (forward_propagate_subreg): Likewise.
19475 * loop-iv.c (get_biv_step_1): Likewise.
19476 * optabs.c (widen_operand): Likewise.
19477 * postreload.c (move2add_valid_value_p): Likewise.
19478 * recog.c (simplify_while_replacing): Likewise.
19479 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19480 (simplify_binary_operation_1): Likewise. Remove redundant
19481 mode equality check.
19483 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19484 Alan Hayward <alan.hayward@arm.com>
19485 David Sherwood <david.sherwood@arm.com>
19487 * combine.c (combine_simplify_rtx): Add checks for
19488 is_a <scalar_int_mode>.
19489 (simplify_if_then_else): Likewise.
19490 (make_field_assignment): Likewise.
19491 (simplify_comparison): Likewise.
19492 * ifcvt.c (noce_try_bitop): Likewise.
19493 * loop-invariant.c (canonicalize_address_mult): Likewise.
19494 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19496 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19497 Alan Hayward <alan.hayward@arm.com>
19498 David Sherwood <david.sherwood@arm.com>
19500 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
19501 is_a <scalar_int_mode> instead of != BLKmode.
19503 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19504 Alan Hayward <alan.hayward@arm.com>
19505 David Sherwood <david.sherwood@arm.com>
19507 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
19508 instead of != VOIDmode.
19509 * combine.c (if_then_else_cond): Likewise.
19510 (change_zero_ext): Likewise.
19511 * dwarf2out.c (mem_loc_descriptor): Likewise.
19512 (loc_descriptor): Likewise.
19513 * rtlanal.c (canonicalize_condition): Likewise.
19514 * simplify-rtx.c (simplify_relational_operation_1): 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 * simplify-rtx.c (simplify_binary_operation_1): Use
19521 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
19523 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19524 Alan Hayward <alan.hayward@arm.com>
19525 David Sherwood <david.sherwood@arm.com>
19527 * wide-int.h (int_traits<unsigned char>) New class.
19528 (int_traits<unsigned short>) Likewise.
19529 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
19530 Use GET_MODE_UNIT_PRECISION and remove redundant test for
19532 * combine.c (set_nonzero_bits_and_sign_copies): Use
19533 is_a <scalar_int_mode>.
19534 (find_split_point): Likewise.
19535 (combine_simplify_rtx): Likewise.
19536 (simplify_logical): Likewise.
19537 (expand_compound_operation): Likewise.
19538 (expand_field_assignment): Likewise.
19539 (make_compound_operation): Likewise.
19540 (extended_count): Likewise.
19541 (change_zero_ext): Likewise.
19542 (simplify_comparison): Likewise.
19543 * dwarf2out.c (scompare_loc_descriptor): Likewise.
19544 (ucompare_loc_descriptor): Likewise.
19545 (minmax_loc_descriptor): Likewise.
19546 (mem_loc_descriptor): Likewise.
19547 (loc_descriptor): Likewise.
19548 * expmed.c (init_expmed_one_mode): Likewise.
19549 * lra-constraints.c (lra_constraint_offset): Likewise.
19550 * optabs.c (prepare_libcall_arg): Likewise.
19551 * postreload.c (move2add_note_store): Likewise.
19552 * reload.c (operands_match_p): Likewise.
19553 * rtl.h (load_extend_op): Likewise.
19554 * rtlhooks.c (gen_lowpart_general): Likewise.
19555 * simplify-rtx.c (simplify_truncation): Likewise.
19556 (simplify_unary_operation_1): Likewise.
19557 (simplify_binary_operation_1): Likewise.
19558 (simplify_const_binary_operation): Likewise.
19559 (simplify_const_relational_operation): Likewise.
19560 (simplify_subreg): Likewise.
19561 * stor-layout.c (bitwise_mode_for_mode): Likewise.
19562 * var-tracking.c (adjust_mems): Likewise.
19563 (prepare_call_arguments): Likewise.
19565 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19566 Alan Hayward <alan.hayward@arm.com>
19567 David Sherwood <david.sherwood@arm.com>
19569 * machmode.h (is_int_mode): New fuction.
19570 * combine.c (find_split_point): Use it.
19571 (combine_simplify_rtx): Likewise.
19572 (simplify_if_then_else): Likewise.
19573 (simplify_set): Likewise.
19574 (simplify_shift_const_1): Likewise.
19575 (simplify_comparison): Likewise.
19576 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
19577 * cse.c (notreg_cost): Likewise.
19578 (cse_insn): Likewise.
19579 * cselib.c (cselib_lookup_1): Likewise.
19580 * dojump.c (do_jump_1): Likewise.
19581 (do_compare_rtx_and_jump): Likewise.
19582 * dse.c (get_call_args): Likewise.
19583 * dwarf2out.c (rtl_for_decl_init): Likewise.
19584 (native_encode_initializer): Likewise.
19585 * expmed.c (emit_store_flag_1): Likewise.
19586 (emit_store_flag): Likewise.
19587 * expr.c (convert_modes): Likewise.
19588 (store_field): Likewise.
19589 (expand_expr_real_1): Likewise.
19590 * fold-const.c (fold_read_from_constant_string): Likewise.
19591 * gimple-ssa-sprintf.c (get_format_string): Likewise.
19592 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
19593 * optabs.c (expand_binop): Likewise.
19594 (expand_unop): Likewise.
19595 (expand_abs_nojump): Likewise.
19596 (expand_one_cmpl_abs_nojump): Likewise.
19597 * simplify-rtx.c (mode_signbit_p): Likewise.
19598 (val_signbit_p): Likewise.
19599 (val_signbit_known_set_p): Likewise.
19600 (val_signbit_known_clear_p): Likewise.
19601 (simplify_relational_operation_1): Likewise.
19602 * tree.c (vector_type_mode): Likewise.
19604 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19605 Alan Hayward <alan.hayward@arm.com>
19606 David Sherwood <david.sherwood@arm.com>
19608 * machmode.h (smallest_mode_for_size): Fix formatting.
19609 (smallest_int_mode_for_size): New function.
19610 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
19611 instead of smallest_mode_for_size.
19612 * combine.c (make_extraction): Likewise.
19613 * config/arc/arc.c (arc_expand_movmem): Likewise.
19614 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
19615 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
19616 * config/s390/s390.c (s390_expand_insv): Likewise.
19617 * config/sparc/sparc.c (assign_int_registers): Likewise.
19618 * config/spu/spu.c (spu_function_value): Likewise.
19619 (spu_function_arg): Likewise.
19620 * coverage.c (get_gcov_type): Likewise.
19621 (get_gcov_unsigned_t): Likewise.
19622 * dse.c (find_shift_sequence): Likewise.
19623 * expmed.c (store_bit_field_1): Likewise.
19624 * expr.c (convert_move): Likewise.
19625 (store_field): Likewise.
19626 * internal-fn.c (expand_arith_overflow): Likewise.
19627 * optabs-query.c (get_best_extraction_insn): Likewise.
19628 * optabs.c (expand_twoval_binop_libfunc): Likewise.
19629 * stor-layout.c (layout_type): Likewise.
19630 (initialize_sizetypes): Likewise.
19631 * targhooks.c (default_get_mask_mode): Likewise.
19632 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
19634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19635 Alan Hayward <alan.hayward@arm.com>
19636 David Sherwood <david.sherwood@arm.com>
19638 * machmode.h (opt_mode::else_blk): New function.
19639 (int_mode_for_mode): Declare.
19640 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
19641 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
19643 * cfgexpand.c (expand_debug_expr): Likewise.
19644 * combine.c (gen_lowpart_or_truncate): Likewise.
19645 (gen_lowpart_for_combine): Likewise.
19646 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
19647 * config/avr/avr.c (avr_to_int_mode): Likewise.
19648 (avr_out_plus_1): Likewise.
19649 (avr_out_plus): Likewise.
19650 (avr_out_round): Likewise.
19651 * config/i386/i386.c (ix86_split_to_parts): Likewise.
19652 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
19653 (s390_expand_vcond): Likewise.
19654 * config/spu/spu.c (spu_split_immediate): Likewise.
19655 (spu_expand_mov): Likewise.
19656 * dse.c (get_stored_val): Likewise.
19657 * expmed.c (store_bit_field_1): Likewise.
19658 (convert_extracted_bit_field): Use int_mode_for_mode instead of
19660 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
19661 (extract_low_bits): Likewise.
19662 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
19663 handling rather than repeating the check.
19664 (emit_group_store): Likewise.
19665 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
19666 * optabs.c (expand_absneg_bit): Likewise.
19667 (expand_copysign_absneg): Likewise.
19668 (expand_copysign_bit): Likewise.
19669 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
19670 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19671 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
19672 * var-tracking.c (prepare_call_arguments): Likewise.
19673 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
19674 int_mode_for_mode instead of mode_for_size.
19675 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
19677 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19678 Alan Hayward <alan.hayward@arm.com>
19679 David Sherwood <david.sherwood@arm.com>
19681 * machmode.h (int_mode_for_size): New function.
19682 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
19683 instead of mode_for_size.
19684 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
19686 * combine.c (expand_field_assignment): Use int_mode_for_size
19687 instead of mode_for_size.
19688 (make_extraction): Likewise.
19689 (simplify_shift_const_1): Likewise.
19690 (simplify_comparison): Likewise.
19691 * dojump.c (do_jump): Likewise.
19692 * dwarf2out.c (mem_loc_descriptor): Likewise.
19693 * emit-rtl.c (init_derived_machine_modes): Likewise.
19694 * expmed.c (flip_storage_order): Likewise.
19695 (convert_extracted_bit_field): Likewise.
19696 * expr.c (copy_blkmode_from_reg): Likewise.
19697 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
19698 * internal-fn.c (expand_mul_overflow): Likewise.
19699 * lower-subreg.c (simple_move): Likewise.
19700 * optabs-libfuncs.c (init_optabs): Likewise.
19701 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19702 * tree.c (vector_type_mode): Likewise.
19703 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
19704 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
19705 * tree-vect-generic.c (expand_vector_parallel): Likewise.
19706 * tree-vect-stmts.c (vectorizable_load): Likewise.
19707 (vectorizable_store): Likewise.
19709 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19710 Alan Hayward <alan.hayward@arm.com>
19711 David Sherwood <david.sherwood@arm.com>
19713 * coretypes.h (pod_mode): New type.
19714 (scalar_int_mode_pod): New typedef.
19715 * machmode.h (pod_mode): New class.
19716 (int_n_data_t::m): Change type to scalar_int_mode_pod.
19717 * genmodes.c (emit_mode_int_n): Update accordingly.
19718 * lower-subreg.h (target_lower_subreg): Change type to
19719 scalar_int_mode_pod.
19720 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
19721 scalar_int_mode_pod.
19723 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19724 Alan Hayward <alan.hayward@arm.com>
19725 David Sherwood <david.sherwood@arm.com>
19727 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
19728 machine_mode to scalar_int_mode.
19729 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
19730 (rs6000_option_override_internal): Remove cast to int.
19731 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
19732 machine_mode to scalar_int_mode.
19733 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
19734 (rs6000_option_override_internal): Remove cast to int.
19735 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
19736 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
19738 * config/s390/s390.c (s390_expand_builtin): Likewise.
19739 * coretypes.h (scalar_int_mode): New type.
19740 (opt_scalar_int_mode): New typedef.
19741 * machmode.h (scalar_int_mode): New class.
19742 (scalar_int_mode::includes_p): New function.
19743 (byte_mode): Change type to scalar_int_mode.
19744 (word_mode): Likewise.
19745 (ptr_mode): Likewise.
19746 * emit-rtl.c (byte_mode): Likewise.
19747 (word_mode): Likewise.
19748 (ptr_mode): Likewise.
19749 (init_derived_machine_modes): Update accordingly.
19750 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
19751 and MODE_PARTIAL_INT.
19752 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
19753 opt_scalar_int_mode.
19755 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19756 Alan Hayward <alan.hayward@arm.com>
19757 David Sherwood <david.sherwood@arm.com>
19759 * target.def (libgcc_floating_mode_supported_p): Take a
19761 * doc/tm.texi: Regenerate.
19762 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
19764 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19765 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
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 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
19773 * doc/tm.texi: Regenerate.
19774 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
19775 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
19776 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
19777 * targhooks.h (default_floatn_mode): Likewise.
19778 * targhooks.c (default_floatn_mode): Likewise.
19779 * tree.c (build_common_tree_nodes): Update accordingly.
19781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19782 Alan Hayward <alan.hayward@arm.com>
19783 David Sherwood <david.sherwood@arm.com>
19785 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
19786 (mode_iterator::iterate_p): Likewise.
19787 (mode_iterator::get_wider): Likewise.
19788 * expr.c (init_expr_target): Use opt_scalar_float_mode.
19790 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19791 Alan Hayward <alan.hayward@arm.com>
19792 David Sherwood <david.sherwood@arm.com>
19794 * coretypes.h (opt_scalar_float_mode): New typedef.
19795 * machmode.h (float_mode_for_size): New function.
19796 * emit-rtl.c (double_mode): Delete.
19797 (init_emit_once): Use float_mode_for_size.
19798 * stor-layout.c (layout_type): Likewise.
19799 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19801 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19802 Alan Hayward <alan.hayward@arm.com>
19803 David Sherwood <david.sherwood@arm.com>
19805 * output.h (assemble_real): Take a scalar_float_mode.
19806 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19807 * config/arm/arm.md (consttable_4): Likewise.
19808 (consttable_8): Likewise.
19809 (consttable_16): Likewise.
19810 * config/mips/mips.md (consttable_float): Likewise.
19811 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19812 * varasm.c (assemble_real): Take a scalar_float_mode.
19813 (output_constant_pool_2): Update accordingly.
19814 (output_constant): Likewise.
19816 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19817 Alan Hayward <alan.hayward@arm.com>
19818 David Sherwood <david.sherwood@arm.com>
19820 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19821 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19822 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19823 (native_encode_real): Likewise.
19824 (native_interpret_real): Likewise.
19825 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19826 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19828 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19829 Alan Hayward <alan.hayward@arm.com>
19830 David Sherwood <david.sherwood@arm.com>
19832 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19833 <scalar_float_mode>. Simplify.
19834 (gen_extend_conv_libfunc): Likewise.
19836 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19837 Alan Hayward <alan.hayward@arm.com>
19838 David Sherwood <david.sherwood@arm.com>
19840 * coretypes.h (scalar_float_mode): New type.
19841 * machmode.h (mode_traits::from_int): Use machine_mode if
19842 USE_ENUM_MODES is defined.
19843 (is_a): New function.
19845 (dyn_cast): Likewise.
19846 (scalar_float_mode): New class.
19847 (scalar_float_mode::includes_p): New function.
19848 (is_float_mode): Likewise.
19849 * gdbhooks.py (MachineModePrinter): New class.
19850 (build_pretty_printer): Use it for scalar_float_mode.
19851 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19852 (format_helper::format_helper): Turn into a template.
19853 * genmodes.c (get_mode_class): New function.
19854 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19855 or machine_mode if none.
19856 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19857 as_a <scalar_float_mode>.
19858 * dwarf2out.c (mem_loc_descriptor): Likewise.
19859 (insert_float): Likewise.
19860 (add_const_value_attribute): Likewise.
19861 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19862 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19863 (expand_unop): Update accordingly.
19864 (expand_abs_nojump): Likewise.
19865 (expand_copysign_absneg): Take a scalar_float_mode.
19866 (expand_copysign_bit): Likewise.
19867 (expand_copysign): Update accordingly.
19869 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19870 Alan Hayward <alan.hayward@arm.com>
19871 David Sherwood <david.sherwood@arm.com>
19873 * coretypes.h (opt_mode): New class.
19874 * machmode.h (opt_mode): Likewise.
19875 (opt_mode::else_void): New function.
19876 (opt_mode::require): Likewise.
19877 (opt_mode::exists): Likewise.
19878 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19879 (GET_MODE_2XWIDER_MODE): Likewise.
19880 (mode_iterator::get_wider): Update accordingly.
19881 (mode_iterator::get_2xwider): Likewise.
19882 (mode_iterator::get_known_wider): Likewise, turning into a template.
19883 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19884 forcing a wider mode to exist.
19885 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19886 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19887 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19888 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19889 * lower-subreg.c (init_lower_subreg): Likewise.
19890 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19891 on the final iteration.
19892 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19893 a wider mode exists before asking for a move pattern.
19894 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19895 forcing a wider mode to exist.
19896 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19897 returning false if no such mode exists.
19898 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19899 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19900 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19901 Avoid checking for a MODE_INT if we already know the mode is not a
19903 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19904 forcing a wider mode to exist.
19905 (expmed_mult_highpart_optab): Likewise.
19906 (expmed_mult_highpart): Likewise.
19907 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19909 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19910 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19911 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19912 * internal-fn.c (expand_mul_overflow): Update use of
19913 GET_MODE_2XWIDER_MODE.
19914 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19915 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19916 GET_MODE_WIDER_MODE.
19917 (convert_plusminus_to_widen): Likewise.
19918 * tree-switch-conversion.c (array_value_type): Likewise.
19919 * var-tracking.c (emit_note_insn_var_location): Likewise.
19920 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19921 Return false inside rather than outside the loop if no wider mode
19923 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19924 and GET_MODE_2XWIDER_MODE
19925 (can_compare_p): Use else_void.
19926 * gdbhooks.py (OptMachineModePrinter): New class.
19927 (build_pretty_printer): Use it for opt_mode.
19929 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19930 Alan Hayward <alan.hayward@arm.com>
19931 David Sherwood <david.sherwood@arm.com>
19933 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19934 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19937 Alan Hayward <alan.hayward@arm.com>
19938 David Sherwood <david.sherwood@arm.com>
19940 * machmode.h (mode_traits): New structure.
19941 (get_narrowest_mode): New function.
19942 (mode_iterator::start): Likewise.
19943 (mode_iterator::iterate_p): Likewise.
19944 (mode_iterator::get_wider): Likewise.
19945 (mode_iterator::get_known_wider): Likewise.
19946 (mode_iterator::get_2xwider): Likewise.
19947 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19948 (FOR_EACH_MODE): Likewise.
19949 (FOR_EACH_MODE_FROM): Likewise.
19950 (FOR_EACH_MODE_UNTIL): Likewise.
19951 (FOR_EACH_WIDER_MODE): Likewise.
19952 (FOR_EACH_2XWIDER_MODE): Likewise.
19953 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19954 * combine.c (simplify_comparison): Likewise
19955 * config/i386/i386.c (type_natural_mode): Likewise.
19956 * cse.c (cse_insn): Likewise.
19957 * dse.c (find_shift_sequence): Likewise.
19958 * emit-rtl.c (init_derived_machine_modes): Likewise.
19959 (init_emit_once): Likewise.
19960 * explow.c (hard_function_value): Likewise.
19961 * expmed.c (extract_fixed_bit_field_1): Likewise.
19962 (extract_bit_field_1): Likewise.
19963 (expand_divmod): Likewise.
19964 (emit_store_flag_1): Likewise.
19965 * expr.c (init_expr_target): Likewise.
19966 (convert_move): Likewise.
19967 (alignment_for_piecewise_move): Likewise.
19968 (widest_int_mode_for_size): Likewise.
19969 (emit_block_move_via_movmem): Likewise.
19970 (copy_blkmode_to_reg): Likewise.
19971 (set_storage_via_setmem): Likewise.
19972 (compress_float_constant): Likewise.
19973 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19974 * optabs-query.c (get_best_extraction_insn): Likewise.
19975 * optabs.c (expand_binop): Likewise.
19976 (expand_twoval_unop): Likewise.
19977 (expand_twoval_binop): Likewise.
19978 (widen_leading): Likewise.
19979 (widen_bswap): Likewise.
19980 (expand_parity): Likewise.
19981 (expand_unop): Likewise.
19982 (prepare_cmp_insn): Likewise.
19983 (prepare_float_lib_cmp): Likewise.
19984 (expand_float): Likewise.
19985 (expand_fix): Likewise.
19986 (expand_sfix_optab): Likewise.
19987 * postreload.c (move2add_use_add2_insn): Likewise.
19988 * reg-stack.c (reg_to_stack): Likewise.
19989 * reginfo.c (choose_hard_reg_mode): Likewise.
19990 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19991 * stor-layout.c (mode_for_size): Likewise.
19992 (smallest_mode_for_size): Likewise.
19993 (mode_for_vector): Likewise.
19994 (finish_bitfield_representative): Likewise.
19995 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19996 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19997 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19998 * var-tracking.c (prepare_call_arguments): Likewise.
20000 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20001 Alan Hayward <alan.hayward@arm.com>
20002 David Sherwood <david.sherwood@arm.com>
20004 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
20005 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
20006 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
20007 * machmode.h (mode_size): Move earlier in file.
20008 (mode_precision): Likewise.
20009 (mode_inner): Likewise.
20010 (mode_nunits): Likewise.
20011 (mode_unit_size): Likewise.
20012 (unit_unit_precision): Likewise.
20013 (mode_wider): Likewise.
20014 (mode_2xwider): Likewise.
20015 (machine_mode): New class.
20016 (mode_to_bytes): New function.
20017 (mode_to_bits): Likewise.
20018 (mode_to_precision): Likewise.
20019 (mode_to_inner): Likewise.
20020 (mode_to_unit_size): Likewise.
20021 (mode_to_unit_precision): Likewise.
20022 (mode_to_nunits): Likewise.
20023 (GET_MODE_SIZE): Use mode_to_bytes.
20024 (GET_MODE_BITSIZE): Use mode_to_bits.
20025 (GET_MODE_PRECISION): Use mode_to_precision.
20026 (GET_MODE_INNER): Use mode_to_inner.
20027 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
20028 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
20029 (GET_MODE_NUNITS): Use mode_to_nunits.
20030 * system.h (ALWAYS_INLINE): New macro.
20031 * config/powerpcspe/powerpcspe-c.c
20032 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
20033 int for arg1_mode and arg2_mode.
20035 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20036 Alan Hayward <alan.hayward@arm.com>
20037 David Sherwood <david.sherwood@arm.com>
20039 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
20040 Prefix mode names with E_ in case statements.
20041 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20042 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
20043 (aarch64_split_simd_move): Likewise.
20044 (aarch64_gen_storewb_pair): Likewise.
20045 (aarch64_gen_loadwb_pair): Likewise.
20046 (aarch64_gen_store_pair): Likewise.
20047 (aarch64_gen_load_pair): Likewise.
20048 (aarch64_get_condition_code_1): Likewise.
20049 (aarch64_constant_pool_reload_icode): Likewise.
20050 (get_rsqrte_type): Likewise.
20051 (get_rsqrts_type): Likewise.
20052 (get_recpe_type): Likewise.
20053 (get_recps_type): Likewise.
20054 (aarch64_gimplify_va_arg_expr): Likewise.
20055 (aarch64_simd_container_mode): Likewise.
20056 (aarch64_emit_load_exclusive): Likewise.
20057 (aarch64_emit_store_exclusive): Likewise.
20058 (aarch64_expand_compare_and_swap): Likewise.
20059 (aarch64_gen_atomic_cas): Likewise.
20060 (aarch64_emit_bic): Likewise.
20061 (aarch64_emit_atomic_swap): Likewise.
20062 (aarch64_emit_atomic_load_op): Likewise.
20063 (aarch64_evpc_trn): Likewise.
20064 (aarch64_evpc_uzp): Likewise.
20065 (aarch64_evpc_zip): Likewise.
20066 (aarch64_evpc_ext): Likewise.
20067 (aarch64_evpc_rev): Likewise.
20068 (aarch64_evpc_dup): Likewise.
20069 (aarch64_gen_ccmp_first): Likewise.
20070 (aarch64_gen_ccmp_next): Likewise.
20071 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
20072 (alpha_emit_xfloating_libcall): Likewise.
20073 (emit_insxl): Likewise.
20074 (alpha_arg_type): Likewise.
20075 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
20076 (arc_preferred_simd_mode): Likewise.
20077 (arc_secondary_reload): Likewise.
20078 (get_arc_condition_code): Likewise.
20079 (arc_print_operand): Likewise.
20080 (arc_legitimate_constant_p): Likewise.
20081 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20082 * config/arc/arc.md (casesi_load): Likewise.
20083 (casesi_compact_jump): Likewise.
20084 * config/arc/predicates.md (proper_comparison_operator): Likewise.
20085 (cc_use_register): Likewise.
20086 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20087 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
20088 (arm_init_iwmmxt_builtins): Likewise.
20089 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
20090 (neon_expand_vector_init): Likewise.
20091 (arm_attr_length_move_neon): Likewise.
20092 (maybe_get_arm_condition_code): Likewise.
20093 (arm_emit_vector_const): Likewise.
20094 (arm_preferred_simd_mode): Likewise.
20095 (arm_output_iwmmxt_tinsr): Likewise.
20096 (thumb1_output_casesi): Likewise.
20097 (thumb2_output_casesi): Likewise.
20098 (arm_emit_load_exclusive): Likewise.
20099 (arm_emit_store_exclusive): Likewise.
20100 (arm_expand_compare_and_swap): Likewise.
20101 (arm_evpc_neon_vuzp): Likewise.
20102 (arm_evpc_neon_vzip): Likewise.
20103 (arm_evpc_neon_vrev): Likewise.
20104 (arm_evpc_neon_vtrn): Likewise.
20105 (arm_evpc_neon_vext): Likewise.
20106 (arm_validize_comparison): Likewise.
20107 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
20108 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
20109 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
20110 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
20111 (c6x_preferred_simd_mode): Likewise.
20112 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
20113 (epiphany_rtx_costs): Likewise.
20114 * config/epiphany/predicates.md (proper_comparison_operator):
20116 * config/frv/frv.c (condexec_memory_operand): Likewise.
20117 (frv_emit_move): Likewise.
20118 (output_move_single): Likewise.
20119 (output_condmove_single): Likewise.
20120 (frv_hard_regno_mode_ok): Likewise.
20121 (frv_matching_accg_mode): Likewise.
20122 * config/h8300/h8300.c (split_adds_subs): Likewise.
20123 (h8300_rtx_costs): Likewise.
20124 (h8300_print_operand): Likewise.
20125 (compute_mov_length): Likewise.
20126 (output_logical_op): Likewise.
20127 (compute_logical_op_length): Likewise.
20128 (compute_logical_op_cc): Likewise.
20129 (h8300_shift_needs_scratch_p): Likewise.
20130 (output_a_shift): Likewise.
20131 (compute_a_shift_length): Likewise.
20132 (compute_a_shift_cc): Likewise.
20133 (expand_a_rotate): Likewise.
20134 (output_a_rotate): Likewise.
20135 * config/i386/i386.c (classify_argument): Likewise.
20136 (function_arg_advance_32): Likewise.
20137 (function_arg_32): Likewise.
20138 (function_arg_64): Likewise.
20139 (function_value_64): Likewise.
20140 (ix86_gimplify_va_arg): Likewise.
20141 (ix86_legitimate_constant_p): Likewise.
20142 (put_condition_code): Likewise.
20143 (split_double_mode): Likewise.
20144 (ix86_avx256_split_vector_move_misalign): Likewise.
20145 (ix86_expand_vector_logical_operator): Likewise.
20146 (ix86_split_idivmod): Likewise.
20147 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
20148 (ix86_build_const_vector): Likewise.
20149 (ix86_build_signbit_mask): Likewise.
20150 (ix86_match_ccmode): Likewise.
20151 (ix86_cc_modes_compatible): Likewise.
20152 (ix86_expand_branch): Likewise.
20153 (ix86_expand_sse_cmp): Likewise.
20154 (ix86_expand_sse_movcc): Likewise.
20155 (ix86_expand_int_sse_cmp): Likewise.
20156 (ix86_expand_vec_perm_vpermi2): Likewise.
20157 (ix86_expand_vec_perm): Likewise.
20158 (ix86_expand_sse_unpack): Likewise.
20159 (ix86_expand_int_addcc): Likewise.
20160 (ix86_split_to_parts): Likewise.
20161 (ix86_vectorize_builtin_gather): Likewise.
20162 (ix86_vectorize_builtin_scatter): Likewise.
20163 (avx_vpermilp_parallel): Likewise.
20164 (inline_memory_move_cost): Likewise.
20165 (ix86_tieable_integer_mode_p): Likewise.
20166 (x86_maybe_negate_const_int): Likewise.
20167 (ix86_expand_vector_init_duplicate): Likewise.
20168 (ix86_expand_vector_init_one_nonzero): Likewise.
20169 (ix86_expand_vector_init_one_var): Likewise.
20170 (ix86_expand_vector_init_concat): Likewise.
20171 (ix86_expand_vector_init_interleave): Likewise.
20172 (ix86_expand_vector_init_general): Likewise.
20173 (ix86_expand_vector_set): Likewise.
20174 (ix86_expand_vector_extract): Likewise.
20175 (emit_reduc_half): Likewise.
20176 (ix86_emit_i387_round): Likewise.
20177 (ix86_mangle_type): Likewise.
20178 (ix86_expand_round_sse4): Likewise.
20179 (expand_vec_perm_blend): Likewise.
20180 (canonicalize_vector_int_perm): Likewise.
20181 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
20182 (expand_vec_perm_1): Likewise.
20183 (expand_vec_perm_interleave3): Likewise.
20184 (expand_vec_perm_even_odd_pack): Likewise.
20185 (expand_vec_perm_even_odd_1): Likewise.
20186 (expand_vec_perm_broadcast_1): Likewise.
20187 (ix86_vectorize_vec_perm_const_ok): Likewise.
20188 (ix86_expand_vecop_qihi): Likewise.
20189 (ix86_expand_mul_widen_hilo): Likewise.
20190 (ix86_expand_sse2_abs): Likewise.
20191 (ix86_expand_pextr): Likewise.
20192 (ix86_expand_pinsr): Likewise.
20193 (ix86_preferred_simd_mode): Likewise.
20194 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
20195 * config/i386/sse.md (*andnot<mode>3): Likewise.
20196 (<mask_codefor><code><mode>3<mask_name>): Likewise.
20197 (*<code><mode>3): Likewise.
20198 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
20199 (ia64_expand_atomic_op): Likewise.
20200 (ia64_arg_type): Likewise.
20201 (ia64_mode_to_int): Likewise.
20202 (ia64_scalar_mode_supported_p): Likewise.
20203 (ia64_vector_mode_supported_p): Likewise.
20204 (expand_vec_perm_broadcast): Likewise.
20205 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
20206 (iq2000_function_arg_advance): Likewise.
20207 (iq2000_function_arg): Likewise.
20208 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
20209 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
20210 (m68k_libcall_value): Likewise.
20211 (m68k_function_value): Likewise.
20212 (sched_attr_op_type): Likewise.
20213 * config/mcore/mcore.c (mcore_output_move): Likewise.
20214 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
20216 (microblaze_function_arg): Likewise.
20217 * config/mips/mips.c (mips16_build_call_stub): Likewise.
20218 (mips_print_operand): Likewise.
20219 (mips_mode_ok_for_mov_fmt_p): Likewise.
20220 (mips_vector_mode_supported_p): Likewise.
20221 (mips_preferred_simd_mode): Likewise.
20222 (mips_expand_vpc_loongson_even_odd): Likewise.
20223 (mips_expand_vec_unpack): Likewise.
20224 (mips_expand_vi_broadcast): Likewise.
20225 (mips_expand_vector_init): Likewise.
20226 (mips_expand_vec_reduc): Likewise.
20227 (mips_expand_msa_cmp): Likewise.
20228 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
20229 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
20230 (cc_flags_for_mode): Likewise.
20231 * config/msp430/msp430.c (msp430_print_operand): Likewise.
20232 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
20233 (nds32_output_casesi_pc_relative): Likewise.
20234 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20235 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
20236 (nvptx_gen_unpack): Likewise.
20237 (nvptx_gen_pack): Likewise.
20238 (nvptx_gen_shuffle): Likewise.
20239 (nvptx_gen_wcast): Likewise.
20240 (nvptx_preferred_simd_mode): Likewise.
20241 * config/pa/pa.c (pa_secondary_reload): Likewise.
20242 * config/pa/predicates.md (base14_operand): Likewise.
20243 * config/powerpcspe/powerpcspe-c.c
20244 (altivec_resolve_overloaded_builtin): Likewise.
20245 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
20247 (rs6000_preferred_simd_mode): Likewise.
20248 (output_vec_const_move): Likewise.
20249 (rs6000_expand_vector_extract): Likewise.
20250 (rs6000_split_vec_extract_var): Likewise.
20251 (reg_offset_addressing_ok_p): Likewise.
20252 (rs6000_legitimate_offset_address_p): Likewise.
20253 (rs6000_legitimize_address): Likewise.
20254 (rs6000_emit_set_const): Likewise.
20255 (rs6000_const_vec): Likewise.
20256 (rs6000_emit_move): Likewise.
20257 (spe_build_register_parallel): Likewise.
20258 (rs6000_darwin64_record_arg_recurse): Likewise.
20259 (swap_selector_for_mode): Likewise.
20260 (spe_init_builtins): Likewise.
20261 (paired_init_builtins): Likewise.
20262 (altivec_init_builtins): Likewise.
20263 (do_load_for_compare): Likewise.
20264 (rs6000_generate_compare): Likewise.
20265 (rs6000_expand_float128_convert): Likewise.
20266 (emit_load_locked): Likewise.
20267 (emit_store_conditional): Likewise.
20268 (rs6000_output_function_epilogue): Likewise.
20269 (rs6000_handle_altivec_attribute): Likewise.
20270 (rs6000_function_value): Likewise.
20271 (emit_fusion_gpr_load): Likewise.
20272 (emit_fusion_p9_load): Likewise.
20273 (emit_fusion_p9_store): Likewise.
20274 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
20275 (fusion_gpr_mem_load): Likewise.
20276 (fusion_addis_mem_combo_load): Likewise.
20277 (fusion_addis_mem_combo_store): Likewise.
20278 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
20279 (fusion_gpr_mem_load): Likewise.
20280 (fusion_addis_mem_combo_load): Likewise.
20281 (fusion_addis_mem_combo_store): Likewise.
20282 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20284 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
20285 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20286 (rs6000_preferred_simd_mode): Likewise.
20287 (output_vec_const_move): Likewise.
20288 (rs6000_expand_vector_extract): Likewise.
20289 (rs6000_split_vec_extract_var): Likewise.
20290 (reg_offset_addressing_ok_p): Likewise.
20291 (rs6000_legitimate_offset_address_p): Likewise.
20292 (rs6000_legitimize_address): Likewise.
20293 (rs6000_emit_set_const): Likewise.
20294 (rs6000_const_vec): Likewise.
20295 (rs6000_emit_move): Likewise.
20296 (rs6000_darwin64_record_arg_recurse): Likewise.
20297 (swap_selector_for_mode): Likewise.
20298 (paired_init_builtins): Likewise.
20299 (altivec_init_builtins): Likewise.
20300 (rs6000_expand_float128_convert): Likewise.
20301 (emit_load_locked): Likewise.
20302 (emit_store_conditional): Likewise.
20303 (rs6000_output_function_epilogue): Likewise.
20304 (rs6000_handle_altivec_attribute): Likewise.
20305 (rs6000_function_value): Likewise.
20306 (emit_fusion_gpr_load): Likewise.
20307 (emit_fusion_p9_load): Likewise.
20308 (emit_fusion_p9_store): Likewise.
20309 * config/rx/rx.c (rx_gen_move_template): Likewise.
20310 (flags_from_mode): Likewise.
20311 * config/s390/predicates.md (s390_alc_comparison): Likewise.
20312 (s390_slb_comparison): Likewise.
20313 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
20314 (s390_vector_mode_supported_p): Likewise.
20315 (s390_cc_modes_compatible): Likewise.
20316 (s390_match_ccmode_set): Likewise.
20317 (s390_canonicalize_comparison): Likewise.
20318 (s390_emit_compare_and_swap): Likewise.
20319 (s390_branch_condition_mask): Likewise.
20320 (s390_rtx_costs): Likewise.
20321 (s390_secondary_reload): Likewise.
20322 (__SECONDARY_RELOAD_CASE): Likewise.
20323 (s390_expand_cs): Likewise.
20324 (s390_preferred_simd_mode): Likewise.
20325 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
20326 * config/sh/sh.c (sh_print_operand): Likewise.
20327 (dump_table): Likewise.
20328 (sh_secondary_reload): Likewise.
20329 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20330 * config/sh/sh.md (casesi_worker_1): Likewise.
20331 (casesi_worker_2): Likewise.
20332 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
20333 (fcc_comparison_operator): Likewise.
20334 * config/sparc/sparc.c (sparc_expand_move): Likewise.
20335 (emit_soft_tfmode_cvt): Likewise.
20336 (sparc_preferred_simd_mode): Likewise.
20337 (output_cbranch): Likewise.
20338 (sparc_print_operand): Likewise.
20339 (sparc_expand_vec_perm_bmask): Likewise.
20340 (vector_init_bshuffle): Likewise.
20341 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
20342 (spu_vector_mode_supported_p): Likewise.
20343 (spu_expand_insv): Likewise.
20344 (spu_emit_branch_or_set): Likewise.
20345 (spu_handle_vector_attribute): Likewise.
20346 (spu_builtin_splats): Likewise.
20347 (spu_builtin_extract): Likewise.
20348 (spu_builtin_promote): Likewise.
20349 (spu_expand_sign_extend): Likewise.
20350 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
20351 (tilegx_simd_int): Likewise.
20352 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
20353 (tilepro_simd_int): Likewise.
20354 * config/v850/v850.c (const_double_split): Likewise.
20355 (v850_print_operand): Likewise.
20356 (ep_memory_offset): Likewise.
20357 * config/vax/vax.c (vax_rtx_costs): Likewise.
20358 (vax_output_int_move): Likewise.
20359 (vax_output_int_add): Likewise.
20360 (vax_output_int_subtract): Likewise.
20361 * config/visium/predicates.md (visium_branch_operator): Likewise.
20362 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
20363 (visium_print_operand_address): Likewise.
20364 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20365 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
20366 (xtensa_expand_conditional_branch): Likewise.
20367 (xtensa_copy_incoming_a7): Likewise.
20368 (xtensa_output_literal): Likewise.
20369 * dfp.c (decimal_real_maxval): Likewise.
20370 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
20372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20373 Alan Hayward <alan.hayward@arm.com>
20374 David Sherwood <david.sherwood@arm.com>
20376 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
20377 (mode_nunits_inline): Likewise.
20378 (mode_inner_inline): Likewise.
20379 (mode_unit_size_inline): Likewise.
20380 (mode_unit_precision_inline): Likewise.
20381 (emit_insn_modes_h): Likewise. Also emit a #define of the
20383 (emit_mode_wider): Add an E_ prefix to mode names.
20384 (emit_mode_complex): Likewise.
20385 (emit_mode_inner): Likewise.
20386 (emit_mode_adjustments): Likewise.
20387 (emit_mode_int_n): Likewise.
20388 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
20389 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
20390 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
20391 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
20392 (CRC32_BUILTIN, ENTRY): Likewise.
20393 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
20394 (aarch64_pop_regs): Likewise.
20395 (aarch64_process_components): Likewise.
20396 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
20397 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
20398 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
20399 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
20400 * config/arm/arm.c (arm_init_libfuncs): Likewise.
20401 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
20403 * config/i386/i386-builtin.def (pcmpestr): Likewise.
20404 (pcmpistr): Likewise.
20405 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
20406 * config/mmix/mmix.c (mmix_output_condition): Likewise.
20407 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20409 * config/rl78/rl78.c (mduc_regs): Likewise.
20410 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20411 (htm_expand_builtin): Likewise.
20412 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
20413 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
20414 * config/xtensa/xtensa.c (print_operand): Likewise.
20415 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
20416 (NUM_MODE_VECTOR_INT): Likewise.
20417 * genoutput.c (null_operand): Likewise.
20418 (output_operand_data): Likewise.
20419 * genrecog.c (print_parameter_value): Likewise.
20420 * lra.c (debug_operand_data): Likewise.
20422 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20424 * dwarf2out.c (scompare_loc_descriptor_wide)
20425 (scompare_loc_descriptor_narrow): New functions, split out from...
20426 (scompare_loc_descriptor): ...here.
20427 * expmed.c (emit_store_flag_int): New function, split out from...
20428 (emit_store_flag): ...here.
20430 2017-08-30 Richard Biener <rguenther@suse.de>
20432 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
20433 (dwarf2out_early_finish): Move setting of AT_pubnames from
20434 early debug output to early finish.
20436 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20438 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
20439 and -mdata-region to the assembler.
20441 2017-08-30 Richard Biener <rguenther@suse.de>
20443 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
20445 (gen_subprogram_die): Add DW_AT_object_pointer only early.
20446 (dwarf2out_early_global_decl): Only generate a DIE for the
20447 abstract origin if it doesn't already exist or is a declaration DIE.
20448 (resolve_addr): Do not add the linkage name twice when
20449 generating a stub DIE for the DW_TAG_GNU_call_site target.
20451 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20453 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20454 Use machine_mode rather than int for arg1_mode.
20456 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20459 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
20460 that the second argument of the built-in functions to unpack
20461 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
20462 switch statement instead a lot of if statements.
20463 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
20464 Allow 64-bit values to be in Altivec registers as well as
20465 traditional floating point registers.
20466 (pack<mode>, FMOVE128_VSX iterator): Likewise.
20468 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
20470 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
20471 MAX_REGS_PER_ADDRESS == 1.
20473 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
20475 * config/i386/i386.opt (flag_fentry): Do not init to -1.
20476 * config/i386/i386.c (ix86_option_override_internal): Simplify
20477 setting of opts->x_flag_entry.
20479 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20480 Jakub Jelinek <jakub@redhat.com>
20481 Richard Biener <rguenther@suse.de>
20483 PR tree-optimization/81503
20484 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
20485 folded constant fits in the target type; reorder tests for clarity.
20487 2017-08-29 Martin Liska <mliska@suse.cz>
20489 * passes.def: Include pass_lower_switch.
20490 * stmt.c (dump_case_nodes): Remove and move to
20491 tree-switch-conversion.
20492 (case_values_threshold): Likewise.
20493 (expand_switch_as_decision_tree_p): Likewise.
20494 (emit_case_decision_tree): Likewise.
20495 (expand_case): Likewise.
20496 (balance_case_nodes): Likewise.
20497 (node_has_low_bound): Likewise.
20498 (node_has_high_bound): Likewise.
20499 (node_is_bounded): Likewise.
20500 (emit_case_nodes): Likewise.
20501 (struct simple_case_node): New struct.
20502 (add_case_node): Remove.
20503 (emit_case_dispatch_table): Use vector instead of case_list.
20504 (reset_out_edges_aux): Remove.
20505 (compute_cases_per_edge): Likewise.
20506 (expand_case): Build list of simple_case_node.
20507 (expand_sjlj_dispatch_table): Use it.
20508 * tree-switch-conversion.c (struct case_node): Moved from
20509 stmt.c and adjusted.
20510 (emit_case_nodes): Likewise.
20511 (node_has_low_bound): Likewise.
20512 (node_has_high_bound): Likewise.
20513 (node_is_bounded): Likewise.
20514 (case_values_threshold): Likewise.
20515 (reset_out_edges_aux): Likewise.
20516 (compute_cases_per_edge): Likewise.
20517 (add_case_node): Likewise.
20518 (dump_case_nodes): Likewise.
20519 (balance_case_nodes): Likewise.
20520 (expand_switch_as_decision_tree_p): Likewise.
20521 (emit_jump): Likewise.
20522 (emit_case_decision_tree): Likewise.
20523 (try_switch_expansion): Likewise.
20524 (do_jump_if_equal): Likewise.
20525 (emit_cmp_and_jump_insns): Likewise.
20526 (fix_phi_operands_for_edge): New function.
20527 (record_phi_operand_mapping): Likewise.
20528 (class pass_lower_switch): New pass.
20529 (pass_lower_switch::execute): New function.
20530 (make_pass_lower_switch): Likewise.
20531 (conditional_probability):
20532 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
20533 * tree-pass.h: Add make_pass_lower_switch.
20535 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20538 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
20541 2017-08-29 Richard Biener <rguenther@suse.de>
20543 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
20544 we do not add a DW_AT_inline attribute twice.
20545 (gen_subprogram_die): Remove code setting DW_AT_inline on
20546 DECL_ABSTRACT_P nodes.
20548 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
20550 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
20551 calls to internal functions.
20552 (gimplify_modify_expr): Likewise.
20553 * tree-call-cdce.c (use_internal_fn): Likewise.
20554 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
20555 (convert_to_divmod): Set the nothrow flag.
20556 * tree-if-conv.c (predicate_mem_writes): Likewise.
20557 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20558 (vectorizable_call): Likewise.
20559 (vectorizable_store): Likewise.
20560 (vectorizable_load): Likewise.
20561 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
20562 (vect_recog_mask_conversion_pattern): Likewise.
20564 2017-08-29 Martin Liska <mliska@suse.cz>
20567 * gcc.c (driver_handle_option): Add new argument.
20568 * opts-common.c (handle_option): Pass
20569 target_option_override_hook.
20570 * opts-global.c (lang_handle_option): Add new option.
20571 (set_default_handlers): Add new argument.
20572 (decode_options): Likewise.
20573 * opts.c (target_handle_option): Likewise.
20574 (common_handle_option): Call target_option_override_hook.
20575 * opts.h (struct cl_option_handler_func): Add hook for
20576 target option override.
20577 (struct cl_option_handlers): Likewise.
20578 (set_default_handlers): Add new argument.
20579 (decode_options): Likewise.
20580 (common_handle_option): Likewise.
20581 (target_handle_option): Likewise.
20582 * toplev.c (toplev::main): Pass targetm.target_option.override
20585 2017-08-29 Richard Biener <rguenther@suse.de>
20586 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
20588 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
20589 life to the active subtree.
20591 2017-08-28 Jeff Law <law@redhat.com>
20593 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
20594 derive_equivalences.
20595 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
20596 Code moved into....
20597 (edge_info::derive_equivalences): New private member function
20599 * tree-ssa-dom.c (class edge_info): Changed from a struct
20600 to a class. Add ctor/dtor, methods and data members.
20601 (edge_info::edge_info): Renamed from allocate_edge_info.
20602 Initialize additional members.
20603 (edge_info::~edge_info): New.
20604 (free_dom_edge_info): Delete the edge info.
20605 (record_edge_info): Use new class & associated member functions.
20606 Tighten forms for testing for edge equivalences.
20607 (record_temporary_equivalences): Iterate over the simple
20608 equivalences rather than assuming there's only one per edge.
20609 (cprop_into_successor_phis): Iterate over the simple
20610 equivalences rather than assuming there's only one per edge.
20611 (optimize_stmt): Use operand_equal_p rather than pointer
20612 equality for mini-DSE code.
20614 2017-08-28 Nathan Sidwell <nathan@acm.org>
20616 * gcc.c (execute): Fold SIGPIPE handling into switch
20617 statement. Adjust internal error message.
20619 2017-08-28 Richard Biener <rguenther@suse.de>
20622 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
20623 Do nothing for removed DIEs.
20625 2017-08-28 Richard Biener <rguenther@suse.de>
20627 PR tree-optimization/81977
20628 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
20631 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
20634 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
20635 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
20636 using targetm.gen_mem_thread_fence.
20638 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
20641 * config/i386/i386.md (*<btsc><mode>): Change operand 2
20642 predicate to register_operand. Reorder operands.
20643 (*btr<mode>): Ditto.
20644 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
20645 (*btr<mode>_mask): Ditto.
20647 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
20649 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
20650 * config/rs6000/xmmintrin.h: New file.
20651 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
20653 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20656 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
20657 parameter and_insn and return it.
20658 (recombine_lvx_pattern): Insert a copy to ensure availability of
20659 the base register of the copied masking operation at the point of
20660 the instruction replacement.
20661 (recombine_stvx_pattern): Likewise.
20663 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20665 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
20666 undocumented switches.
20667 (-mpower9-dform-vector): Likewise.
20668 (-mpower9-dform): Likewise.
20669 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
20670 comments to delete references to -mpower9-dform* switches.
20671 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
20672 Delete reference to -mpower9-dform* switches, test for
20673 -mpower9-vector instead.
20674 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
20675 (OTHER_P9_VECTOR_MASKS): Likewise.
20676 (POWERPC_MASKS): Likewise.
20677 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
20678 tests against -mpower9-dform* to -mpower9-vector. Delete code
20679 that checked for -mpower9-dform* consistancy with other options.
20680 Add test for -mpower9-misc to enable other power9 switches.
20681 (rs6000_init_hard_regno_mode_ok): Likewise.
20682 (rs6000_option_override_internal): Likewise.
20683 (rs6000_emit_prologue): Likewise.
20684 (rs6000_emit_epilogue): Likewise.
20685 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
20686 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
20687 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
20689 (emit_fusion_p9_store): Likewise.
20690 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
20691 resetting these macros if the assembler does not support ISA 3.0
20693 (TARGET_P9_DFORM_VECTOR): Likewise.
20694 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
20695 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
20697 2017-08-25 Alan Modra <amodra@gmail.com>
20699 PR rtl-optimization/81747
20700 * cse.c (cse_extended_basic_block): Don't attempt to record
20701 equivalences for degenerate conditional branches that branch
20702 to their fall-through.
20704 2017-08-24 Martin Sebor <msebor@redhat.com>
20706 PR middle-end/81908
20707 * gimple-fold.c (size_must_be_zero_p): New function.
20708 (gimple_fold_builtin_memory_op): Call it.
20710 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
20712 * config/rs6000/mm_malloc.h: New file.
20714 2017-08-24 Bin Cheng <bin.cheng@arm.com>
20716 PR tree-optimization/81913
20717 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
20718 analysis when either IVs in condition can wrap.
20720 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
20722 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
20723 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
20725 2017-08-24 Richard Biener <rguenther@suse.de>
20728 * targhooks.c (default_target_can_inline_p): Properly
20729 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
20730 is present and always compare.
20731 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
20732 infer -mfpmath=sse from TARGET_SSE_P.
20733 (ix86_can_inline_p): Properly use target_option_default_node when
20734 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
20736 2017-08-24 Richard Biener <rguenther@suse.de>
20739 * dwarf2out.c (output_die): Handle flag_generate_offload like
20741 (output_comp_unit): Likewise.
20742 (gen_array_type_die): Likewise.
20743 (dwarf2out_early_finish): Likewise.
20744 (note_variable_value_in_expr): Likewise.
20745 (dwarf2out_finish): Likewise. Adjust assert.
20746 * cgraphunit.c (symbol_table::compile): Move setting of
20747 flag_generate_offload earlier ...
20748 (symbol_table::finalize_compilation_unit): ... here, before
20749 early debug finalization.
20751 2017-08-24 Richard Biener <rguenther@suse.de>
20753 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
20754 and ipa-fnsummary.h.
20755 (ix86_can_inline_p): When ix86_fpmath flags do not match
20756 check whether the callee uses FP math at all.
20758 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
20760 PR middle-end/81931
20761 * tree-ssanames.c (get_nonzero_bits): Use element_precision
20762 instead of TYPE_PRECISION.
20764 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
20765 Alan Hayward <alan.hayward@arm.com>
20766 David Sherwood <david.sherwood@arm.com>
20768 * combine.c (make_extraction): Use subreg_offset_from_lsb.
20770 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
20772 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
20774 (ix86_frame::stack_realign_allocate): New field.
20775 (struct machine_frame_state): Modify comments.
20776 (machine_frame_state::sp_realigned_fp_end): New field.
20777 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
20778 layout calculation.
20779 (sp_valid_at): Add assertion to assure no attempt to access invalid
20780 offset of a realigned stack.
20781 (fp_valid_at): Likewise.
20782 (choose_baseaddr): Modify comments.
20783 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
20784 ix86_expand_prologue.
20785 (ix86_expand_prologue): Modify stack realignment and allocation.
20786 (ix86_expand_epilogue): Modify comments.
20787 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
20788 avx2_runtime, avx512f, and avx512f_runtime.
20790 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20792 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20793 (mstackrealign): Do not init to -1.
20794 * config/i386/i386.c (ix86_option_override_internal):
20795 Check opts_set, not opts when setting default value of
20796 opts->x_ix86_force_align_arg_pointer.
20798 2017-08-23 Richard Biener <rguenther@suse.de>
20800 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20801 lang_hooks.decl_printable_name.
20802 * print-rtl-function.c (print_rtx_function): Likewise.
20803 * tree-pretty-print.c (dump_function_header): Likewise.
20805 2017-08-23 Richard Biener <rguenther@suse.de>
20808 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20809 -g0 at compile-time.
20811 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20813 PR middle-end/19706
20814 * doc/sourcebuild.texi (Other hardware attributes):
20817 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20819 PR middle-end/19706
20820 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20821 Fix single-use check.
20823 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20825 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20827 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20829 * doc/install.texi: Modify to add more details on running selected
20832 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20834 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20835 is combined with -mabi=ms.
20836 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20839 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20841 PR tree-optimization/81488
20842 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20843 and cached_basis fields.
20844 (MAX_SPREAD): New constant.
20845 (alloc_cand_and_find_basis): Initialize new fields.
20846 (clear_visited): New function.
20847 (create_phi_basis_1): Rename from create_phi_basis, set visited
20848 and cached_basis fields.
20849 (create_phi_basis): New wrapper function.
20850 (phi_add_costs_1): Rename from phi_add_costs, add spread
20851 parameter, set visited field, short-circuit when limits reached.
20852 (phi_add_costs): New wrapper function.
20853 (record_phi_increments_1): Rename from record_phi_increments, set
20855 (record_phi_increments): New wrapper function.
20856 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20857 (phi_incr_cost): New wrapper function.
20858 (all_phi_incrs_profitable_1): Rename from
20859 all_phi_incrs_profitable, set visited field.
20860 (all_phi_incrs_profitable): New wrapper function.
20862 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20863 Alan Hayward <alan.hayward@arm.com>
20864 David Sherwood <david.sherwood@arm.com>
20866 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20867 that takes the outer and inner modes.
20868 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20869 comparison as the canonical test for a paradoxical subreg.
20870 * combine.c (simplify_set): Use paradoxical_subreg_p.
20871 (make_extraction): Likewise.
20872 (force_to_mode): Likewise.
20873 (rtx_equal_for_field_assignment_p): Likewise.
20874 (gen_lowpart_for_combine): Likewise.
20875 (simplify_comparison): Likewise.
20876 * cse.c (equiv_constant): Likewise.
20877 * expmed.c (store_bit_field_1): Likewise.
20878 * final.c (alter_subreg): Likewise.
20879 * fwprop.c (propagate_rtx): Likewise.
20880 (forward_propagate_subreg): Likewise.
20881 * ira-conflicts.c (ira_build_conflicts): Likewise.
20882 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20883 * lra-constraints.c (curr_insn_transform): Likewise.
20884 (split_reg): Likewise.
20885 * lra-eliminations.c (move_plus_up): Likewise.
20886 (lra_eliminate_regs_1): Likewise.
20887 * recog.c (general_operand): Likewise.
20888 * ree.c (combine_reaching_defs): Likewise.
20889 * reload.c (push_reload): Likewise.
20890 (find_reloads): Likewise.
20891 * reload1.c (elimination_effects): Likewise.
20892 (compute_reload_subreg_offset): Likewise.
20893 (choose_reload_regs): Likewise.
20894 * rtlanal.c (subreg_lsb_1): Likewise.
20895 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20896 (simplify_subreg): Likewise.
20897 * var-tracking.c (track_loc_p): Likewise.
20898 * emit-rtl.c (byte_lowpart_offset): Likewise.
20899 (paradoxical_subreg_p): Delete out-of-line definition.
20901 2017-08-22 Jeff Law <law@redhat.com>
20903 PR tree-optimization/81741
20904 PR tree-optimization/71947
20905 * tree-ssa-dom.c: Include tree-inline.h.
20906 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20907 equivalences if one is more expensive to compute than the other.
20908 * tree-ssa-scopedtables.h (class const_or_copies): Make
20909 record_const_or_copy_raw method private.
20910 (class avail_exprs_stack): New method simplify_binary_operation.
20911 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20912 avail_exprs_stack::simplify_binary_operation as needed.
20913 (avail_exprs_stack::simplify_binary_operation): New function.
20915 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20917 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20918 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20919 (DOT_SYMBOLS): Likewise.
20920 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20921 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20922 (RS6000_ABI_NAME): Likewise.
20923 (TARGET_CMODEL): Likewise.
20924 (TOC_SECTION_ASM_OP): Likewise.
20925 (SET_CMODEL): New macro.
20926 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20928 2017-08-22 Richard Biener <rguenther@suse.de>
20930 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20931 to allow for free-lang-data replacements similar to verify_type_variant.
20933 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20934 Alan Hayward <alan.hayward@arm.com>
20935 David Sherwood <david.sherwood@arm.com>
20937 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20938 VOIDmode for the LABEL_REF.
20940 2017-08-22 Richard Biener <rguenther@suse.de>
20942 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20944 2017-08-22 Richard Biener <rguenther@suse.de>
20946 * common.opt (feliminate-dwarf2-dups): Ignore.
20947 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20948 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20949 same_die_p_wrap, compute_section_prefix,
20950 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20951 (comdat_symbol_id, comdat_symbol_number): Likewise.
20952 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20954 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20955 (output_die): Mark unreachable path unreachable.
20956 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20957 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20958 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20959 (dwarf2out_early_finish): Likewise.
20961 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20963 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20965 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20968 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20969 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20970 (avr_attribute_table) <io, io_low, address>: Initialize
20971 .decl_required with true.
20973 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20975 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20976 undocumented debugging options.
20977 (-mvsx-scalar-double): Likewise.
20978 (-mallow-df-permute): Likewise.
20979 (-mvectorize-builtins): Likewise.
20980 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20981 (rs6000_builtin_vectorized_function): Likewise.
20982 (rs6000_builtin_md_vectorized_function): Likewise.
20983 (rs6000_opt_vars): Likewise.
20985 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20988 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20989 (*btrq_imm): Rename from *btrq.
20990 (*btcq_imm): Rename from *btcq.
20991 (btsc): New code attribute.
20992 (*<btsc><mode>): New insn pattern.
20993 (*btr<mode>): Ditto.
20994 (*<btsc><mode>_mask): New insn_and_split pattern.
20995 (*btr<mode>_mask): Ditto.
20997 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20998 Alan Hayward <alan.hayward@arm.com>
20999 David Sherwood <david.sherwood@arm.com>
21001 * function.c (pad_below): Simplify padding calculation.
21003 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21004 Alan Hayward <alan.hayward@arm.com>
21005 David Sherwood <david.sherwood@arm.com>
21007 * target.def (function_prologue): Remove frame size argument.
21008 (function_epilogue): Likewise.
21009 * doc/tm.texi: Regenerate.
21010 * final.c (final_start_function): Update call to function_prologue.
21011 (final_end_function): Update call to function_epilogue.
21012 (default_function_pro_epilogue): Remove frame size argument.
21013 * output.h (default_function_pro_epilogue): Likewise.
21014 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
21015 (arm_output_function_prologue): Likewise.
21016 * config/frv/frv.c (frv_function_prologue): Likewise.
21017 (frv_function_epilogue): Likewise.
21018 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
21019 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
21020 (ia64_output_function_epilogue): Likewise.
21021 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
21022 (m32r_output_function_epilogue): Likewise.
21023 * config/microblaze/microblaze.c (microblaze_function_prologue)
21024 (microblaze_function_epilogue): Likewise.
21025 * config/mips/mips.c (mips_output_function_prologue): Likewise.
21026 (mips_output_function_epilogue): Likewise.
21027 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
21028 (mmix_target_asm_function_epilogue): Likewise.
21029 * config/msp430/msp430.c (msp430_start_function): Likewise.
21030 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
21031 (nds32_asm_function_epilogue): Likewise.
21032 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
21033 * config/pa/pa.c (pa_output_function_prologue): Likewise.
21034 (pa_output_function_epilogue): Likewise.
21035 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
21036 (rs6000_output_function_epilogue): Likewise.
21037 * config/rl78/rl78.c (rl78_start_function): Likewise.
21038 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
21039 (rs6000_output_function_epilogue): Likewise.
21040 * config/rx/rx.c (rx_output_function_prologue): Likewise.
21041 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
21042 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
21043 (sparc_asm_function_epilogue): Likewise.
21045 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21047 * tree.h (type_has_mode_precision_p): New function.
21048 * convert.c (convert_to_integer_1): Use it.
21049 * expr.c (expand_expr_real_2): Likewise.
21050 (expand_expr_real_1): Likewise.
21051 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
21052 * match.pd: Likewise.
21053 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
21054 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
21055 * tree-tailcall.c (process_assignment): Likewise.
21056 * tree-vect-loop.c (vectorizable_reduction): Likewise.
21057 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
21058 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
21059 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
21060 (vectorizable_assignment): Likewise.
21061 (vectorizable_shift): Likewise.
21062 (vectorizable_operation): Likewise.
21063 * tree-vrp.c (register_edge_assert_for_2): Likewise.
21065 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
21067 * match.pd: Add pow (C, x) simplification.
21069 2017-08-21 Richard Biener <rguenther@suse.de>
21071 PR tree-optimization/81900
21072 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
21073 for blocks with abnormal predecessors.
21074 (compute_antic): Do not set visited flag prematurely.
21076 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
21079 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
21081 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21083 * stor-layout.h (vector_type_mode): Move to...
21084 * tree.h (vector_type_mode): ...here.
21085 * stor-layout.c (vector_type_mode): Move to...
21086 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
21088 2017-08-21 Richard Biener <rguenther@suse.de>
21090 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
21091 register_external_die hooks.
21092 (debug_false_tree_charstarstar_uhwistar): Declare.
21093 (debug_nothing_tree_charstar_uhwi): Likewise.
21094 * debug.c (do_nothing_debug_hooks): Adjust.
21095 (debug_false_tree_charstarstar_uhwistar): New do nothing.
21096 (debug_nothing_tree_charstar_uhwi): Likewise.
21097 * dbxout.c (dbx_debug_hooks): Adjust.
21098 (xcoff_debug_hooks): Likewise.
21099 * sdbout.c (sdb_debug_hooks): Likewise.
21100 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21101 * dwarf2out.c (macinfo_label_base): New global.
21102 (dwarf2out_register_external_die): New function for the
21103 register_external_die hook.
21104 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
21105 (dwarf2_debug_hooks): Use them.
21106 (dwarf2_lineno_debug_hooks): Adjust.
21107 (struct die_struct): Add with_offset flag.
21108 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
21109 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
21110 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
21111 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
21112 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
21113 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
21114 defining section names for the early LTO debug variants.
21115 (reset_indirect_string): New helper.
21116 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
21117 (print_dw_val): Add support for offsetted symbol references.
21118 (get_ultimate_context): Split out from is_cxx.
21119 (is_cxx): Use get_ultimate_context.
21120 (is_fortran): Add decl overload.
21121 (compute_comp_unit_symbol): Split out worker from
21122 compute_section_prefix.
21123 (compute_section_prefix): Call compute_comp_unit_symbol and
21124 set comdat_type_p here.
21125 (output_die): Skip DIE symbol output for the LTO added one.
21126 Handle DIE symbol references with offset.
21127 (output_comp_unit): Guard section name mangling properly.
21128 For LTO debug sections emit a symbol at the section beginning
21129 which we use to refer to its DIEs.
21130 (add_abstract_origin_attribute): For DIEs registered via
21131 dwarf2out_register_external_die directly refer to the early
21132 DIE rather than indirectly through the shadow one we created.
21133 Remove obsolete call to dwarf2out_abstract_function for
21134 non-function/block origins.
21135 (gen_array_type_die): When generating early LTO debug do
21136 not emit DW_AT_string_length.
21137 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
21138 late when in LTO. As suggested place a gcc_unreachable for
21139 the DECL_ABSTRACT_P case.
21140 (gen_subprogram_die): Avoid another specification DIE
21141 for early built declarations/definitions for the late LTO case.
21142 (gen_variable_die): Add type references for late duplicated VLA dies
21144 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
21145 we have the abstract instance already.
21146 (process_scope_var): Adjust decl DIE contexts in LTO which
21147 first puts them in limbo.
21148 (gen_decl_die): Do not generate type DIEs late apart from
21149 types for VLAs or for decls we do not yet have a DIE. Do not
21150 call dwarf2out_abstract_function late.
21151 (dwarf2out_early_global_decl): Make sure to create DIEs
21152 for abstract instances of a decl first.
21153 (dwarf2out_late_global_decl): Adjust comment.
21154 (output_macinfo_op): With multiple macro sections use
21155 macinfo_label_base to distinguish labels.
21156 (output_macinfo): Likewise. Update macinfo_label_base.
21157 Pass in the line info label.
21158 (note_variable_value_in_expr): When generating LTO resolve
21159 all variable values here by generating DIEs as needed.
21160 (init_sections_and_labels): Add early LTO debug flag parameter
21161 and generate different sections and names if set. Add generation
21162 counter for the labels so we can have multiple of them.
21163 (reset_dies): Helper to allow DIEs to be output multiple times.
21164 (dwarf2out_finish): When outputting DIEs to the fat part of an
21165 LTO object first reset DIEs.
21166 (dwarf2out_early_finish): Output early DIEs when generating LTO.
21167 (modified_type_die): Check for decl_ultimate_origin being self
21169 (gen_type_die_with_usage): Likewise.
21170 (gen_typedef_die): Allow decl_ultimate_origin being self.
21171 (set_decl_abstract_flags): Remove.
21172 (set_block_abstract_flags): Likewise.
21173 (dwarf2out_abstract_function): Treat the early generated DIEs
21174 as the abstract copy and only add DW_AT_inline and
21175 DW_AT_artificial here and call set_decl_origin_self.
21176 If the DIE has an abstract origin don't do anything.
21177 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
21178 if we have none yet (Go fails to build one, PR78628).
21179 (variably_modified_type_p): Prevent endless recursion for Ada
21180 cyclic pointer types.
21181 * lto-streamer-in.c: Include debug.h.
21182 (dref_queue): New global.
21183 (lto_read_tree_1): Stream in DIE references.
21184 (lto_input_tree): Register DIE references.
21185 (input_function): Stream DECL_DEBUG_ARGS.
21186 * lto-streamer-out.c: Include debug.h.
21187 (lto_write_tree_1): Output DIE references.
21188 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
21189 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
21190 (output_function): Stream DECL_DEBUG_ARGS.
21191 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
21192 Stream DECL_ABSTRACT_ORIGIN.
21193 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
21194 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
21195 DECL_CONTEXT for file-scope decls.
21196 * lto-streamer.h (struct dref_entry): Declare.
21197 (dref_queue): Likewise.
21198 * cfgexpand.c (pass_expand::execute): Do not call the
21199 outlining_inline_function hook here.
21200 * lto-wrapper.c (debug_obj): New global.
21201 (tool_cleanup): Unlink it if required.
21202 (debug_objcopy): New function.
21203 (run_gcc): Handle early debug sections in the IL files by
21204 extracting them to separate files, partially linkin them and
21205 feeding the result back as result to the linker.
21206 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
21207 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
21208 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
21209 sections into a separate segment.
21210 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
21212 (darwin_asm_dwarf_section): Likewise.
21213 (darwin_asm_output_dwarf_offset): Likewise.
21214 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
21216 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21217 Alan Hayward <alan.hayward@arm.com>
21218 David Sherwood <david.sherwood@arm.com>
21220 * read-md.h (md_reader::record_potential_iterator_use): Replace
21221 pointer argument with an rtx and an index.
21222 * read-rtl.c (iterator_group::apply_iterator): Likewise.
21223 (apply_mode_iterator): Likewise.
21224 (apply_code_iterator): Likewise.
21225 (apply_int_iterator): Likewise.
21226 (apply_subst_iterator): Likewise.
21227 (record_iterator_use): Likewise.
21228 (record_attribute_use): Likewise.
21229 (md_reader::record_potential_iterator_use): Likewise. Update calls
21230 to record_iterator_use and apply_iterator.
21231 (iterator_use): Replace ptr with x and index.
21232 (attribute_use): Likewise.
21233 (apply_attribute_uses): Update calls to apply_iterator.
21234 (apply_iterators): Likewise. Update initialization of iterator_use.
21235 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
21236 and record_potential_iterator_use.
21237 (rtx_reader::read_rtx_operand): Likewise.
21239 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21240 Alan Hayward <alan.hayward@arm.com>
21241 David Sherwood <david.sherwood@arm.com>
21243 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
21246 2017-08-21 Richard Biener <rguenther@suse.de>
21248 PR middle-end/81884
21249 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
21250 at struct end conservatively when comparing common bases.
21252 2017-08-21 Richard Biener <rguenther@suse.de>
21254 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
21255 (mem_ref_in_stmt): Remove.
21256 (determine_max_movement): Use ref index to get at the reference.
21257 (invariantness_dom_walker::before_dom_children): Deal with
21258 lim data already initialized.
21259 (gather_mem_refs_stmt): Initialize lim data and record ref index.
21261 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21263 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
21264 (TARGET_ISA_ROUND): Ditto.
21265 (TARGET_ROUND): Ditto.
21266 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
21267 * config/i386/i386.md: Ditto.
21268 * config/i386/sse.md: Ditto.
21269 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
21270 with OPTION_MASK_ISA_SSE4_1.
21272 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21275 * doc/extend.texi (x86 Built-in Functions): Correct the name of
21276 __builtin_ia32_lzcnt_u16.
21278 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
21281 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
21282 (rs6000_set_current_function): Rewrite function to use it.
21284 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
21287 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
21288 and TYPE_WARN_IF_NOT_ALIGN.
21289 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
21290 (handle_warn_if_not_align): New.
21291 (place_union_field): Call handle_warn_if_not_align.
21292 (place_field): Call handle_warn_if_not_align.
21293 Copy TYPE_WARN_IF_NOT_ALIGN.
21294 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
21295 (layout_type): Likewise.
21296 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
21298 (tree_decl_common): Add warn_if_not_align.
21299 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
21300 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
21301 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
21302 (DECL_WARN_IF_NOT_ALIGN): Likewise.
21303 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
21304 * doc/extend.texi: Document warn_if_not_aligned attribute.
21305 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
21307 2017-08-17 Martin Liska <mliska@suse.cz>
21310 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
21311 (get_data_dependence): Use it as pointer type.
21312 (distribute_loop): Likewise.
21314 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21316 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
21317 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
21318 (p8_vmrgow_<mode>_direct): New define_insn.
21319 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
21320 handle endianness for vmrgew and vmrgow permute patterns.
21322 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21324 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
21325 * config/rs6000/rs6000-cpus.def: Remove comment.
21326 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
21327 (POWERPC_MASKS): Likewise.
21328 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
21329 use of TARGET_VSX_TIMODE.
21330 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
21331 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
21332 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
21333 (rs6000_option_override_internal): Remove dead code.
21334 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
21335 (rs6000_legitimize_reload_address): Likewise.
21336 (rs6000_legitimate_address_p): Likewise.
21337 (rs6000_opt_masks): Delete "vsx-timode".
21338 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
21339 from function comment.
21340 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
21341 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
21342 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
21344 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
21345 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
21346 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
21347 (VSX_TI): Likewise.
21349 (define_peephole2): Likewise.
21351 2017-08-17 Martin Sebor <msebor@redhat.com>
21354 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
21355 past the end of an array.
21356 (test_pp_format): Add test cases.
21358 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
21360 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
21361 missing ECF_NOTHROW flags.
21363 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21366 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
21367 operands residing in integer registers.
21368 (*vsx_le_perm_load_<mode>): Likewise.
21369 (*vsx_le_perm_store_<mode>): Likewise.
21370 (define_peephole2): Add peepholes to optimize the above.
21372 2017-08-17 Marek Polacek <polacek@redhat.com>
21374 PR middle-end/81814
21375 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
21376 to mimic what shorten_compare did. Change the return type to bool.
21377 (fold_cond_expr_with_comparison): Update call to
21378 operand_equal_for_comparison_p.
21379 (fold_ternary_loc): Likewise.
21381 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
21383 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
21385 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
21387 2017-08-17 Richard Biener <rguenther@suse.de>
21389 * tree-ssa-structalias.c (solve_graph): When propagating
21390 to successors update the graphs succ edges and avoid duplicate work.
21392 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
21395 * config/i386/i386.c (ix86_option_override_internal): Save target
21396 specific options after ix86_stack_protector_guard_reg was changed.
21398 2017-08-17 Richard Biener <rguenther@suse.de>
21400 PR tree-optimization/81827
21401 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
21402 (new_var_info): Initialize it conservatively.
21403 (get_call_vi): Mark register vars.
21404 (new_scalar_tmp_constraint_exp): Likewise.
21405 (handle_rhs_call): Likewise.
21406 (handle_const_call): Likewise.
21407 (create_function_info_for): Likewise.
21408 (solve_constraints): Sort varinfos to separate register from
21409 non-register vars to pack points-to solution bitmaps during
21412 2017-08-17 Marek Polacek <polacek@redhat.com>
21414 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
21416 2017-08-17 Richard Biener <rguenther@suse.de>
21418 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
21419 to true when overflow is undefined and we saturated the result.
21421 2017-08-17 Alan Modra <amodra@gmail.com>
21424 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
21425 Don't use store multiple if only one reg needs saving.
21426 (interesting_frame_related_regno): New function.
21427 (rs6000_frame_related): Don't emit frame info for regs that
21429 (rs6000_emit_epilogue): Likewise.
21431 2017-08-16 Nathan Sidwell <nathan@acm.org>
21433 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
21434 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
21435 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
21436 * tree.c (free_lang_data_in_type): Use else-if chain. Always
21437 clear TYPE_LANG_1. Remove obsolete member-function stripping.
21438 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
21439 (verify_type): Adjust for TYPE_BINFO move.
21440 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
21441 process TYPE_BINFO directly.
21442 (hash_tree): Likewise.
21443 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21445 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21448 2017-08-16 David Malcolm <dmalcolm@redhat.com>
21450 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
21452 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
21455 * config/i386/i386.md (*anddi_1_btr): Change predicates of
21456 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
21457 Add ix86_binary_operator_ok to insn constraint.
21458 (*iordi_1_bts): Ditto.
21459 (*xordi_1_btc): Ditto.
21460 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
21461 Update corresponding peephole2 pattern.
21465 2017-08-16 Bin Cheng <bin.cheng@arm.com>
21467 PR tree-optimization/81832
21468 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
21469 copy loop header which has IFN_LOOP_DIST_ALIAS call.
21471 2017-08-16 Marek Polacek <polacek@redhat.com>
21474 * fold-const.c (fold_indirect_ref_1): Restore original behavior
21475 regarding size_zero_node.
21477 2017-08-16 Martin Liska <mliska@suse.cz>
21480 * config.gcc: Respect previously set extra_objs in case
21483 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
21485 PR tree-optimization/81835
21486 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
21487 the phi SSA_NAME. Check that the condition in a COND_EXPR does
21488 not depend on the phi.
21490 2017-08-16 Alan Modra <amodra@gmail.com>
21492 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
21495 2017-08-16 Alan Modra <amodra@gmail.com>
21497 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
21498 (save_reg_p): ..into this. Update all callers.
21499 (first_reg_to_save): Simplify.
21501 2017-08-16 Alan Modra <amodra@gmail.com>
21503 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
21506 2017-08-15 Joseph Myers <joseph@codesourcery.com>
21510 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
21511 constant count if that count is less than 32.
21513 2017-08-15 Nathan Sidwell <nathan@acm.org>
21515 * gcc.c (execute): Emit friendlier message if inferior is killed
21516 by an external cause.
21518 2017-08-15 Richard Biener <rguenther@suse.de>
21520 PR tree-optimization/81790
21521 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
21522 CONSTRUCTORs from simplifying and VN.
21524 2017-08-14 Martin Sebor <msebor@redhat.com>
21526 * builtin-attrs.def: Add comments.
21528 2017-08-14 Martin Sebor <msebor@redhat.com>
21531 * doc/extend.texi (attribute nonstring): Document new attribute.
21533 2017-08-14 Martin Sebor <msebor@redhat.com>
21536 * tree-diagnostic.c (default_tree_printer): Handle %G.
21537 * gimple-pretty-print.h (percent_G_format): Declare new function.
21538 * gimple-pretty-print.c (percent_G_format): Define.
21539 * tree-pretty-print.c (percent_K_format): Add argument.
21541 2017-08-14 Martin Sebor <msebor@redhat.com>
21543 PR translation/79998
21544 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
21545 Remove a stray space.
21547 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
21550 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
21551 (*iordi_1_bts): Ditto.
21552 (*xordi_1_btc): Ditto.
21554 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21557 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
21558 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21560 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
21561 quoted strings, and make more translator-friendly.
21562 (darwin_rs6000_override_options): Likewise.
21563 (rs6000_option_override_internal): Likewise.
21564 (rs6000_return_in_memory): Fix overlong line.
21565 (init_cmulative_args): Use quoted strings, and make more
21566 translator-friendly.
21567 (rs6000_pass_by_reference): Fix overlong line.
21568 (def_builtin): Use quoted strings.
21569 (altivec_expand_predicate_builtin): Use quoted strings, and make
21570 more translator-friendly.
21571 (htm_expand_builtin): Use quoted strings.
21572 (cpu_expand_builtin): Use quoted strings, and make more
21573 translator-friendly.
21574 (altivec_expand_builtin): Likewise.
21575 (paired_expand_predicate_builtin): Likewise.
21576 (rs6000_invalid_builtin): Likewise.
21577 (builtin_function_type): Use quoted strings.
21578 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
21579 more translator-friendly.
21580 (rs6000_trampoline_init): Likewise.
21581 (rs6000_handle_altivec_attribute): Likewise.
21582 (rs6000_inner_target_options): Use quoted strings.
21583 (rs6000_disable_incompatible_switches): Likewise.
21584 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
21585 strings, and make more translator-friendly.
21586 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
21588 2017-08-14 Bin Cheng <bin.cheng@arm.com>
21590 PR tree-optimization/81799
21591 * tree-loop-distribution.c (version_loop_by_alias_check): Force
21592 cond_expr to simple gimple operand.
21594 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
21596 PR middle-end/46932
21597 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
21599 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
21603 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
21604 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
21605 TARGET_GASISR_PROLOGUES.
21606 * config/avr/avr.c (avr_option_override): Same.
21607 (avr_pass_pre_proep::execute): Same.
21609 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
21612 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
21613 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
21615 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
21617 * config/i386/i386.md (*load_tp_<mode>): Redefine as
21618 define_insn_and_split. Split to a memory load from 0 in
21619 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
21620 using PTR mode iterator.
21621 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
21622 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
21623 (*add_tp_<mode>): Redefine as define_insn_and_split.
21624 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
21625 address space. Merge with *add_tp_x32 using PTR mode iterator.
21626 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
21627 Split to an add with a memory load from 0 in
21628 DEFAULT_TLS_SEG_REG address space.
21630 2017-08-12 Andrew Pinski <apinski@cavium.com>
21632 * config/aarch64/aarch64-option-extensions.def (rdma):
21633 Fix feature string to what Linux prints out in /proc/cpuinfo.
21635 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
21638 * dwarf2out.c (modified_type_die): For C typedef types that have
21639 an ultimate origin, process the ultimate origin instead of the
21641 (gen_typedef_die): Assert that input DECLs have no ultimate
21643 (gen_type_die_with_usage): For typedef variants that have an
21644 ultimate origin, just call gen_decl_die on the original DECL.
21645 (process_scope_var): Avoid creating DIEs for local typedefs and
21646 concrete static variables.
21648 2017-08-12 Alan Modra <amodra@gmail.com>
21652 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
21653 match gnu-user.h startfile.
21654 (ENDFILE_LINUX_SPEC): Similarly.
21656 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
21659 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
21661 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
21663 2017-08-11 Tamar Christina <tamar.christina@arm.com>
21664 * config/aarch64/aarch64.md (mov<mode>): Change.
21665 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21666 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
21667 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
21669 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
21671 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
21672 for storage order barriers.
21674 2017-08-11 Martin Liska <mliska@suse.cz>
21677 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
21678 variables of void type.
21680 2017-08-11 Martin Liska <mliska@suse.cz>
21682 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
21683 TARGET_SUPPORTS_ALIASES.
21684 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
21685 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
21686 (optimize_weakref): Likewise.
21687 * symtab.c (symtab_node::noninterposable_alias): Likewise.
21688 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
21689 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
21691 2017-08-11 Martin Liska <mliska@suse.cz>
21694 * config/i386/i386.c (make_resolver_func): Do complete
21695 refactoring of the function.
21697 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21700 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
21701 * config/i386/i386.c (ix86_stack_protect_guard): Use
21702 ix86_stack_protect_guard_symbol_str to generate varible declaration.
21703 * doc/invoke.texi (x86 Options): Document
21704 -mstack-protector-guard-symbol= option.
21706 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21708 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
21709 * config/i386/i386.c (ix86_split_stack_guard): New function.
21710 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
21711 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
21712 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
21713 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
21714 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
21715 (split_stack_space_check): Call ix86_split_stack_guard.
21717 2017-08-10 Martin Sebor <msebor@redhat.com>
21719 * print-tree.c (print_node): Print location using the established
21721 Replace spaces with colons.
21722 (debug_raw, debug): Ditto.
21724 2017-08-10 Martin Sebor <msebor@redhat.com>
21727 * pretty-print.c (pp_format): Correct the handling of %s precision.
21729 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
21732 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21734 (ix86_finalize_stack_frame_flags): This. Also clear
21735 frame_pointer_needed if -fno-omit-frame-pointer is used without
21737 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21738 with ix86_finalize_stack_frame_flags.
21739 (ix86_expand_epilogue): Likewise.
21740 (ix86_expand_split_stack_prologue): Likewise.
21741 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
21743 2017-08-10 Martin Liska <mliska@suse.cz>
21746 * c-attribs.c (handle_target_attribute):
21747 Report warning for an empty string argument of target attribute.
21749 2017-08-09 Jakub Jelinek <jakub@redhat.com>
21752 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
21754 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
21755 or DECL_NONLOCAL labels.
21756 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
21757 or DECL_NONLOCAL labels here.
21759 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
21761 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
21762 to indicate when early gimple folding has been disabled.
21763 (rs6000_gimple_fold_builtin): Add debug content.
21764 (rs6000_invalid_builtin): Fix whitespace.
21765 (rs6000_expand_builtin): Fix whitespace.
21766 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
21768 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
21771 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
21772 SAVE_MULTIPLE if not all the registers that saves, should be saved.
21774 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
21776 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
21777 (qdf24xx): Likewise.
21778 * config/aarch64/aarch64.md: Include falkor.md.
21779 * config/aarch64/falkor.md: New.
21781 2017-08-09 Marek Polacek <polacek@redhat.com>
21784 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
21785 * diagnostic.c (emit_diagnostic): Add a comment.
21786 (emit_diagnostic_valist): New function.
21788 2017-08-09 Marek Polacek <polacek@redhat.com>
21791 * input.c (make_location): New overload.
21792 * input.h (make_location): Declare.
21794 2017-08-08 Alan Modra <amodra@gmail.com>
21795 H.J. Lu <hongjiu.lu@intel.com>
21798 * gcc.c (NO_PIE_SPEC): Delete.
21799 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21801 (LINK_PIE_SPEC): ..to here.
21802 (LINK_COMMAND_SPEC): Support -no-pie.
21803 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21804 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21806 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21807 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21808 (ENDFILE_CRTEND_SPEC): Similarly.
21810 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21813 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21814 (mstack-protector-guard-offset=): Ditto.
21815 * config/i386/i386.c (ix86_option_override): Handle
21816 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21818 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21819 ix86_stack_protect_guard_offset variables.
21820 (TARGET_STACK_PROTECT_GUARD): Always define.
21821 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21822 and -mstack-protector-guard-offset= options.
21824 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21826 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21827 boundary case for the last candidate.
21829 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21831 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21832 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21834 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21836 PR middle-end/19706
21837 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21838 * config/aarch64/aarch64-builtins.c
21839 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21840 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21841 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21843 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21844 Andrew Pinski <pinskia@gmail.com>
21846 PR middle-end/19706
21847 * internal-fn.def (XORSIGN): New.
21848 * optabs.def (xorsign_optab): New.
21849 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21850 (convert_expand_mult_copysign): New.
21851 (pass_optimize_widening_mul::execute): Call
21852 convert_expand_mult_copysign.
21854 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21856 PR tree-optimization/81354
21857 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21858 Insert on edges rather than explicitly creating landing pads.
21859 (analyze_candidates_and_replace): Commit edge inserts.
21861 2017-08-08 Richard Biener <rguenther@suse.de>
21863 PR middle-end/81719
21864 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21865 (expand_simple_operations): Also look through ADDR_EXPRs with
21866 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21868 2017-08-08 Richard Biener <rguenther@suse.de>
21870 PR tree-optimization/81723
21871 * tree-vect-slp.c (struct bst_traits): New hash traits.
21872 (bst_fail): New global.
21873 (vect_build_slp_tree_2): New worker, split out from ...
21874 (vect_build_slp_tree): ... this now wrapping it with using
21875 bst_fail set to cache SLP tree build fails. Properly handle
21877 (vect_analyze_slp_instance): Allocate and free bst_fail.
21879 2017-08-08 Martin Liska <mliska@suse.cz>
21882 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21883 LABEL_DECLs that can be from a different function.
21885 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21887 PR tree-optimization/81744
21888 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21889 loop's number of iterations.
21891 2017-08-08 Martin Liska <mliska@suse.cz>
21893 * asan.c: Include header files.
21894 * attribs.c (build_decl_attribute_variant): New function moved
21896 (build_type_attribute_qual_variant): Likewise.
21897 (cmp_attrib_identifiers): Likewise.
21898 (simple_cst_list_equal): Likewise.
21899 (omp_declare_simd_clauses_equal): Likewise.
21900 (attribute_value_equal): Likewise.
21901 (comp_type_attributes): Likewise.
21902 (build_type_attribute_variant): Likewise.
21903 (lookup_ident_attribute): Likewise.
21904 (remove_attribute): Likewise.
21905 (merge_attributes): Likewise.
21906 (merge_type_attributes): Likewise.
21907 (merge_decl_attributes): Likewise.
21908 (merge_dllimport_decl_attributes): Likewise.
21909 (handle_dll_attribute): Likewise.
21910 (attribute_list_equal): Likewise.
21911 (attribute_list_contained): Likewise.
21912 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21913 (lookup_attribute_by_prefix): Likewise.
21914 * bb-reorder.c: Include header files.
21915 * builtins.c: Likewise.
21916 * calls.c: Likewise.
21917 * cfgexpand.c: Likewise.
21918 * cgraph.c: Likewise.
21919 * cgraphunit.c: Likewise.
21920 * convert.c: Likewise.
21921 * dwarf2out.c: Likewise.
21922 * final.c: Likewise.
21923 * fold-const.c: Likewise.
21924 * function.c: Likewise.
21925 * gimple-expr.c: Likewise.
21926 * gimple-fold.c: Likewise.
21927 * gimple-pretty-print.c: Likewise.
21928 * gimple.c: Likewise.
21929 * gimplify.c: Likewise.
21930 * hsa-common.c: Likewise.
21931 * hsa-gen.c: Likewise.
21932 * internal-fn.c: Likewise.
21933 * ipa-chkp.c: Likewise.
21934 * ipa-cp.c: Likewise.
21935 * ipa-devirt.c: Likewise.
21936 * ipa-fnsummary.c: Likewise.
21937 * ipa-inline.c: Likewise.
21938 * ipa-visibility.c: Likewise.
21940 * lto-cgraph.c: Likewise.
21941 * omp-expand.c: Likewise.
21942 * omp-general.c: Likewise.
21943 * omp-low.c: Likewise.
21944 * omp-offload.c: Likewise.
21945 * omp-simd-clone.c: Likewise.
21946 * opts-global.c: Likewise.
21947 * passes.c: Likewise.
21948 * predict.c: Likewise.
21949 * sancov.c: Likewise.
21950 * sanopt.c: Likewise.
21951 * symtab.c: Likewise.
21952 * toplev.c: Likewise.
21953 * trans-mem.c: Likewise.
21954 * tree-chkp.c: Likewise.
21955 * tree-eh.c: Likewise.
21956 * tree-into-ssa.c: Likewise.
21957 * tree-object-size.c: Likewise.
21958 * tree-parloops.c: Likewise.
21959 * tree-profile.c: Likewise.
21960 * tree-ssa-ccp.c: Likewise.
21961 * tree-ssa-live.c: Likewise.
21962 * tree-ssa-loop.c: Likewise.
21963 * tree-ssa-sccvn.c: Likewise.
21964 * tree-ssa-structalias.c: Likewise.
21965 * tree-ssa.c: Likewise.
21966 * tree-streamer-in.c: Likewise.
21967 * tree-vectorizer.c: Likewise.
21968 * tree-vrp.c: Likewise.
21969 * tsan.c: Likewise.
21970 * ubsan.c: Likewise.
21971 * varasm.c: Likewise.
21972 * varpool.c: Likewise.
21973 * tree.c: Remove functions moved to attribs.[ch].
21974 * tree.h: Likewise.
21975 * config/aarch64/aarch64.c: Add attrs.h header file.
21976 * config/alpha/alpha.c: Likewise.
21977 * config/arc/arc.c: Likewise.
21978 * config/arm/arm.c: Likewise.
21979 * config/avr/avr.c: Likewise.
21980 * config/bfin/bfin.c: Likewise.
21981 * config/c6x/c6x.c: Likewise.
21982 * config/cr16/cr16.c: Likewise.
21983 * config/cris/cris.c: Likewise.
21984 * config/darwin.c: Likewise.
21985 * config/epiphany/epiphany.c: Likewise.
21986 * config/fr30/fr30.c: Likewise.
21987 * config/frv/frv.c: Likewise.
21988 * config/ft32/ft32.c: Likewise.
21989 * config/h8300/h8300.c: Likewise.
21990 * config/i386/winnt.c: Likewise.
21991 * config/ia64/ia64.c: Likewise.
21992 * config/iq2000/iq2000.c: Likewise.
21993 * config/lm32/lm32.c: Likewise.
21994 * config/m32c/m32c.c: Likewise.
21995 * config/m32r/m32r.c: Likewise.
21996 * config/m68k/m68k.c: Likewise.
21997 * config/mcore/mcore.c: Likewise.
21998 * config/microblaze/microblaze.c: Likewise.
21999 * config/mips/mips.c: Likewise.
22000 * config/mmix/mmix.c: Likewise.
22001 * config/mn10300/mn10300.c: Likewise.
22002 * config/moxie/moxie.c: Likewise.
22003 * config/msp430/msp430.c: Likewise.
22004 * config/nds32/nds32-isr.c: Likewise.
22005 * config/nds32/nds32.c: Likewise.
22006 * config/nios2/nios2.c: Likewise.
22007 * config/nvptx/nvptx.c: Likewise.
22008 * config/pa/pa.c: Likewise.
22009 * config/pdp11/pdp11.c: Likewise.
22010 * config/powerpcspe/powerpcspe.c: Likewise.
22011 * config/riscv/riscv.c: Likewise.
22012 * config/rl78/rl78.c: Likewise.
22013 * config/rx/rx.c: Likewise.
22014 * config/s390/s390.c: Likewise.
22015 * config/sh/sh.c: Likewise.
22016 * config/sol2.c: Likewise.
22017 * config/sparc/sparc.c: Likewise.
22018 * config/spu/spu.c: Likewise.
22019 * config/stormy16/stormy16.c: Likewise.
22020 * config/tilegx/tilegx.c: Likewise.
22021 * config/tilepro/tilepro.c: Likewise.
22022 * config/v850/v850.c: Likewise.
22023 * config/vax/vax.c: Likewise.
22024 * config/visium/visium.c: Likewise.
22025 * config/xtensa/xtensa.c: Likewise.
22027 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
22030 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
22031 constraints since the -mupper-regs-* switches have been
22033 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
22034 into a vector from a double word element that was extracted from
22035 another vector, and eliminate extra XXPERMDI instructions.
22036 (vsx_concat_<mode>_2): Likewise.
22037 (vsx_concat_<mode>_3): Likewise.
22038 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
22039 concat to allow optimizing inserts from previous extracts.
22041 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
22043 * config/i386/i386.c (ix86_stack_protect_guard): Generate
22044 memory reference to a SSP offset in TLS address space.
22045 (ix86_print_operand) <case '@'>: Remove.
22046 (ix86_print_operand_punct_valid_p): Remove '@' code.
22047 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
22048 UNSPEC_SP_TLS_TEST.
22049 (stack_tls_protect_set_<mode>): Remove.
22050 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
22051 (stack_tls_protect_test_<mode>): Remove.
22052 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
22054 2017-08-07 Olivier Hainque <hainque@adacore.com>
22057 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
22059 2017-08-07 Douglas Rupp <rupp@adacore.com>
22061 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
22062 variable was referenced as multidir in command.
22064 2017-08-07 Jakub Jelinek <jakub@redhat.com>
22067 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
22070 2017-08-07 Martin Liska <mliska@suse.cz>
22072 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
22073 * config/rl78/rl78.c: Add include of attribs.h.
22074 * config/sh/sh.c: Likewise.
22075 * config/v850/v850.c: Likewise.
22077 2017-08-07 Tom de Vries <tom@codesourcery.com>
22079 PR middle-end/78266
22080 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
22082 2017-08-07 Martin Liska <mliska@suse.cz>
22084 * config/mips/mips.c: Include attribs.h.
22086 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
22089 * doc/invoke.texi: Document change in behvaior for -Ofast for
22092 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
22094 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
22096 (aarch64_pop_regs): Likewise.
22097 (aarch64_gen_load_pair): Likewise.
22098 (aarch64_save_callee_saves): Likewise.
22099 (aarch64_restore_callee_saves): Likewise.
22101 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22103 * config/i386/i386.c: Revert the last change.
22105 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22108 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
22110 (ix86_finalize_stack_frame_flags): This. Also clear
22111 frame_pointer_needed if -fno-omit-frame-pointer is used without
22113 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
22114 with ix86_finalize_stack_frame_flags.
22115 (ix86_expand_epilogue): Likewise.
22116 (ix86_expand_split_stack_prologue): Likewise.
22118 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22121 * config/i386/i386.c (get_builtin_code_for_version): Set priority
22122 to P_AES for Westmere.
22124 2017-08-07 Jonathan Yong <10walls@gmail.com>
22126 * config/i386/mingw.opt (fset-stack-executable): Removed.
22127 * config/i386/cygming.opt (fset-stack-executable): Moved
22129 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
22131 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
22133 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
22135 2017-08-07 Marek Polacek <polacek@redhat.com>
22137 PR middle-end/81737
22138 * fold-const.c (fold_indirect_ref_1): Check type_domain.
22140 2017-08-07 Martin Liska <mliska@suse.cz>
22142 * attribs.h (canonicalize_attr_name): New function.
22143 (cmp_attribs): Move from c-format.c and adjusted.
22144 (is_attribute_p): Moved from tree.h.
22145 * tree-inline.c: Add new includes.
22146 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
22147 (private_is_attribute_p): Remove.
22148 (private_lookup_attribute): Likewise.
22149 (private_lookup_attribute_by_prefix): Simplify.
22150 (remove_attribute): Use is_attribute_p.
22151 * tree.h: Remove removed declarations.
22153 2017-08-07 Jakub Jelinek <jakub@redhat.com>
22155 PR middle-end/81698
22156 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
22157 instead of computing it in the function. Formatting fix.
22158 (expand_case): Don't rely on default_edge being the first edge,
22159 clear it if removing it, pass default_edge to
22160 emit_case_dispatch_table.
22161 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
22164 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
22166 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
22167 insn in the function, emit NOP after the insn.
22169 2017-08-06 Tom de Vries <tom@codesourcery.com>
22171 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
22174 2017-08-06 Tom de Vries <tom@codesourcery.com>
22176 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
22179 2017-08-04 Yury Gribov <tetra2005@gmail.com>
22181 PR tree-optimization/57371
22182 * match.pd: New pattern.
22184 2017-08-04 Marek Polacek <polacek@redhat.com>
22186 PR middle-end/81695
22187 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
22188 perform the computation in offset_int.
22190 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22192 PR tree-optimization/81136
22193 * tree-vectorizer.h: Include tree-hash-traits.h.
22194 (vec_base_alignments): New typedef.
22195 (vec_info): Add a base_alignments field.
22196 (vect_record_base_alignments): Declare.
22197 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
22199 (DR_IS_CONDITIONAL_IN_STMT): New macro.
22200 (create_data_ref): Add an is_conditional_in_stmt argument.
22201 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
22202 the is_conditional_in_stmt field.
22203 (data_ref_loc): Add an is_conditional_in_stmt field.
22204 (get_references_in_stmt): Set the is_conditional_in_stmt field.
22205 (find_data_references_in_stmt): Update call to create_data_ref.
22206 (graphite_find_data_references_in_stmt): Likewise.
22207 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
22208 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
22209 (vect_record_base_alignment): New function.
22210 (vect_record_base_alignments): Likewise.
22211 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
22212 for nested statements even if we fail to compute a misalignment.
22213 Use pooled base alignments for unconditional references.
22214 (vect_find_same_alignment_drs): Compare base addresses instead
22216 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
22217 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
22219 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22221 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
22222 Add an explicit name for the enum. Use auto_vec for slp_instances
22223 and grouped_stores.
22224 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
22226 (_bb_vec_info): Add a constructor and destructor.
22227 (vinfo_for_stmt): Return NULL for uids of -1 as well.
22228 (destroy_loop_vec_info): Delete.
22229 (vect_destroy_datarefs): Likewise.
22230 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
22231 (vec_info::vec_info): New function.
22232 (vec_info::~vec_info): Likewise.
22233 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
22234 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
22235 destroy_loop_vec_info.
22236 * tree-vect-loop.c (new_loop_vec_info): Replace with...
22237 (_loop_vec_info::_loop_vec_info): ...this.
22238 (destroy_loop_vec_info): Replace with...
22239 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
22240 the stmt_vec_infos. Leave handling of vec_info information to its
22241 destructor. Remove explicit vector releases.
22242 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
22243 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
22244 * tree-vect-slp.c (new_bb_vec_info): Replace with...
22245 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
22246 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
22247 (destroy_bb_vec_info): Replace with...
22248 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
22249 information to its destructor.
22250 (vect_slp_analyze_bb_1): Use new and delete instead of
22251 new_bb_vec_info and destroy_bb_vec_info.
22252 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
22255 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22257 * tree-data-ref.h (subscript): Add access_fn field.
22258 (data_dependence_relation): Add could_be_independent_p.
22259 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
22260 (same_access_functions): Move to tree-data-ref.c.
22261 * tree-data-ref.c (ref_contains_union_access_p): New function.
22262 (access_fn_component_p): Likewise.
22263 (access_fn_components_comparable_p): Likewise.
22264 (dr_analyze_indices): Add a reference to access_fn_component_p.
22265 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
22267 (constant_access_functions): Likewise.
22268 (add_other_self_distances): Likewise.
22269 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
22270 (initialize_data_dependence_relation): Use XCNEW and remove
22271 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
22272 of access functions that have the same type. Allow the
22273 subsequence to end with different bases in some circumstances.
22274 Record the chosen access functions in SUB_ACCESS_FN.
22275 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
22276 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
22277 (subscript_dependence_tester_1): Likewise dra and drb.
22278 (build_classic_dist_vector): Update calls accordingly.
22279 (subscript_dependence_tester): Likewise.
22280 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
22281 DDR_COULD_BE_INDEPENDENT_P.
22282 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
22283 comp_alias_ddrs instead of may_alias_ddrs.
22284 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
22286 (vect_analyze_data_ref_dependence): Use it if
22287 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
22288 distance vectors if that fails.
22289 (dependence_distance_ge_vf): New function.
22290 (vect_prune_runtime_alias_test_list): Use it. Don't clear
22291 LOOP_VINFO_MAY_ALIAS_DDRS.
22293 2017-08-04 Richard Biener <rguenther@suse.de>
22295 PR middle-end/81705
22296 * fold-const.c (fold_binary_loc): Properly restrict
22297 minus_var0 && minus_var1 case when associating undefined overflow
22300 2017-08-04 Tom de Vries <tom@codesourcery.com>
22302 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
22304 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22306 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22307 Don't start diagnostic messages with a capital letter.
22308 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22310 (rs6000_invalid_builtin): Likewise.
22311 (rs6000_trampoline_init): Likewise.
22313 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22316 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
22317 after setting changeable df flags.
22319 2017-08-03 Richard Biener <rguenther@suse.de>
22321 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
22322 up if the use is in USE - X.
22324 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
22326 * toplev.c (dumpfile.h): New include.
22327 (internal_error_reentered): New static function. Use it...
22328 (internal_error_function): ...here to handle reentered internal_error.
22330 2017-08-03 Richard Biener <rguenther@suse.de>
22332 PR middle-end/81148
22333 * fold-const.c (split_tree): Add minus_var and minus_con
22334 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
22335 here but always use minus_*.
22336 (associate_trees): Assert we never associate with MINUS_EXPR
22337 and NULL first operand. Do not recurse for PLUS_EXPR operands
22338 when associating as MINUS_EXPR either.
22339 (fold_binary_loc): Track minus_var and minus_con.
22341 2017-08-03 Tom de Vries <tom@codesourcery.com>
22344 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
22345 ACCEL_COMPILER, apply finish_options on
22346 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
22348 2017-08-03 Tom de Vries <tom@codesourcery.com>
22351 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
22352 function_entry_patch_area_size > 0.
22354 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22357 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
22358 instead of 10??LU, perform unit multiplication in wide_int,
22359 don't change alloc_object_size_limit if the limit is larger
22362 PR tree-optimization/81655
22363 PR tree-optimization/81588
22364 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
22365 the case when ranges[i].low and high are 1 for unsigned type with
22368 PR middle-end/81052
22369 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
22370 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
22372 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22374 * tree-vrp.h: Add include guard.
22376 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
22379 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
22380 (ud2): New insn pattern.
22381 * config/i386/i386.c (ix86_expand_epilogue):
22382 For naked functions, generate ud2 instead of trap insn.
22384 2017-08-02 Marek Polacek <polacek@redhat.com>
22387 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
22389 2017-08-02 Tom de Vries <tom@codesourcery.com>
22390 Cesar Philippidis <cesar@codesourcery.com>
22392 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
22393 Add missing edge probabilities.
22395 2017-08-02 Tamar Christina <tamar.christina@arm.com>
22397 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
22398 Correct endianness.
22400 2017-08-02 Jakub Jelinek <jakub@redhat.com>
22402 PR middle-end/79499
22403 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
22404 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
22405 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
22406 prologue_seq sequences - if any.
22408 2017-08-02 Richard Biener <rguenther@suse.de>
22410 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
22411 via vectors if supported, integer extracts via punning if supported
22412 or otherwise vector extracts.
22414 2017-08-02 Richard Biener <rguenther@suse.de>
22416 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
22418 (bitmap_insert_into_set): ... this.
22420 2017-08-02 Richard Biener <rguenther@suse.de>
22422 PR tree-optimization/81633
22424 2015-08-17 Alan Hayward <alan.hayward@arm.com>
22426 PR tree-optimization/71752
22427 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
22429 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
22431 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
22432 (machine_function::call_ms2sysv_pad_out): Remove field.
22433 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
22434 (ix86_compute_frame_layout): Likewise.
22436 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
22439 * config/i386/i386.c (ix86_set_func_type): Disallow naked
22440 attribute with interrupt attribute.
22442 2017-08-01 Andrew Pinski <apinski@cavium.com>
22444 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
22445 BIT_INSERT_EXPR's operand 1
22446 to see if the types precision matches.
22448 2017-08-01 Martin Liska <mliska@suse.cz>
22450 PR middle-end/70140
22451 * builtins.c (expand_builtin_memcpy_args): Remove.
22452 (expand_builtin_memcpy): Call newly added function
22453 expand_builtin_memory_copy_args.
22454 (expand_builtin_memcpy_with_bounds): Likewise.
22455 (expand_builtin_mempcpy): Remove last argument.
22456 (expand_builtin_mempcpy_with_bounds): Likewise.
22457 (expand_builtin_memory_copy_args): New function created from
22458 expand_builtin_mempcpy_args with small modifications.
22459 (expand_builtin_mempcpy_args): Remove.
22460 (expand_builtin_stpcpy): Remove unused argument.
22461 (expand_builtin): Likewise.
22462 (expand_builtin_with_bounds): Likewise.
22464 2017-08-01 Martin Liska <mliska@suse.cz>
22467 Make mempcpy more optimal (PR middle-end/70140).
22469 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22472 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
22473 __builtin_vec_cmpne verify both arguments are compatible vectors
22474 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
22475 verify arg1_type is a pointer or array type. For __builtin_vec_st,
22476 move computation of aligned to after checking the argument types.
22480 * config/rs6000/vsx.md (vextract_fp_from_shorth,
22481 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
22484 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
22485 Doug Rupp <rupp@adacore.com>
22486 Olivier Hainque <hainque@adacore.com>
22488 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
22489 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
22491 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
22492 for TARGET_OS_CPP_BUILTIN.
22493 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
22494 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
22496 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
22497 passing required abi options to the assembler for EABI configurations.
22498 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
22499 of .text.hot and .text.unlikely sections for kernel modules when
22500 using ARM style exceptions.
22501 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
22502 options. Add EXTRA_CC1_SPEC.
22503 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
22505 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
22507 (ARM_TARGET2_DWARF_FORMAT): Define.
22508 * config/arm/t-vxworks: Adjust multilib control to removal of the
22509 Diab command line options.
22511 2017-08-01 Martin Liska <mliska@suse.cz>
22513 PR gcov-profile/81561
22514 * gcov.c (unblock): Make unblocking safe as we need to preserve
22515 index correspondence of blocks and block_lists.
22517 2017-08-01 Richard Biener <rguenther@suse.de>
22519 PR tree-optimization/81181
22520 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
22521 (compute_antic): ... end of iteration here.
22523 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
22525 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
22526 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
22527 (ftree-slp-vectorize): Likewise.
22528 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
22529 can no longer be set independent of flag_tree_loop_vectorize.
22530 * omp-general.c (emp_max_vf): Likewise.
22531 * opts.c (enable_fdo_optimizations): Remove references to
22532 flag_tree_vectorize, these are now implicit.
22533 (common_handle_option): Remove handling for OPT_ftree_vectorize,
22534 and leave it for the options machinery.
22536 2017-08-01 Martin Liska <mliska@suse.cz>
22538 PR middle-end/70140
22539 * builtins.c (expand_builtin_memcpy_args): Remove.
22540 (expand_builtin_memcpy): Call newly added function
22541 expand_builtin_memory_copy_args.
22542 (expand_builtin_memcpy_with_bounds): Likewise.
22543 (expand_builtin_mempcpy): Remove last argument.
22544 (expand_builtin_mempcpy_with_bounds): Likewise.
22545 (expand_builtin_memory_copy_args): New function created from
22546 expand_builtin_mempcpy_args with small modifications.
22547 (expand_builtin_mempcpy_args): Remove.
22548 (expand_builtin_stpcpy): Remove unused argument.
22549 (expand_builtin): Likewise.
22550 (expand_builtin_with_bounds): Likewise.
22552 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22555 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
22556 print "ds:" only for immediates in generic address space.
22558 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22561 * config/i386/i386.c (ix86_funciton_naked): New prototype.
22562 (ix86_function_ok_for_sibcall): Return false for naked functions.
22564 2017-08-01 Richard Biener <rguenther@suse.de>
22566 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
22567 (compute_antic): Seed worklist with exit block predecessors.
22568 * cfganal.c (dfs_find_deadend): For a cycle return the source
22569 of the edge closing it.
22571 2017-08-01 Tamar Christina <tamar.christina@arm.com>
22573 * config/aarch64/aarch64.c
22574 (aarch64_can_const_movi_rtx_p): Move 0 check.
22576 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22578 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
22579 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
22581 * match.pd: Ditto in address comparison pattern.
22583 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22585 PR tree-optimization/81627
22586 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
22587 closed ssa form for store-store chain.
22589 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22591 PR tree-optimization/81620
22592 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
22593 for store-store chain.
22595 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22597 PR tree-optimization/81588
22598 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
22599 ranges[i].in_p, invert comparison code ccode. For >/>=,
22600 swap rhs1 and rhs2 and comparison code unconditionally,
22601 for </<= don't do that. Don't swap rhs1/rhs2 again if
22602 ranges[i].in_p, instead invert comparison code ccode if
22603 opcode or oe->rank is BIT_IOR_EXPR.
22606 * optabs.def (vec_extract_optab, vec_init_optab): Change from
22607 a direct optab to conversion optab.
22608 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
22609 with GET_MODE_INNER as last argument instead of optab_handler.
22610 * expmed.c (extract_bit_field_1): Likewise. Use vector from
22611 vector extraction if possible and optab is available.
22612 * expr.c (store_constructor): Use convert_optab_handler instead
22613 of optab_handler. Use vector initialization from smaller
22614 vectors if possible and optab is available.
22615 * tree-vect-stmts.c (vectorizable_load): Likewise.
22616 * doc/md.texi (vec_extract, vec_init): Document that the optabs
22617 now have two modes.
22618 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
22619 of vec_init from half-sized vectors with the same element mode.
22620 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
22621 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
22622 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
22623 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
22624 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
22625 after mode in gen_vec_extract* calls.
22626 (vec_extract<mode>): Renamed to ...
22627 (vec_extract<mode><ssescalarmodelower>): ... this.
22628 (vec_extract<mode><ssehalfvecmodelower>): New expander.
22629 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
22630 element mode after mode in gen_vec_init* calls.
22631 (VEC_INIT_HALF_MODE): New mode iterator.
22632 (vec_init<mode>): Renamed to ...
22633 (vec_init<mode><ssescalarmodelower>): ... this.
22634 (vec_init<mode><ssehalfvecmodelower>): New expander.
22635 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
22636 (vec_extractv2sfsf): ... this.
22637 (vec_initv2sf): Renamed to ...
22638 (vec_initv2sfsf): ... this.
22639 (vec_extractv2si): Renamed to ...
22640 (vec_extractv2sisi): ... this.
22641 (vec_initv2si): Renamed to ...
22642 (vec_initv2sisi): ... this.
22643 (vec_extractv4hi): Renamed to ...
22644 (vec_extractv4hihi): ... this.
22645 (vec_initv4hi): Renamed to ...
22646 (vec_initv4hihi): ... this.
22647 (vec_extractv8qi): Renamed to ...
22648 (vec_extractv8qiqi): ... this.
22649 (vec_initv8qi): Renamed to ...
22650 (vec_initv8qiqi): ... this.
22651 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
22652 (vec_init<mode>): Renamed to ...
22653 (vec_init<mode><VEC_base_l>): ... this.
22654 (vec_extract<mode>): Renamed to ...
22655 (vec_extract<mode><VEC_base_l>): ... this.
22656 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
22657 (vec_initv2sfsf): ... this.
22658 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
22659 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22660 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
22661 element mode after mode in gen_vec_init* calls.
22662 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
22663 (vec_init<mode><Vel>): ... this.
22664 (vec_extract<mode>): Renamed to ...
22665 (vec_extract<mode><Vel>): ... this.
22666 * config/aarch64/iterators.md (Vel): New mode attribute.
22667 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
22668 Add element mode after mode in gen_vec_extract* calls.
22669 * config/s390/vector.md (non_vec_l): New mode attribute.
22670 (vec_extract<mode>): Renamed to ...
22671 (vec_extract<mode><non_vec_l>): ... this.
22672 (vec_init<mode>): Renamed to ...
22673 (vec_init<mode><non_vec_l>): ... this.
22674 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
22675 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
22677 * config/arm/iterators.md (V_elem_l): New mode attribute.
22678 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
22679 (vec_extract<mode><V_elem_l>): ... this.
22680 (vec_extractv2di): Renamed to ...
22681 (vec_extractv2didi): ... this.
22682 (vec_init<mode>): Renamed to ...
22683 (vec_init<mode><V_elem_l>): ... this.
22684 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
22685 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
22686 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
22687 Add element mode after gen_vec_extract* calls.
22688 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
22689 (vec_init<mode><unitmode>): ... this.
22690 (vec_extract<mode>): Renamed to ...
22691 (vec_extract<mode><unitmode>): ... this.
22692 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
22693 (vec_init<mode><unitmode>): ... this.
22694 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
22695 (vec_initv2sfsf): ... this.
22696 (vec_extractv2sf): Renamed to ...
22697 (vec_extractv2sfsf): ... this.
22698 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
22699 Add element mode after gen_vec_extract* calls.
22700 * config/mips/mips.md (unitmode): New mode iterator.
22701 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
22702 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
22703 * config/spu/spu.md (inner_l): New mode attribute.
22704 (vec_init<mode>): Renamed to ...
22705 (vec_init<mode><inner_l>): ... this.
22706 (vec_extract<mode>): Renamed to ...
22707 (vec_extract<mode><inner_l>): ... this.
22708 * config/sparc/sparc.md (veltmode): New mode iterator.
22709 (vec_init<VMALL:mode>): Renamed to ...
22710 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
22711 * config/ia64/vect.md (vec_initv2si): Renamed to ...
22712 (vec_initv2sisi): ... this.
22713 (vec_initv2sf): Renamed to ...
22714 (vec_initv2sfsf): ... this.
22715 (vec_extractv2sf): Renamed to ...
22716 (vec_extractv2sfsf): ... this.
22717 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
22718 (vec_init<mode>): Renamed to ...
22719 (vec_init<mode><VEC_base_l>): ... this.
22720 (vec_extract<mode>): Renamed to ...
22721 (vec_extract<mode><VEC_base_l>): ... this.
22722 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
22723 (vec_initv2sfsf): ... this.
22724 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
22725 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22726 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
22727 gen_vec_init* calls.
22729 2017-08-01 Richard Biener <rguenther@suse.de>
22731 PR tree-optimization/81297
22732 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
22733 TREE_OVERFLOW from INTEGER_CSTs.
22735 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
22737 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
22739 2017-07-31 Carl Love <cel@us.ibm.com>
22741 * config/rs6000/rs6000-c: Add support for built-in functions
22742 vector signed char vec_xl_be (signed long long, signed char *);
22743 vector unsigned char vec_xl_be (signed long long, unsigned char *);
22744 vector signed int vec_xl_be (signed long long, signed int *);
22745 vector unsigned int vec_xl_be (signed long long, unsigned int *);
22746 vector signed long long vec_xl_be (signed long long, signed long long *);
22747 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
22748 vector signed short vec_xl_be (signed long long, signed short *);
22749 vector unsigned short vec_xl_be (signed long long, unsigned short *);
22750 vector double vec_xl_be (signed long long, double *);
22751 vector float vec_xl_be (signed long long, float *);
22752 * config/rs6000/altivec.h (vec_xl_be): Add #define.
22753 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
22754 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
22756 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
22757 (altivec_expand_builtin): Add switch statement to call
22758 altivec_expand_xl_be for each builtin.
22759 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
22760 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
22761 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
22762 __builtin_vsx_le_be_v16qi.
22763 * doc/extend.texi: Update the built-in documentation file for the
22764 new built-in functions.
22766 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22769 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
22771 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22773 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22775 * config.gcc: Add z14.
22776 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
22777 CPU model numbers for z13s and z14.
22778 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
22780 * config/s390/s390-opts.h (enum processor_type): Rename
22781 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22782 * config/s390/s390.c (processor_table): Add field for CPU name to
22783 be passed to Binutils.
22784 (s390_asm_output_machine_for_arch): Use the new field in
22785 processor_table for Binutils.
22786 (s390_expand_builtin): Replace arch12 with z14.
22787 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22788 (s390_get_sched_attrmask): Likewise.
22789 (s390_get_unit_mask): Likewise.
22790 * config/s390/s390.opt: Add z14 to processor_type enum.
22792 2017-07-31 Martin Jambor <mjambor@suse.cz>
22795 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22796 regardless of optimization level.
22798 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22799 Martin Liska <mliska@suse.cz>
22801 * predict.def: Remove old comment and adjust probability.
22802 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22803 PREDICT statements.
22805 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22808 * config/i386/i386.c (ix86_function_naked): New function.
22809 (ix86_can_use_return_insn_p): Return false for naked functions.
22810 (ix86_expand_prologue): Skip prologue for naked functions.
22811 (ix86_expand_epilogue): Skip epilogue for naked functions
22812 and emit trap instruction.
22813 (ix86_warn_func_return): New function.
22814 (ix86_attribute_table): Add "naked" attribute specification.
22815 (TARGET_WARN_FUNC_RETURN): Define.
22816 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22818 2017-07-31 Martin Liska <mliska@suse.cz>
22820 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22821 (dump_gimple_bb_header): Always dump BB info.
22822 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22824 2017-07-31 Martin Liska <mliska@suse.cz>
22827 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22828 also with current_function_decl non-null equality.
22830 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22833 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22834 change type to the element type, instead add eltype variable and
22835 use it where we are interested in the element type.
22837 PR tree-optimization/81603
22838 * ipa-polymorphic-call.c
22839 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22840 offset arithmetic in offset_int, bail out if the resulting bit offset
22841 doesn't fit into shwi.
22843 2017-07-31 Martin Liska <mliska@suse.cz>
22845 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22846 (gimplify_save_expr): Fix comment.
22848 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22851 * config/i386/i386.c (ix86_function_arg): Update arguments for
22853 (ix86_compute_frame_layout): Set the initial stack offset to
22854 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22855 INCOMING_FRAME_SP_OFFSET.
22856 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22857 stack before exception handler returns.
22858 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22859 the 'ERROR_CODE' for exception handler.
22861 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22863 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22864 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22865 (ASM_OUTPUT_REG_POP): Ditto.
22866 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22867 instead of asm_fprintf to output pure string.
22869 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22871 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22872 to imported_module_or_decl hook.
22873 (debug_nothing_tree_tree_tree_bool): Remove.
22874 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22875 * debug.c (do_nothing_debug_hooks): Use
22876 debug_nothing_tree_tree_tree_bool_bool instead of
22877 debug_nothing_tree_tree_tree_bool.
22878 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22879 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22880 * sdbout.c (sdb_debug_hooks): Likewise.
22881 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22882 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22884 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22885 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22886 attribute, don't add anything.
22888 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22890 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22891 (fold_build2_stat_loc): Likewise.
22892 (fold_build3_stat_loc): Likewise.
22893 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22894 (fold_build1_loc): Remove macro.
22895 (fold_build2_loc): Likewise.
22896 (fold_build3_loc): Likewise.
22898 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22900 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22901 (gimple_build_debug_bind_source_stat): Likewise.
22902 * gimple.h (gimple_build_debug_bind): Remove macro.
22903 (gimple_build_debug_bind_source): Likewise.
22905 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22907 * bitmap.c (bitmap_alloc): Adjust.
22908 (bitmap_gc_alloc): Likewise.
22909 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22911 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22913 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22914 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22915 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22916 (bitmap_gc_alloc_stat): Likewise.
22917 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22919 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22921 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22922 * rtl.h (shallow_copy_rtx): Remove macro.
22924 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22926 * emit-rtl.c (gen_raw_REG): Adjust.
22927 * gengenrtl.c (gendef): Likewise.
22928 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22929 * rtl.h (rtx_alloc): Remove macro.
22931 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22933 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22934 (build_tree_list_stat): Likewise.
22935 * tree.h (build_tree_list): Remove macro.
22936 (build_tree_list_vec): Likewise.
22938 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22940 * tree.c (make_vector_stat): Remove _stat from name.
22941 (build_vector_stat): Likewise.
22942 * tree.h (make_vector_stat): Remove macro.
22943 (build_vector_stat): Likewise.
22945 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22947 * tree.h (build_var_debug_value): Remove prototype.
22949 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22951 * tree.c (tree_cons_stat): Remove _stat from name.
22952 * tree.h (tree_cons): Remove macro.
22954 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22956 * tree.c (build_vl_exp_stat): Remove _stat from name.
22957 * tree.h (build_vl_exp): Remove macro.
22959 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22961 * tree.c (build_decl_stat): Remove _stat from name.
22962 * tree.h (build_decl): Remove macro.
22964 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22966 * gimple.c (gimple_build_with_ops_stat): Adjust.
22967 (gimple_alloc_stat): Remove _stat from name.
22968 * gimple.h (gimple_alloc): Remove macro.
22970 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22972 * tree.c (make_tree_vec_stat): Remove _stat from name.
22973 (grow_tree_vec_stat): Likewise.
22974 * tree.h (make_tree_vec_stat): Adjust prototype.
22975 (grow_tree_vec_stat): Likewise.
22976 (make_tree_vec): Remove macro.
22977 (grow_tree_vec): Likewise.
22979 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22981 * fold-const.c (fold_build1_stat_loc): Adjust.
22982 (fold_build2_stat_loc): Likewise.
22983 (fold_build3_stat_loc): Likewise.
22984 * tree.c (build0_stat): Remove _stat from name.
22985 (build1_stat): Likewise.
22986 (build2_stat): Likewise.
22987 (build3_stat): Likewise.
22988 (build4_stat): Likewise.
22989 (build5_stat): Likewise.
22990 * tree.h (build1_loc): Remove macro, and rename _stat function
22992 (build2_loc): Likewise.
22993 (build3_loc): Likewise.
22994 (build4_loc): Likewise.
22995 (build5_loc): Likewise.
22997 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22999 * tree.c (make_int_cst_stat): Remove _stat from name.
23000 * tree.h (make_int_cst_stat): Adjust prototype.
23001 (make_int_cst): Remove macro.
23003 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23005 * tree.c (make_tre_binfo_stat): Remove _stat from name.
23006 * tree.h (make_tree_binfo_stat): Adjust prototype.
23007 (make_tree_binfo): Remove.
23009 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23011 * tree.c (copy_node_stat): Rename to copy_node.
23012 (build_distinct_type_copy): Adjust.
23013 * tree.h (copy_node_stat): Adjust prototype.
23014 (copy_node): Remove macro.
23016 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23018 * tree.c (make_node_stat): rename to make_node.
23019 (build_tree_list_stat): Adjust.
23020 (build0_stat): Likewise.
23021 (build2_stat): Likewise.
23022 (build3_stat): Likewise.
23023 (build4_stat): Likewise.
23024 (build5_stat): Likewise.
23025 (build_decl_stat): Likewise.
23026 * tree.h (make_node_stat): Adjust prototype.
23027 (make_node): remove macro.
23029 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
23031 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
23032 (PPC_FEATURE2_SCV): Likewise.
23033 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
23035 2017-07-28 Tamar Christina <tamar.christina@arm.com>
23037 * config/aarch64/aarch64.c
23038 (aarch64_internal_mov_immediate): Add new special pattern.
23039 * config/aarch64/aarch64.md (*movdi_aarch64):
23040 Add reg/32bit const mov case.
23042 2017-07-28 Tamar Christina <tamar.christina@arm.com>
23043 Richard Sandiford <richard.sandiford@linaro.org>
23045 * config/aarch64/aarch64.md (mov<mode>): Generalize.
23046 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
23047 Add integer and movi cases.
23048 (movi-split-hf-df-sf split, fp16): New.
23049 (enabled): Added TARGET_FP_F16INST.
23050 * config/aarch64/iterators.md (GPF_HF): New.
23051 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
23053 2017-07-28 Tamar Christina <tamar.christina@arm.com>
23055 * config/aarch64/aarch64.c
23056 (aarch64_simd_container_mode): Add prototype.
23057 (aarch64_expand_mov_immediate): Add HI support.
23058 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
23059 (aarch64_can_const_movi_rtx_p): New.
23060 (aarch64_preferred_reload_class):
23061 Remove restrictions of using FP registers for certain SIMD operations.
23062 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
23063 (aarch64_valid_floating_const): Add integer move validation.
23064 (aarch64_simd_imm_scalar_p): Remove.
23065 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
23066 (aarch64_legitimate_constant_p): Expand list of supported cases.
23067 * config/aarch64/aarch64-protos.h
23068 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
23069 (aarch64_reinterpret_float_as_int): New.
23070 (aarch64_simd_imm_scalar_p): Remove.
23071 * config/aarch64/constraints.md (Uvi): New.
23072 (Dd): Split into Ds and new Dd.
23073 * config/aarch64/aarch64.md (*movsi_aarch64):
23075 (*movdi_aarch64): Add SIMD mov case.
23077 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23079 * tree-predcom.c: (struct chain): Handle store-store chain in which
23080 stores for elimination only store loop invariant values.
23081 (execute_pred_commoning_chain): Ditto.
23082 (prepare_initializers_chain_store_elim): Ditto.
23083 (prepare_finalizers): Ditto.
23084 (is_inv_store_elimination_chain): New function.
23085 (initialize_root_vars_store_elim_1): New function.
23087 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23089 * tree-predcom.c: Revise general description of the pass.
23090 (enum chain_type): New enum type for store elimination.
23091 (struct chain): New field supporting store elimination.
23092 (struct component): Ditto.
23093 (dump_chain): Dump store-stores chain.
23094 (release_chain): Release resources.
23095 (split_data_refs_to_components): Compute and create component
23096 contains only stores for elimination.
23097 (get_chain_last_ref_at): New function.
23098 (make_invariant_chain): Initialization.
23099 (make_rooted_chain): Specify chain type in parameter and record it.
23100 (add_looparound_copies): Skip for store-stores chain.
23101 (determine_roots_comp): Compute type of chain and pass it to
23103 (initialize_root_vars_store_elim_2): New function.
23104 (finalize_eliminated_stores): New function.
23105 (remove_stmt): Handle store for elimination.
23106 (execute_pred_commoning_chain): Execute predictive commoning on
23107 store-store chains.
23108 (determine_unroll_factor): Skip unroll for store-stores chain.
23109 (prepare_initializers_chain_store_elim): New function.
23110 (prepare_initializers_chain): Hanlde store-store chain.
23111 (prepare_finalizers_chain, prepare_finalizers): New function.
23112 (tree_predictive_commoning_loop): Return integer value indicating
23113 if loop is unrolled or lcssa form is corrupted.
23114 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
23116 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23118 * tree-predcom.c (initialize_root): Delete.
23119 (execute_pred_commoning_chain): Initialize root vars and replace
23120 reference of non-combined chain directly, rather than call above
23123 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23125 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
23126 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
23128 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23130 * tree-predcom.c (struct chain): New field init_seq.
23131 (release_chain): Release init_seq.
23132 (prepare_initializers_chain): Record intialization stmts in above
23134 (insert_init_seqs): New function.
23135 (tree_predictive_commoning_loop): Call insert_init_seqs.
23137 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23139 * tree-predcom.c (determine_roots_comp): Skip trivial components.
23141 2017-07-28 Richard Biener <rguenther@suse.de>
23143 * match.pd: Remove superfluous :c.
23144 * genmatch.c (simplify::id): Add member.
23145 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
23147 (current_id): New global.
23148 (dt_node::parent): Move from ...
23149 (dt_operand::parent): ... here. Add for_id member.
23150 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
23151 (decision_tree::find_node): Relax order requirement when
23152 merging DT_TRUE nodes to ones inbetween the current simplify
23153 and the one we try to merge with. Add diagnostic whenever
23154 we need to enforce pattern order by not merging.
23155 (decision_tree::insert): Set current_id.
23156 (decision_tree::print_node): Dump parent node and for_id.
23157 (parser::last_id): Add member.
23158 (parser::push_simplify): Assign unique id.
23159 (parser::parser): Initialize last_id.
23161 2017-07-28 Martin Liska <mliska@suse.cz>
23164 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
23165 gimple_build_debug_bind.
23167 2017-07-28 Richard Biener <rguenther@suse.de>
23169 PR tree-optimization/81502
23170 * match.pd: Add pattern combining BIT_INSERT_EXPR with
23172 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
23174 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
23175 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
23176 for BIT_FIELD_REF args.
23177 * fold-const.c (make_bit_field_ref): Likewise.
23178 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
23180 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23183 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
23184 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
23185 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
23186 Or it into SANITIZER_UNDEFINED.
23187 * ubsan.c: Include gimple-fold.h and varasm.h.
23188 (ubsan_expand_ptr_ifn): New function.
23189 (instrument_pointer_overflow): New function.
23190 (maybe_instrument_pointer_overflow): New function.
23191 (instrument_object_size): Formatting fix.
23192 (pass_ubsan::execute): Call instrument_pointer_overflow
23193 and maybe_instrument_pointer_overflow.
23194 * internal-fn.c (expand_UBSAN_PTR): New function.
23195 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
23196 * sanitizer.def (__ubsan_handle_pointer_overflow,
23197 __ubsan_handle_pointer_overflow_abort): New builtins.
23198 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
23199 * internal-fn.def (UBSAN_PTR): New internal function.
23200 * opts.c (sanitizer_opts): Add pointer-overflow.
23201 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
23202 * fold-const.c (build_range_check): Compute pointer range check in
23203 integral type if pointer arithmetics would be needed. Formatting
23206 2017-07-28 Martin Liska <mliska@suse.cz>
23209 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
23210 parameters that are of a variable-length.
23212 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
23214 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
23216 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
23217 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
23218 (CRT_CALL_STATIC_FUNCTION): Likewise.
23219 (ASM_DEFAULT_SPEC): New define.
23220 (ASM_SPEC32): Likewise.
23221 (ASM_SPEC64): Likewise.
23222 (ASM_SPEC_COMMON): Likewise.
23223 (ASM_SPEC): Likewise.
23224 (INVALID_64BIT): Likewise.
23225 (LINK_OS_DEFAULT_SPEC): Likewise.
23226 (LINK_OS_SPEC32): Likewise.
23227 (LINK_OS_SPEC64): Likewise.
23228 (POWERPC_LINUX): Likewise.
23229 (PTRDIFF_TYPE): Likewise.
23230 (RESTORE_FP_PREFIX): Likewise.
23231 (RESTORE_FP_SUFFIX): Likewise.
23232 (SAVE_FP_PREFIX): Likewise.
23233 (SAVE_FP_SUFFIX): Likewise.
23234 (SIZE_TYPE): Likewise.
23235 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
23236 (TARGET_64BIT): Likewise.
23237 (TARGET_64BIT): Likewise.
23238 (TARGET_AIX): Likewise.
23239 (WCHAR_TYPE_SIZE): Likewise.
23240 (WCHAR_TYPE): Undefine.
23241 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
23242 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
23243 (CPP_OS_RTEMS_SPEC): Delete.
23244 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
23245 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
23247 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
23249 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23251 PR tree-optimization/81578
23252 * tree-parloops.c (build_new_reduction): Bail out if
23253 reduction_code isn't one of the standard OpenMP reductions.
23254 Move the details printing after that decision.
23256 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23258 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
23259 related to reload_in_progress.
23260 (splat_input_operand): Likewise.
23261 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
23263 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
23265 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
23266 (TARGET_INSTANTIATE_DECLS): Likewise.
23267 (legitimate_indexed_address_p): Delete reload_in_progress code.
23268 (rs6000_debug_legitimate_address_p): Likewise.
23269 (rs6000_eliminate_indexed_memrefs): Likewise.
23270 (rs6000_emit_le_vsx_store): Likewise.
23271 (rs6000_emit_move_si_sf_subreg): Likewise.
23272 (rs6000_emit_move): Likewise.
23273 (register_to_reg_type): Likewise.
23274 (rs6000_pre_atomic_barrier): Likewise.
23275 (rs6000_machopic_legitimize_pic_address): Likewise.
23276 (rs6000_allocate_stack_temp): Likewise.
23277 (rs6000_address_for_fpconvert): Likewise.
23278 (rs6000_address_for_altivec): Likewise.
23279 (rs6000_secondary_memory_needed_rtx): Delete function.
23280 (rs6000_check_sdmode): Likewise.
23281 (rs6000_alloc_sdmode_stack_slot): Likewise.
23282 (rs6000_instantiate_decls): Likewise.
23283 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
23284 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
23285 Delete reload_in_progress.
23286 (*vec_reload_and_plus_<mptrsize>): Likewise.
23287 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
23288 (vsx_div_v2di): Likewise.
23289 (vsx_udiv_v2di): Likewise.
23291 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23293 * config/rs6000/rs6000.opt (mlra): Replace with stub.
23294 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
23295 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
23296 (rs6000_debug_reg_global): Delete print of LRA status.
23297 (rs6000_option_override_internal): Delete dead LRA related code.
23298 (rs6000_lra_p): Delete function.
23299 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
23301 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
23303 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
23304 * config/riscv/rtems.h: New file.
23306 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23307 Sudakshina Das <sudi.das@arm.com>
23309 * config/aarch64/aarch64.md
23310 (define_split for and<mode>3nr_compare): Move
23311 non aarch64_logical_operand to a register.
23312 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
23313 register immediate operand to a register.
23314 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
23316 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23318 PR middle-end/81564
23319 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
23321 2017-07-27 Richard Biener <rguenther@suse.de>
23323 PR tree-optimization/81573
23324 PR tree-optimization/81494
23325 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
23326 multi defuse cycle case.
23328 2017-07-27 Richard Biener <rguenther@suse.de>
23330 PR tree-optimization/81571
23331 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
23334 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
23336 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
23337 earlier and only if MASK_FPU is set. Adjust formatting.
23339 2017-07-27 Martin Liska <mliska@suse.cz>
23341 * opt-functions.awk: Add validation of value of Init.
23342 * optc-gen.awk: Pass new argument.
23344 2017-07-27 Martin Liska <mliska@suse.cz>
23346 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
23347 Fix wrong condition.
23349 2017-07-27 Martin Liska <mliska@suse.cz>
23351 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
23352 BBs and edges seen by autoFDO.
23354 2017-07-27 Richard Biener <rguenther@suse.de>
23356 PR tree-optimization/81502
23357 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
23358 with incompatible but same sized type.
23359 (execute_update_addresses_taken): Likewise.
23361 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
23363 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
23364 flag_tree_loop_vectorize rather than flag_tree_vectorize.
23366 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23369 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
23370 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
23371 Change s_operand to memory_operand.
23373 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
23375 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
23376 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
23377 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
23378 Emit instructions rather than returning an expression. Handle TFmode
23379 and KFmode by casting to TImode.
23380 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
23381 (rs6000_emit_le_vsx_store): Likewise.
23382 * config/rs6000/vsx.md (VSX_TI): New iterator.
23383 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
23384 (*vsx_le_undo_permute_<mode>): Likewise.
23385 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
23386 emit the split sequence.
23387 (*vsx_le_perm_store_<mode>): Likewise.
23389 2017-07-27 Jakub Jelinek <jakub@redhat.com>
23391 PR tree-optimization/81555
23392 PR tree-optimization/81556
23393 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
23394 if true, force CHANGED for the recursive invocation.
23395 (reassociate_bb): Remember original length of ops array, pass
23396 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
23398 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
23399 attributes for noipa attribute. For naked attribute use
23400 lookup_attribute first before lookup_attribute_spec.
23401 * final.c (rest_of_handle_final): Disable IPA RA for functions with
23403 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
23404 for functions with noipa attribute.
23405 (cgraph_externally_visible_p): Return true for functions with noipa
23407 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
23408 for functions with noipa attribute.
23409 * doc/extend.texi: Document noipa function attribute.
23410 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
23411 also for functions with noipa attribute.
23412 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
23414 2017-07-26 Andrew Pinski <apinski@cavium.com>
23416 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
23417 vec_unalign_load_cost and vec_unalign_store_cost.
23419 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
23421 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
23422 -mvsx-small-integer option.
23423 (ISA_3_0_MASKS_IEEE): Likewise.
23424 (OTHER_VSX_VECTOR_MASKS): Likewise.
23425 (POWERPC_MASKS): Likewise.
23426 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
23427 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
23428 code, only testing for DImode being allowed in non-VSX floating
23430 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
23431 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
23433 (rs6000_option_override_internal): Delete -mvsx-small-integer.
23434 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
23435 TARGET_P8_VECTOR test.
23436 (rs6000_secondary_reload_simple_move): Likewise.
23437 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
23438 since TARGET_P9_VECTOR was already tested.
23439 (rs6000_opt_masks): Remove -mvsx-small-integer.
23440 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
23441 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23443 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
23444 test for TARGET_VEXTRACTUB was used, and that uses
23446 (p9 extract splitter): Likewise.
23447 (vsx_extract_<mode>_di_p9): Likewise.
23448 (vsx_extract_<mode>_store_p9): Likewise.
23449 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
23450 for TARGET_P9_VECTOR was used. Delete code that is now dead with
23451 the elimination of TARGET_VSX_SMALL_INTEGER.
23452 (vsx_extract_<mode>_p8): Likewise.
23453 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
23454 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
23455 (vsx_set_<mode>_p9): Likewise.
23456 (vsx_set_v4sf_p9): Likewise.
23457 (vsx_set_v4sf_p9_zero): Likewise.
23458 (vsx_insert_extract_v4sf_p9): Likewise.
23459 (vsx_insert_extract_v4sf_p9_2): Likewise.
23460 * config/rs6000/rs6000.md (sign extend splitter): Change
23461 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
23462 (floatsi<mode>2_lfiwax_mem): Likewise.
23463 (floatunssi<mode>2_lfiwzx_mem): Likewise.
23464 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
23465 since a test for TARGET_P9_VECTOR was used.
23466 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23467 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
23468 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23469 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23470 TARGET_P8_VECTOR test.
23471 (fix_trunc<mode>si2_stfiwx): Likewise.
23472 (fix_trunc<mode>si2_internal): Likewise.
23473 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
23474 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23476 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23477 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23478 TARGET_P8_VECTOR test.
23479 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23480 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
23481 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23483 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23484 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
23485 since a test for TARGET_P9_VECTOR was used.
23486 (splitter for loading small constants): Likewise.
23488 2017-07-26 Andrew Pinski <apinski@cavium.com>
23490 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
23493 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
23496 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
23497 (fp_valid_at): Likewise.
23499 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23501 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
23502 (qdf24xx_addrcost_table): Likewise.
23503 (cortexa57_tunings): Update to use generic_branch_cost.
23504 (cortexa72_tunings): Likewise.
23505 (cortexa73_tunings): Likewise.
23506 (qdf24xx_tunings): Likewise.
23508 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23510 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
23511 (thunderx2t99_branch_cost): Likewise.
23512 (cortexa35_tunings): Update to use generic_branch_cost.
23513 (cortexa53_tunings): Likewise.
23514 (cortexa57_tunings): Likewise.
23515 (cortexa72_tunings): Likewise.
23516 (cortexa73_tunings): Likewise.
23517 (thunderx2t99_tunings): Likewise.
23519 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23521 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
23522 (sparc_option_override): Honour MASK_FSMULD.
23523 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
23524 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
23525 * config/sparc/sparc.opt (mfsmuld): New option.
23526 * doc/invoke.texi (mfsmuld): Document option.
23528 2017-07-26 Marek Polacek <polacek@redhat.com>
23530 PR middle-end/70992
23531 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
23533 2017-07-26 Richard Biener <rguenther@suse.de>
23535 * gimple-match-head.c (do_valueize): Return OP if valueize
23537 (get_def): New helper to get at the def stmt of a SSA name
23538 if valueize allows.
23539 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
23540 do_valueize to get at the def stmt.
23541 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
23543 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
23545 PR middle-end/46932
23546 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
23548 2017-07-26 Martin Liska <mliska@suse.cz>
23551 * function.c (expand_function_start): Make expansion of
23552 nonlocal_goto_save_area after parm_birth_insn.
23554 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23556 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
23557 from all CPU target flags enable members.
23559 2017-07-26 Richard Biener <rguenther@suse.de>
23561 * genmatch.c (dt_simplify::gen): Make iterator vars const.
23562 (decision_tree::gen): Make 'type' const.
23563 (write_predicate): Likewise.
23565 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23567 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
23568 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
23569 (rs6000_option_override_internal): Likewise.
23570 (rs6000_expand_vector_set): Likewise.
23571 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
23572 (TARGET_UPPER_REGS_SF): Likewise.
23573 (TARGET_UPPER_REGS_DI): Likewise.
23574 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
23575 (TARGET_DIRECT_MOVE_64BIT): Likewise.
23576 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
23577 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23578 (Splitters for DI constants in Altivec registers): Likewise.
23579 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
23580 (vsx_set_v4sf_p9): Likewise.
23581 (vsx_set_v4sf_p9_zero): Likewise.
23582 (vsx_insert_extract_v4sf_p9): Likewise.
23583 (vsx_insert_extract_v4sf_p9_2): Likewise.
23585 2017-07-25 Carl Love <cel@us.ibm.com>
23587 * doc/extend.texi: Update the built-in documentation file for the
23588 existing built-in functions
23589 vector signed char vec_cnttz (vector signed char);
23590 vector unsigned char vec_cnttz (vector unsigned char);
23591 vector signed short vec_cnttz (vector signed short);
23592 vector unsigned short vec_cnttz (vector unsigned short);
23593 vector signed int vec_cnttz (vector signed int);
23594 vector unsigned int vec_cnttz (vector unsigned int);
23595 vector signed long long vec_cnttz (vector signed long long);
23596 vector unsigned long long vec_cnttz (vector unsigned long long);
23598 2017-07-25 Andrew Pinski <apinski@cavium.com>
23600 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
23601 accesses where the use is for the first operand of a BIT_INSERT.
23603 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23606 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23607 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23609 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23611 * config/i386/gstabs.h: Delete.
23612 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
23614 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
23616 * config/i386/i386.c (ix86_decompose_address): Do not check for
23617 register RTX when looking at index_reg or base_reg.
23618 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
23620 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
23622 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
23623 to update EH info here.
23625 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23627 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
23629 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23631 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
23633 2017-07-25 Torsten Duwe <duwe@suse.de>
23635 * common.opt: Introduce -fpatchable-function-entry
23636 command line option, and its variables function_entry_patch_area_size
23637 and function_entry_patch_area_start.
23638 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
23639 including a two-value parser.
23640 * target.def (print_patchable_function_entry): New target hook.
23641 * targhooks.h (default_print_patchable_function_entry): New function.
23642 * targhooks.c (default_print_patchable_function_entry): Likewise.
23643 * toplev.c (process_options): Switch off IPA-RA if
23644 patchable function entries are being generated.
23645 * varasm.c (assemble_start_function): Look at the
23646 patchable-function-entry command line switch and current
23647 function attributes and maybe generate NOP instructions by
23648 calling the print_patchable_function_entry hook.
23649 * doc/extend.texi: Document patchable_function_entry attribute.
23650 * doc/invoke.texi: Document -fpatchable_function_entry
23651 command line option.
23652 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
23654 * doc/tm.texi: Re-generate.
23656 2017-07-25 Jakub Jelinek <jakub@redhat.com>
23659 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
23660 TARGET_AVX512DQ rather than TARGET_AVX512BW.
23662 2017-07-25 Tamar Christina <tamar.christina@arm.com>
23664 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
23666 2017-07-25 Richard Biener <rguenther@suse.de>
23668 PR tree-optimization/81455
23669 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
23670 not walk in cycles when looking for guards.
23672 2017-07-25 Richard Biener <rguenther@suse.de>
23674 PR tree-optimization/81529
23675 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
23676 when optimizing backedge uses.
23678 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
23680 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
23682 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
23683 to dl_section_ref. On AIX, append an expression to subtract
23684 the size of the section length to dl_section_ref.
23686 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
23688 * configure.ac: If any of the config.* scripts fail, exit 1.
23689 * configure: Regenerate.
23691 2017-07-25 Richard Biener <rguenther@suse.de>
23693 PR middle-end/81546
23694 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
23695 of immediate uses, be more verbose on errors.
23697 2017-07-25 Richard Biener <rguenther@suse.de>
23699 PR tree-optimization/81510
23700 * tree-vect-loop.c (vect_is_simple_reduction): When the
23701 reduction stmt is not inside the loop bail out.
23703 2017-07-25 Richard Biener <rguenther@suse.de>
23705 PR tree-optimization/81303
23706 * tree-vect-loop-manip.c (vect_loop_versioning): Build
23707 profitability check against LOOP_VINFO_NITERSM1.
23709 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23711 * domwalk.c (cmp_bb_postorder): Simplify.
23712 (sort_bbs_postorder): New function. Use it...
23713 (dom_walker::walk): ...here to optimize common cases.
23715 2017-07-25 Martin Liska <mliska@suse.cz>
23718 * ipa-visibility.c (function_and_variable_visibility): Make the
23719 redirection just on target that supports aliasing.
23720 Fix GNU coding style.
23722 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
23725 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
23727 * config/arm/rtems.h: Likewise.
23728 * config/bfin/rtems.h: Likewise.
23729 * config/i386/rtemself.h: Likewise.
23730 * config/lm32/rtems.h: Likewise.
23731 * config/m32c/rtems.h: Likewise.
23732 * config/m68k/rtemself.h: Likewise.
23733 * config/microblaze/rtems.h: Likewise.
23734 * config/mips/rtems.h: Likewise.
23735 * config/moxie/rtems.h: Likewise.
23736 * config/nios2/rtems.h: Likewise.
23737 * config/powerpcspe/rtems.h: Likewise.
23738 * config/rs6000/rtems.h: Likewise.
23739 * config/rtems.h: Likewise.
23740 * config/sh/rtems.h: Likewise.
23741 * config/sh/rtemself.h: Likewise.
23742 * config/sparc/rtemself.h: Likewise.
23744 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23747 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
23748 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
23749 * tree-ssa-structalias.c (alias_get_name): Same.
23751 2017-07-25 Bin Cheng <bin.cheng@arm.com>
23754 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
23755 instructions if no du chain is found.
23757 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23759 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
23761 2017-07-25 Richard Biener <rguenther@suse.de>
23763 PR middle-end/81505
23764 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
23767 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23769 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
23770 upper-regs options.
23771 (ISA_2_7_MASKS_SERVER): Likewise.
23772 (ISA_3_0_MASKS_IEEE): Likewise.
23773 (OTHER_P8_VECTOR_MASKS): Likewise.
23774 (OTHER_VSX_VECTOR_MASKS): Likewise.
23775 (POWERPC_MASKS): Likewise.
23776 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
23777 duplicate list of options.
23778 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
23779 explicit -mupper-regs options.
23780 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
23781 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
23782 alias for -mupper-regs-df.
23783 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
23784 (rs6000_init_hard_regno_mode_ok): Likewise.
23785 (rs6000_option_override_internal): Likewise.
23786 (rs6000_opt_masks): Likewise.
23787 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
23788 options in terms of whether -mvsx or -mpower8-vector was used.
23789 (TARGET_UPPER_REGS_DI): Likewise.
23790 (TARGET_UPPER_REGS_SF): Likewise.
23791 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23792 -mupper-regs-* options.
23794 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23796 * passes.c (emergency_dump_function): Print some empty lines and a
23797 header before the RTL dump.
23799 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23801 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23803 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23806 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23807 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23809 2017-07-24 Carl Love <cel@us.ibm.com>
23811 * config/rs6000/rs6000-c.c: Add support for built-in functions
23812 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23813 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23814 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23815 vec_extract_fp_from_shortl): Add defines for the two builtins.
23816 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23817 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23819 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23820 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23821 * doc/extend.texi: Update the built-in documentation file for the
23822 new built-in function.
23824 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23827 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23829 * doc/generic.texi: Likewise.
23830 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23831 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23833 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23835 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23836 (aarch64_mls_elt_merge<mode>): Likewise.
23838 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23840 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23841 having __cxa_atexit.
23843 2017-07-23 Michael Collison <michael.collison@arm.com>
23845 * config/arm/arm.c (arm_option_override): Deprecate
23846 use of -mstructure-size-boundary.
23847 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23848 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23850 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23853 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23854 Reduce cost estimate for direct moves.
23856 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23859 * config/i386/i386.c (ix86_option_override_internal): Disable
23860 BMI, BMI2 and TBM instructions for -m16.
23862 2017-07-21 Carl Love <cel@us.ibm.com>
23864 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23865 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23866 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23867 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23868 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23869 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23870 VMULOSW): New enum "unspec" values.
23871 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23872 altivec_vmulosw): New patterns.
23873 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23874 VMULOSW): Add definitions.
23876 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23878 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23879 (qdf24xx): Likewise.
23880 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23881 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23882 (AARCH64_FL_V8_1): Renumber.
23883 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23884 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23885 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23886 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23887 rdma to feature modifiers list.
23889 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23891 PR middle-end/56727
23892 * ipa-visibility (function_and_variable_visibility): Convert
23893 recursive PLT call to direct call if appropriate.
23895 2017-07-21 Andrew Pinski <apinski@cavium.com>
23897 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23898 operand 1 to see if the types precision matches.
23899 * fold-const.c (operand_equal_p): Likewise.
23901 2017-07-21 Richard Biener <rguenther@suse.de>
23903 PR tree-optimization/81303
23904 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23905 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23906 (vect_peeling_hash_get_lowest_cost): Adjust.
23907 (vect_enhance_data_refs_alignment): Likewise. Use
23908 vect_get_peeling_costs_all_drs to compute the penalty for no
23909 peeling to match up costs.
23911 2017-07-21 Richard Biener <rguenther@suse.de>
23913 PR tree-optimization/81500
23914 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23915 we didn't identify a reduction path.
23917 2017-07-21 Tom de Vries <tom@codesourcery.com>
23918 Cesar Philippidis <cesar@codesourcery.com>
23920 PR gcov-profile/81442
23921 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23924 2017-07-21 Tom de Vries <tom@codesourcery.com>
23927 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23929 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23930 nvptx_override_options_after_change.
23932 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23934 * dwarf2out.c (output_file_names): Avoid double testing for
23935 dwarf_version >= 5.
23937 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23939 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23941 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23943 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23945 (try_crossjump_to_edge): Do not punt on partitioned functions.
23947 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23949 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23950 Put all BBs reachable only via paths crossing cold region to cold
23952 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23954 2016-07-21 Richard Biener <rguenther@suse.de>
23956 PR tree-optimization/81303
23957 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23958 into account prologue and epilogue iterations when raising
23959 min_profitable_iters to sth at least covering one vector iteration.
23961 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23963 * config/arm/arm.c (arm_test_cpu_arch_dat):
23966 2017-07-20 Nathan Sidwell <nathan@acm.org>
23968 Remove TYPE_METHODS.
23969 * tree.h (TYPE_METHODS): Delete.
23970 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23971 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23972 (dbxout_type_methods): Scan TYPE_FIELDS.
23973 (dbxout_type): Don't check TYPE_METHODS here.
23974 * function.c (use_register_for_decl): Always ignore register for
23975 class types when not optimizing.
23976 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23977 * tree.c (free_lang_data_in_type): Stitch out member functions and
23978 templates from TYPE_FIELDS.
23979 (build_distinct_type_copy, verify_type_variant,
23980 verify_type): Member fns are on TYPE_FIELDS.
23981 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23982 * tree-pretty-print.c (dump_generic_node): Likewise.
23984 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23987 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23988 V2TImode and V4TImode.
23989 (ix86_expand_vector_extract): Likewise.
23990 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23991 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23992 (ssescalarmode): Handle V4TImode and V2TImode.
23993 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23994 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23995 (VEXTRACTI128_MODE): New mode iterator.
23996 (splitter for *vec_extractv?ti first element): New.
23997 (VEC_INIT_MODE): New mode iterator.
23998 (vec_init<mode>): Consolidate 3 expanders into one using
23999 VEC_INIT_MODE mode iterator.
24001 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
24003 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
24004 non_spilled_static_chain_regno_p.
24006 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
24008 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
24010 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
24012 * bb-reorder.c (connect_traces): Allow copying of blocks within
24015 2017-07-20 Richard Biener <rguenther@suse.de>
24017 * gimple.h (gimple_phi_result): Add gphi * overload.
24018 (gimple_phi_result_ptr): Likewise.
24019 (gimple_phi_arg): Likewise. Adjust index assert to only
24020 allow actual argument accesses rather than all slots available
24022 (gimple_phi_arg_def): Add gphi * overload.
24023 * tree-phinodes.c (make_phi_node): Initialize only actual
24025 (resize_phi_node): Clear memory not covered by old node,
24026 do not initialize excess argument slots.
24027 (reserve_phi_args_for_new_edge): Initialize new argument slot
24030 2017-07-20 Bin Cheng <bin.cheng@arm.com>
24032 PR tree-optimization/81388
24034 2016-07-21 Bin Cheng <bin.cheng@arm.com>
24036 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
24037 by removing computation of may_be_zero.
24039 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24040 Tom de Vries <tom@codesourcery.com>
24042 PR middle-end/81030
24043 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
24044 when gimple level profile disagrees with what RTL expander did.
24046 2017-07-20 Richard Biener <rguenther@suse.de>
24048 PR tree-optimization/61171
24049 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
24050 (vect_analyze_stmt): Add slp instance parameter.
24051 (vectorizable_reduction): Likewise.
24052 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
24053 (vect_is_simple_reduction): Deal with chains not detected
24054 as SLP reduction chain, specifically not properly associated
24055 chains containing a mix of plus/minus.
24056 (get_reduction_op): Remove.
24057 (get_initial_defs_for_reduction): Simplify, pass in whether
24058 this is a reduction chain, pass in the SLP node for the PHIs.
24059 (vect_create_epilog_for_reduction): Get the SLP instance as
24061 (vectorizable_reduction): Get the SLP instance as arg.
24062 During analysis remember the SLP node with the PHIs in the
24063 instance. Simplify getting at the vectorized reduction PHIs.
24064 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
24065 through SLP instance.
24066 (vect_slp_analyze_operations): Likewise.
24067 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
24068 (vect_transform_stmt): Likewise.
24070 2017-07-20 Tom de Vries <tom@codesourcery.com>
24072 PR tree-optimization/81489
24073 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
24074 read of phi arg location to before loop that modifies phi.
24076 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24078 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
24081 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24083 PR middle-end/81331
24084 * except.c (execute): Fix ordering issue.
24086 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
24088 PR rtl-optimization/81423
24089 * combine.c (make_compound_operation_int): Don't try to optimize
24090 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
24092 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
24094 PR rtl-optimization/81423
24095 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
24096 with a constant that is -1 in the truncated to mode.
24098 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24100 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
24101 (determine_unlikely_bbs): ... here.
24102 * predict.h (propagate_unlikely_bbs_forward): Declare.
24103 * cfgexpand.c (pass_expand::execute): Use it.
24104 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
24106 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
24107 propagate_unlikely_bbs_forward.
24109 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24111 PR middle-end/81331
24112 * except.c (maybe_add_nop_after_section_switch): New function.
24115 2017-07-19 Tom de Vries <tom@codesourcery.com>
24117 * gimple.h (gimple_phi_set_arg): Make assert more strict.
24119 2017-07-19 Tom de Vries <tom@codesourcery.com>
24121 * gimple.h (gimple_phi_arg): Make assert more strict.
24123 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
24125 * config.gcc (powerpc*-*-*): Add mmintrin.h.
24126 * config/rs6000/mmintrin.h: New file.
24127 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
24129 2017-07-19 Jakub Jelinek <jakub@redhat.com>
24131 PR tree-optimization/81346
24132 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
24134 2017-07-19 Tom de Vries <tom@codesourcery.com>
24136 * config/nvptx/nvptx.md (VECIM): Add V2DI.
24138 2017-07-19 Tom de Vries <tom@codesourcery.com>
24140 * config/nvptx/nvptx-modes.def: Add V2DImode.
24141 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
24142 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
24143 (nvptx_output_mov_insn): Handle lack of mov.b128.
24144 (nvptx_print_operand): Handle 'H' and 'L' codes.
24145 (nvptx_vector_mode_supported): Allow V2DImode.
24146 (nvptx_preferred_simd_mode): New function.
24147 (nvptx_data_alignment): New function.
24148 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
24149 nvptx_preferred_simd_mode.
24150 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
24152 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
24154 2017-07-19 Tom de Vries <tom@codesourcery.com>
24156 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
24157 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
24158 (nvptx_vector_mode_supported): New function. Allow V2SImode.
24159 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
24160 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
24161 (mov<VECIM>_insn): New define_insn.
24162 (define_expand "mov<VECIM>): New define_expand.
24164 2017-07-19 Tom de Vries <tom@codesourcery.com>
24166 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
24168 2017-07-19 Jakub Jelinek <jakub@redhat.com>
24170 PR tree-optimization/81346
24171 * fold-const.h (fold_div_compare, range_check_type): Declare.
24172 * fold-const.c (range_check_type): New function.
24173 (build_range_check): Use range_check_type.
24174 (fold_div_compare): No longer static, rewritten into
24175 a match.pd helper function.
24176 (fold_comparison): Don't call fold_div_compare here.
24177 * match.pd (X / C1 op C2): New optimization using fold_div_compare
24178 as helper function.
24180 2017-07-19 Nathan Sidwell <nathan@acm.org>
24182 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
24183 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
24184 * tree.c (find_decls_types_r, verify_type): Use
24185 TYPE_{MIN,MAX}_VALUE_RAW.
24186 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
24187 (hash_tree): Likewise.
24188 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
24190 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
24193 2017-07-18 Tom de Vries <tom@codesourcery.com>
24195 PR middle-end/81464
24196 * omp-expand.c (expand_omp_for_static_chunk): Handle
24197 equal-argument loop exit phi.
24199 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
24202 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
24203 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
24204 operand 2 predicate.
24205 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
24206 operand 2 predicate.
24207 (ror,rol -> rorx splitters): Use const_int_operand as
24208 operand 2 predicate.
24210 2017-06-18 Richard Biener <rguenther@suse.de>
24212 PR tree-optimization/81410
24213 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
24214 the gap in the ! slp_perm SLP case after each group.
24216 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24218 PR middle-end/81463
24219 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
24222 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24224 PR middle-end/81462
24225 * predict.c (set_even_probabilities): Cleanup; do not affect
24226 probabilities that are already known.
24227 (combine_predictions_for_bb): Call even when count is set.
24229 2017-07-18 Nathan Sidwell <nathan@acm.org>
24231 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
24234 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24237 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
24238 optimization for loop niter analysis.
24240 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
24243 * config/avr/avr.c (avr_optimize_casesi): Don't use
24244 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
24246 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
24248 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
24249 body_cost_vec from _vect_peel_extended_info.
24250 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
24251 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
24254 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24256 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
24258 2017-07-18 Richard Biener <rguenther@suse.de>
24260 PR tree-optimization/80620
24261 PR tree-optimization/81403
24262 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
24263 info when re-using a VN table entry.
24265 2017-07-18 Richard Biener <rguenther@suse.de>
24267 PR tree-optimization/81418
24268 * tree-vect-loop.c (vectorizable_reduction): Properly compute
24269 vectype_in. Verify that with lane-reducing reduction operations
24270 we have a single def-use cycle.
24272 2017-07-17 Carl Love <cel@us.ibm.com>
24274 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
24276 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24277 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24278 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24279 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24280 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24281 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24282 VMULOSW): New enum "unspec" values.
24283 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24284 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24285 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24286 altivec_vmulosw): New patterns.
24287 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24288 VMULOSW): Add definitions.
24290 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
24292 * config/alpha/alpha.c: Include predict.h.
24294 2017-07-17 Yury Gribov <tetra2005@gmail.com>
24296 * tree-vrp.c (compare_assert_loc): Fix comparison function
24297 to return predictable results.
24299 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24301 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
24303 (subdi3): Likewise.
24304 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
24305 * doc/invoke.texi (mexpand-adddi): Update text.
24307 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24309 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
24310 that also clobbers the CC register. The old expand code is moved
24312 (*arc_clzsi2): ... here.
24313 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
24314 the CC register. The old expand code is moved to ...
24315 (arc_ctzsi2): ... here.
24317 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24319 * config/arc/arc.opt (mindexed-loads): Use initial value
24320 TARGET_INDEXED_LOADS_DEFAULT.
24321 (mauto-modify-reg): Use initial value
24322 TARGET_AUTO_MODIFY_REG_DEFAULT.
24323 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24324 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24325 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24326 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24328 2017-07-17 Martin Liska <mliska@suse.cz>
24331 * opts.c (finish_options): Do not allow -fgnu-tm
24332 w/ -fsanitize={kernel-,}address. Say sorry.
24334 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24337 * tree-loop-distribution.c (classify_partition): Only assert on
24338 numer of iterations.
24339 (merge_dep_scc_partitions): Delete prameter. Update function call.
24340 (distribute_loop): Remove code handling loop with unknown niters.
24341 (pass_loop_distribution::execute): Skip loop with unknown niters.
24343 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24346 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
24347 function sort_partitions_by_post_order.
24349 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24351 PR tree-optimization/81374
24352 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
24353 the max index of basic blocks, rather than number of basic blocks.
24355 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24357 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
24359 (arc_legitimate_pic_operand_p): Likewise.
24360 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
24362 (arc_needs_pcl_p): Likewise.
24363 (arc_legitimate_pc_offset_p): Likewise.
24364 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
24365 function is also used in constrains.md.
24366 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
24367 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
24368 PLUS. Only return true/false in known cases, otherwise assert.
24369 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
24370 is already called in arc_legitimate_constant_p.
24371 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
24373 (LEGITIMATE_PIC_OPERAND_P): Use
24374 arc_raw_symbolic_reference_mentioned_p function.
24375 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
24380 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24381 Andrew Burgess <andrew.burgess@embecosm.com>
24383 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
24384 (arc_return_address_register): New function.
24385 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
24386 (arc_handle_fndecl_attribute): Add naked attribute.
24387 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
24388 (TARGET_WARN_FUNC_RETURN): Likewise.
24389 (arc_allocate_stack_slots_for_args): New function.
24390 (arc_warn_func_return): Likewise.
24391 (machine_function): Change type fn_type.
24392 (arc_compute_function_type): Consider new naked function type,
24393 change function return type.
24394 (arc_must_save_register): Adapt to handle new
24395 arc_compute_function_type's return type.
24396 (arc_expand_prologue): Likewise.
24397 (arc_expand_epilogue): Likewise.
24398 (arc_return_address_regs): Delete.
24399 (arc_return_address_register): New function.
24400 (arc_epilogue_uses): Use above function.
24401 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
24402 (arc_function_type): Change encoding, add naked type.
24403 (ARC_INTERRUPT_P): Change to handle the new encoding.
24404 (ARC_FAST_INTERRUPT_P): Likewise.
24405 (ARC_NORMAL_P): Define.
24406 (ARC_NAKED_P): Likewise.
24407 (arc_compute_function_type): Delete prototype.
24408 * config/arc/arc.md (in_ret_delay_slot): Use
24409 arc_return_address_register function.
24410 (simple_return): Likewise.
24411 (p_return_i): Likewise.
24413 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24415 PR tree-optimization/81428
24416 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
24417 can't be built for those types.
24419 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24421 Remove stuff dead since r239246.
24423 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
24424 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
24425 (avr_inform_devices): Remove dead stuff.
24427 2017-07-17 Tamar Christina <tamar.christina@arm.com>
24429 * config/arm/arm_neon.h: Fix softp typo.
24431 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24433 PR tree-optimization/81365
24434 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
24435 aggregate moves onto bb predecessor edges, make sure there are no
24436 loads that could alias the lhs in between the start of bb and the
24439 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24442 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
24443 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
24444 [LSHIFTRT, outer_code = TRUNCATE]: Same.
24446 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24448 PR tree-optimization/81396
24449 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
24450 (init_symbolic_number): Initialize it to 1.
24451 (perform_symbolic_merge): Add n_ops from both operands into the new
24453 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
24454 without base_addr as useless if they need more than one operation.
24455 (bswap_replace): Handle !bswap case for NULL base_addr.
24457 2017-07-17 Tom de Vries <tom@codesourcery.com>
24460 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
24463 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
24465 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
24466 conditional builtin define __FIX_LEON3FT_B2BST.
24468 2017-07-17 Daniel Cederman <cederman@gaisler.com>
24470 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
24471 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
24474 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24476 PR rtl-optimization/81424
24477 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
24478 to remove potential trapping from operands if -fnon-call-exceptions.
24480 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24482 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
24483 profile_proability for scalling.
24484 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
24486 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24488 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
24490 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24492 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
24493 fixpoint arithmetics.
24495 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24497 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
24498 fixpoint arithmetics.
24500 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24502 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
24503 fixpoint arithmetics.
24505 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24507 * profile-count.h (profile_probability::from_reg_br_prob_note,
24508 profile_probability::to_reg_br_prob_note): New functions.
24509 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
24510 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
24511 * predict.c (probability_reliable_p): Update.
24512 (edge_probability_reliable_p): Update.
24513 (br_prob_note_reliable_p): Update.
24514 (invert_br_probabilities): Update.
24515 (add_reg_br_prob_note): New function.
24516 (combine_predictions_for_insn): Update.
24517 * asan.c (asan_clear_shadow): Update.
24518 * cfgbuild.c (compute_outgoing_frequencies): Update.
24519 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24520 (update_br_prob_note): Update.
24521 (rtl_verify_edges): Update.
24522 (purge_dead_edges): Update.
24523 (fixup_reorder_chain): Update.
24524 * emit-rtl.c (try_split): Update.
24525 * ifcvt.c (cond_exec_process_insns): Update.
24526 (cond_exec_process_if_block): Update.
24527 (dead_or_predicable): Update.
24528 * internal-fn.c (expand_addsub_overflow): Update.
24529 (expand_neg_overflow): Update.
24530 (expand_mul_overflow): Update.
24531 * loop-doloop.c (doloop_modify): Update.
24532 * loop-unroll.c (compare_and_jump_seq): Update.
24533 * optabs.c (emit_cmp_and_jump_insn_1): Update.
24534 * predict.h: Update.
24535 * reorg.c (mostly_true_jump): Update.
24537 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
24538 * config/alpha/alpha.c (emit_unlikely_jump): Update.
24539 * config/arc/arc.c: (emit_unlikely_jump): Update.
24540 * config/arm/arm.c: (emit_unlikely_jump): Update.
24541 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
24542 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
24543 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
24544 (ix86_print_operand): Update.
24545 (ix86_split_fp_branch): Update.
24546 (predict_jump): Update.
24547 * config/ia64/ia64.c (ia64_print_operand): Update.
24548 * config/mmix/mmix.c (mmix_print_operand): Update.
24549 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
24550 (rs6000_expand_split_stack_prologue): Update.
24551 * config/rs6000/rs6000.c: Update.
24552 * config/s390/s390.c (s390_expand_vec_strlen): Update.
24553 (s390_expand_vec_movstr): Update.
24554 (s390_expand_cs_tdsi): Update.
24555 (s390_expand_split_stack_prologue): Update.
24556 * config/sh/sh.c (sh_print_operand): Update.
24557 (expand_cbranchsi4): Update.
24558 (expand_cbranchdi4): Update.
24559 * config/sparc/sparc.c (output_v9branch): Update.
24560 * config/spu/spu.c (get_branch_target): Update.
24561 (ea_load_store_inline): Update.
24562 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
24563 * config/tilepro/tilepro.c: Update.
24565 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24567 * gimplify.c (mostly_copy_tree_r): Revert latest change.
24568 (gimplify_save_expr): Likewise.
24570 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24572 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
24574 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24576 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
24578 * timevar.def (TV_IPA_FNSUMMARY): Define.
24580 2017-07-16 Daniel Cederman <cederman@gaisler.com>
24582 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
24583 to back store errata sensitive sequence from being generated.
24584 (sqrtdf2_fix): Likewise.
24586 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24588 * tree-ssa-threadupdate.c (compute_path_counts,
24589 update_joiner_offpath_counts): Use profile_probability.
24591 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
24594 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24596 * config/arm/arm-c.c (arm_cpu_builtins): Define
24597 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24599 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
24601 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24602 array entries to represent __ieee128 versions of the
24603 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
24604 scalar_extract_sig, and scalar_insert_exp built-in functions.
24605 (altivec_resolve_overloaded_builtin): Add special case handling
24606 for the __builtin_scalar_insert_exp function, as represented by
24607 the P9V_BUILTIN_VEC_VSIEDP constant.
24608 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
24609 exponent support for __ieee128 argument.
24610 (VSESQP): Add scalar extract signature support for __ieee128
24612 (VSTDCNQP): Add scalar test negative support for __ieee128
24614 (VSIEQP): Add scalar insert exponent support for __int128 argument
24615 with __ieee128 result.
24616 (VSIEQPF): Add scalar insert exponent support for __ieee128
24617 argument with __ieee128 result.
24618 (VSTDCQP): Add scalar test data class support for __ieee128
24620 (VSTDCNQP): Add overload support for scalar test negative with
24621 __ieee128 argument.
24622 (VSTDCQP): Add overload support for scalar test data class
24623 __ieee128 argument.
24624 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
24625 UNSPEC_VSX_SXSIGDP.
24626 (UNSPEC_VSX_SIEXPQP): New constant.
24627 (xsxexpqp): New insn for VSX scalar extract exponent quad
24629 (xsxsigqp): New insn for VSX scalar extract significand quad
24631 (xsiexpqpf): New insn for VSX scalar insert exponent quad
24632 precision with floating point argument.
24633 (xststdcqp): New expand for VSX scalar test data class quad
24635 (xststdcnegqp): New expand for VSX scalar test negative quad
24637 (xststdcqp): New insn to match expansions for VSX scalar test data
24638 class quad precision and VSX scalar test negative quad precision.
24639 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
24640 special case operand checking to enforce that second operand of
24641 VSX scalar test data class with quad precision argument is a 7-bit
24643 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24644 prototypes and descriptions of __ieee128 versions of
24645 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
24646 scalar_test_data_class, and scalar_test_neg built-in functions.
24648 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24650 PR tree-optimization/81162
24651 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
24652 replace a negate with an add.
24654 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
24656 * doc/invoke.texi (arm/-mcpu): Document +crypto.
24658 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24660 * config/arm/arm-c.c (arm_cpu_builtins): Define
24661 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24663 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24665 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
24666 (armv8-r): Set ARM Cortex-R52 as default CPU.
24667 * config/arm/arm-tables.opt: Regenerate.
24668 * config/arm/arm-tune.md: Regenerate.
24669 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24671 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
24672 extension for -mcpu=cortex-r52.
24674 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24676 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
24677 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
24678 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
24679 (fp-armv8): Define it as FP_ARMv8 only.
24680 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
24681 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
24683 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
24684 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
24685 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
24686 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
24687 than TARGET_FPU_ARMV8.
24688 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
24689 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
24690 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
24692 * config/arm/neon.md (neon_vrint): Likewise.
24693 (neon_vcvt): Likewise.
24694 (neon_<fmaxmin_op><mode>): Likewise.
24695 (<fmaxmin><mode>3): Likewise.
24696 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
24697 * config/arm/predicates.md (arm_cond_move_operator): Check against
24698 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
24700 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
24702 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
24703 to top of function.
24705 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24707 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
24708 loop in comment with memset.
24710 2017-07-14 Martin Liska <mliska@suse.cz>
24712 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
24713 * dwarf2out.c (is_java): Remove the function.
24714 (output_pubname): Remove usage of the function.
24715 (lower_bound_default): Remove usage of DW_LANG_Java.
24716 (gen_compile_unit_die): Likewise.
24717 * gcc.c: Remove compiler defaults for .java and .zip files.
24718 * gimple-expr.c (remove_suffix): Change as there's no longer
24719 extension than 4-letter one.
24720 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
24721 (gimplify_save_expr): Likewise.
24722 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
24723 as it's possible even for other languages than Java.
24724 * langhooks.h (struct lang_hooks): Remove Java from a comment.
24725 * lto-opts.c (lto_write_options): Remove reference to Java.
24726 * opts.c (strip_off_ending): Update file extension handling.
24727 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
24728 * tree-eh.c (lower_resx): Likewise.
24729 * tree.c (free_lang_data_in_type): Remove dead code.
24730 (find_decls_types_r): Likewise.
24731 (build_common_builtin_nodes): Remove Java from a comment.
24732 (verify_type): Remove dead code.
24733 * varasm.c (assemble_external): Remove Java from a comment.
24735 2017-07-14 Martin Liska <mliska@suse.cz>
24737 * opts.c (finish_options): Add quotes.
24738 (common_handle_option): Likewise.
24740 2017-07-14 Martin Liska <mliska@suse.cz>
24742 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
24743 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
24744 Remove N_SO_PASCAL.
24745 * dwarf2out.c (lower_bound_default): Do not handle
24747 (gen_compile_unit_die): Likewise.
24748 * gcc.c: Remove default extension binding for GNU Pascal.
24749 * stmt.c: Remove Pascal language from a comment.
24750 * xcoffout.c: Likewise.
24752 2017-07-13 David Malcolm <dmalcolm@redhat.com>
24755 * diagnostic-show-locus.c (fixit_cmp): New function.
24756 (layout::layout): Sort m_fixit_hints.
24757 (column_range::column_range): Assert that the values are valid.
24758 (struct char_span): New struct.
24759 (correction::overwrite): New method.
24760 (struct source_line): New struct.
24761 (line_corrections::add_hint): Add assertions. Reimplement memcpy
24762 calls in terms of classes source_line and char_span, and
24763 correction::overwrite.
24764 (selftest::test_overlapped_fixit_printing_2): New function.
24765 (selftest::diagnostic_show_locus_c_tests): Call it.
24767 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
24769 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
24770 early if there is no lhs.
24772 2017-07-13 Martin Liska <mliska@suse.cz>
24774 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
24775 (gen_reference_type_die): Likewise.
24776 * stor-layout.c: Remove Pascal-related comment.
24778 2017-07-13 Martin Liska <mliska@suse.cz>
24780 * opts.c (finish_options): Add quotes to error messages.
24781 (parse_sanitizer_options): Likewise.
24783 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
24785 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
24787 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
24789 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24791 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24793 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24795 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24797 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24800 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24801 provides the hardware capability bits, define the macro
24802 __BUILTIN_CPU_SUPPORTS__.
24803 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24804 if GLIBC does not provide the hardware capability bits. Add a
24805 gcc_unreachable call if the built-in cpu function is neither
24806 __builtin_cpu_is nor __builtin_cpu_supports.
24807 (rs6000_get_function_versions_dispatcher): Change the warning
24808 that an old GLIBC is used which does not export the capability
24809 bits to be an error.
24810 * doc/extend.texi (target_clones attribute): Document the
24811 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24812 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24813 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24814 the macros defined by GCC if the newer GLIBC is available.
24816 2017-07-12 Jeff Law <law@redhat.com>
24818 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24819 remaining includes slightly.
24820 * config/riscv/riscv-builtins.c: Include profile-count.h.
24822 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24825 * config/avr/avr.c (avr_set_current_function): In diagnostic
24826 messages: Quote keywords and (parts of) identifiers.
24827 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24830 2017-07-12 Carl Love <cel@us.ibm.com>
24832 * config/rs6000/rs6000-c.c: Add support for built-in functions
24833 vector bool char vec_revb (vector bool char);
24834 vector bool short vec_revb (vector short char);
24835 vector bool int vec_revb (vector bool int);
24836 vector bool long long vec_revb (vector bool long long);
24837 * doc/extend.texi: Update the built-in documentation file for the
24838 new built-in functions.
24840 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24842 * config/s390/s390.md: Remove movcc splitter.
24844 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24846 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24847 load/store on condition.
24849 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24852 * config/avr/avr.c (avr_encode_section_info)
24853 [progmem && !TREE_READONLY]: Error if progmem object needs
24856 2017-07-11 Michael Collison <michael.collison@arm.com>
24858 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24861 2017-07-11 Carl Love <cel@us.ibm.com>
24863 * config/rs6000/rs6000-c.c: Add support for builtins
24864 vector unsigned int vec_parity_lsbb (vector signed int);
24865 vector unsigned int vec_parity_lsbb (vector unsigned int);
24866 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24867 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24868 vector unsigned long long vec_parity_lsbb (vector signed long long);
24869 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24870 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24871 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24872 * doc/extend.texi: Update the built-in documentation file for the
24873 new built-in functions.
24875 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24877 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24878 (layout::m_primary_loc): New field.
24879 (layout::layout): Initialize new field. Move location filtering
24880 logic from here to...
24881 (layout::maybe_add_location_range): ...this new method. Add
24882 support for filtering to just the lines already specified by other
24884 (layout::will_show_line_p): New method.
24885 (gcc_rich_location::add_location_if_nearby): New method.
24886 (selftest::test_add_location_if_nearby): New test function.
24887 (selftest::diagnostic_show_locus_c_tests): Call it.
24888 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24891 2017-07-11 Tom de Vries <tom@codesourcery.com>
24893 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24894 (bb_first_real_insn): New function.
24895 (nvptx_single): Add extra initialization of broadcasted condition
24898 2017-07-11 Nathan Sidwell <nathan@acm.org>
24900 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24902 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24904 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24905 Binutils doc as TEXI will mess them up.
24906 * doc/invoke.texi (AVR Options): Same here.
24908 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24910 * config/sparc/sparc.opt (mfix-ut700): New option.
24911 (mfix-gr712rc): Likewise.
24912 (sparc_fix_b2bst): New variable.
24913 * doc/invoke.texi (SPARC options): Document them.
24914 (ARM options): Fix warnings.
24915 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24916 instructions to prevent sequences that can trigger the store-store
24917 errata for certain LEON3FT processors.
24918 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24919 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24920 * config/sparc/sparc.md (fix_b2bst): New attribute.
24921 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24923 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24926 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24928 (*rsqrtsf2_sse): Ditto.
24930 (div<mode>3): Macroize insn from divdf3 and divsf3
24931 using MODEF mode iterator.
24933 2017-07-10 Martin Sebor <msebor@redhat.com>
24935 PR tree-optimization/80397
24936 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24937 instead of testing for equality to INTEGER_TYPE.
24939 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24941 * config.gcc: Remove uclibc from arc target spec.
24943 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24945 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24947 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24950 * lto-wrapper.c (remove_option): New function.
24951 (merge_and_complain): Merge PIC/PIE options more realistically.
24953 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24955 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24959 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24960 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24961 * config.in: Regenerate.
24962 * configure: Regenerate.
24963 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24964 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24965 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24966 (TARGET_GASISR_PROLOGUES): ...target mask.
24967 * common/config/avr/avr-common.c
24968 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24969 Set -mgas-isr-prologues.
24970 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24971 INSERT_PASS_BEFORE for it.
24972 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24973 * config/avr/avr.c (avr_option_override)
24974 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24975 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24976 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24977 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24978 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24980 (make_avr_pass_pre_proep): New function.
24981 (emit_push_sfr) <treg>: Add argument to function and use it
24982 instead of TMP_REG.
24983 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24984 and set machine->gasisr.yes.
24985 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24986 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24987 __gcc_isr.n_pushed to .L__stack_usage.
24988 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24989 (avr_asm_final_postscan_insn): ...this new static function.
24990 * config/avr/avr.h (machine_function)
24991 <is_no_gccisr, use_L__stack_usage>: New fields.
24992 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24993 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24994 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24995 (gasisr, *gasisr): New expander and insn.
24996 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24997 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24998 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
25000 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
25002 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
25005 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
25007 Move jump-tables out of .text again.
25010 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
25011 (ASM_OUTPUT_ADDR_VEC): New function.
25012 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
25013 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
25014 INSN_ADDRESSes as asm comment.
25015 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
25016 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
25017 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
25018 * config/avr/avr.md (*tablejump): Adjust comment.
25019 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
25020 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
25022 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
25023 (avr_output_addr_vec): New proto.
25024 (avr_log_t) <insn_addresses>: New field.
25026 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
25029 * config/i386/i386.c (ix86_function_arg_advance): Set
25030 outgoing_args_on_stack to true if there are outgoing arguments
25032 (ix86_function_arg): Likewise.
25033 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
25034 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
25035 * config/i386/i386.h (machine_function): Add
25036 outgoing_args_on_stack.
25038 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
25040 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
25041 supporting pthreds.
25042 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
25044 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
25046 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
25047 (REAL_H): Remove $(MACHMODE_H).
25048 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
25050 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
25051 $(MACHMODE_H) and double-int.h.
25052 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
25054 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
25057 2017-07-07 Andrew Pinski <apinski@cavium.com>
25059 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
25060 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
25062 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
25064 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
25065 Add warning if GCC was not configured to link against a GLIBC that
25066 exports the hardware capability bits.
25067 (make_resolver_func): Make resolver function private and not a
25068 COMDAT function. Create the name with clone_function_name instead
25069 of make_unique_name.
25072 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
25073 correct operand in doing the split.
25075 2017-07-07 Carl Love <cel@us.ibm.com>
25077 * config/rs6000/rs6000-c: Add support for built-in function
25078 vector unsigned short vec_pack_to_short_fp32 (vector float,
25080 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
25081 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
25082 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
25083 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
25084 (convert_4f32_8i16): Add define_expand.
25085 * doc/extend.texi: Update the built-in documentation file for the
25086 new built-in function.
25088 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25090 * config/sparc/m8.md: New file.
25091 * config/sparc/sparc.md: Include m8.md.
25093 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25095 * config/sparc/sparc.opt: New option -mvis4b.
25096 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
25097 (sparc_option_override): Handle VIS4B.
25098 (enum sparc_builtins): Define
25099 SPARC_BUILTIN_DICTUNPACK{8,16,32},
25100 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
25101 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
25102 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
25103 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
25104 (check_constant_argument): New function.
25105 (sparc_vis_init_builtins): Define builtins
25106 __builtin_vis_dictunpack{8,16,32},
25107 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
25108 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
25109 __builtin_vis_fpcmpde{8,16,32}shl and
25110 __builtin_vis_fpcmpur{8,16,32}shl.
25111 (sparc_expand_builtin): Check that the constant operands to
25112 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
25113 constant and in range.
25114 * config/sparc/sparc-c.c (sparc_target_macros): Handle
25116 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
25117 (SPARC_IMM5_P): Likewise.
25118 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
25119 (enabled): Handle vis4b.
25120 (UNSPEC_DICTUNPACK): New unspec.
25121 (UNSPEC_FPCMPSHL): Likewise.
25122 (UNSPEC_FPUCMPSHL): Likewise.
25123 (UNSPEC_FPCMPDESHL): Likewise.
25124 (UNSPEC_FPCMPURSHL): Likewise.
25125 (cpu_feature): New CPU feature `vis4b'.
25126 (dictunpack{8,16,32}): New insns.
25127 (FPCSMODE): New mode iterator.
25128 (fpcscond): New code iterator.
25129 (fpcsucond): Likewise.
25130 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
25131 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
25132 (fpcmpde{8,16,32}{si,di}shl): Likewise.
25133 (fpcmpur{8,16,32}{si,di}shl): Likewise.
25134 * config/sparc/constraints.md: Define constraints `q' for unsigned
25135 2-bit integer constants and `t' for unsigned 5-bit integer
25137 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
25139 (imm5_operand_dictunpack16): Likewise.
25140 (imm5_operand_dictunpack32): Likewise.
25141 (imm2_operand): Likewise.
25142 * doc/invoke.texi (SPARC Options): Document -mvis4b.
25143 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
25144 ditunpack* and fpcmp*shl builtins.
25146 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25148 * config.gcc: Handle m8 in --with-{cpu,tune} options.
25149 * config.in: Add HAVE_AS_SPARC6 define.
25150 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
25152 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
25154 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
25155 (CPP_CPU_SPEC): Handle m8.
25156 (ASM_CPU_SPEC): Likewise.
25157 * config/sparc/sparc-opts.h (enum processor_type): Add
25159 * config/sparc/sparc.c (m8_costs): New struct.
25160 (sparc_option_override): Handle TARGET_CPU_m8.
25161 (sparc32_initialize_trampoline): Likewise.
25162 (sparc64_initialize_trampoline): Likewise.
25163 (sparc_issue_rate): Likewise.
25164 (sparc_register_move_cost): Likewise.
25165 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
25166 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
25167 (ASM_CPU64_DEFAULT_SPEC): Likewise.
25168 (CPP_CPU_SPEC): Handle M8.
25169 (ASM_CPU_SPEC): Likewise.
25170 (AS_M8_FLAG): Define.
25171 * config/sparc/sparc.md: Add m8 to the cpu attribute.
25172 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
25173 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
25175 * configure: Regenerate.
25176 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
25179 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25181 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
25183 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
25184 ("*movdi_insn_sp32"): Do not set v3pipe.
25185 ("*movsi_insn"): Likewise.
25186 ("*movdi_insn_sp64"): Likewise.
25187 ("*movsf_insn"): Likewise.
25188 ("*movdf_insn_sp32"): Likewise.
25189 ("*movdf_insn_sp64"): Likewise.
25190 ("*zero_extendsidi2_insn_sp64"): Likewise.
25191 ("*sign_extendsidi2_insn"): Likewise.
25192 ("*mov<VM32:mode>_insn"): Likewise.
25193 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25194 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25195 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25196 ("<vlop:code><VL:mode>3"): Likewise.
25197 ("*not_<vlop:code><VL:mode>3"): Likewise.
25198 ("*nand<VL:mode>_vis"): Likewise.
25199 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
25200 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
25201 ("one_cmpl<VL:mode>2"): Likewise.
25202 ("faligndata<VM64:mode>_vis"): Likewise.
25203 ("alignaddrsi_vis"): Likewise.
25204 ("alignaddrdi_vis"): Likweise.
25205 ("alignaddrlsi_vis"): Likewise.
25206 ("alignaddrldi_vis"): Likewise.
25207 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25208 ("bmaskdi_vis"): Likewise.
25209 ("bmasksi_vis"): Likewise.
25210 ("bshuffle<VM64:mode>_vis"): Likewise.
25211 ("cmask8<P:mode>_vis"): Likewise.
25212 ("cmask16<P:mode>_vis"): Likewise.
25213 ("cmask32<P:mode>_vis"): Likewise.
25214 ("pdistn<P:mode>_vis"): Likewise.
25215 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25217 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25219 * config/sparc/sparc.md ("subtype"): New insn attribute.
25220 ("*wrgsr_sp64"): Set insn subtype.
25221 ("*rdgsr_sp64"): Likewise.
25222 ("alignaddrsi_vis"): Likewise.
25223 ("alignaddrdi_vis"): Likewise.
25224 ("alignaddrlsi_vis"): Likewise.
25225 ("alignaddrldi_vis"): Likewise.
25226 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25227 ("fexpand_vis"): Likewise.
25228 ("fpmerge_vis"): Likewise.
25229 ("faligndata<VM64:mode>_vis"): Likewise.
25230 ("bshuffle<VM64:mode>_vis"): Likewise.
25231 ("cmask8<P:mode>_vis"): Likewise.
25232 ("cmask16<P:mode>_vis"): Likewise.
25233 ("cmask32<P:mode>_vis"): Likewise.
25234 ("fchksm16_vis"): Likewise.
25235 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
25236 ("fmean16_vis"): Likewise.
25237 ("fp<plusminus_insn>64_vis"): Likewise.
25238 ("<plusminus_insn>v8qi3"): Likewise.
25239 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25240 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
25241 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
25242 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
25243 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
25244 ("*movqi_insn"): Likewise.
25245 ("*movhi_insn"): Likewise.
25246 ("*movsi_insn"): Likewise.
25247 ("movsi_pic_gotdata_op"): Likewise.
25248 ("*movdi_insn_sp32"): Likewise.
25249 ("*movdi_insn_sp64"): Likewise.
25250 ("movdi_pic_gotdata_op"): Likewise.
25251 ("*movsf_insn"): Likewise.
25252 ("*movdf_insn_sp32"): Likewise.
25253 ("*movdf_insn_sp64"): Likewise.
25254 ("*zero_extendhisi2_insn"): Likewise.
25255 ("*zero_extendqihi2_insn"): Likewise.
25256 ("*zero_extendqisi2_insn"): Likewise.
25257 ("*zero_extendqidi2_insn"): Likewise.
25258 ("*zero_extendhidi2_insn"): Likewise.
25259 ("*zero_extendsidi2_insn_sp64"): Likewise.
25260 ("ldfsr"): Likewise.
25261 ("prefetch_64"): Likewise.
25262 ("prefetch_32"): Likewise.
25263 ("tie_ld32"): Likewise.
25264 ("tie_ld64"): Likewise.
25265 ("*tldo_ldub_sp32"): Likewise.
25266 ("*tldo_ldub1_sp32"): Likewise.
25267 ("*tldo_ldub2_sp32"): Likewise.
25268 ("*tldo_ldub_sp64"): Likewise.
25269 ("*tldo_ldub1_sp64"): Likewise.
25270 ("*tldo_ldub2_sp64"): Likewise.
25271 ("*tldo_ldub3_sp64"): Likewise.
25272 ("*tldo_lduh_sp32"): Likewise.
25273 ("*tldo_lduh1_sp32"): Likewise.
25274 ("*tldo_lduh_sp64"): Likewise.
25275 ("*tldo_lduh1_sp64"): Likewise.
25276 ("*tldo_lduh2_sp64"): Likewise.
25277 ("*tldo_lduw_sp32"): Likewise.
25278 ("*tldo_lduw_sp64"): Likewise.
25279 ("*tldo_lduw1_sp64"): Likewise.
25280 ("*tldo_ldx_sp64"): Likewise.
25281 ("*mov<VM32:mode>_insn"): Likewise.
25282 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25283 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25285 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25287 * config/sparc/sparc.md ("type"): New insn type viscmp.
25288 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
25290 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
25291 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
25292 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25293 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
25295 ("n7_vis_logical_11cycle"): Likewise.
25296 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
25297 * config/sparc/niagara2.md ("niag3_vis": Likewise.
25298 * config/sparc/niagara.md ("niag_vis"): Likewise.
25299 * config/sparc/ultra3.md ("us3_fga"): Likewise.
25300 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
25302 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25304 * config/sparc/sparc.md: New instruction type `bmask'.
25305 (bmaskdi_vis): Use the `bmask' type.
25306 (bmasksi_vis): Likewise.
25307 * config/sparc/ultra3.md (us3_array): Likewise.
25308 * config/sparc/niagara7.md (n7_array): Likewise.
25309 * config/sparc/niagara4.md (n4_array): Likewise.
25310 * config/sparc/niagara2.md (niag2_vis): Likewise.
25311 (niag3_vis): Likewise.
25312 * config/sparc/niagara.md (niag_vis): Likewise.
25314 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25316 * ipa-comdats.c: Remove optimize check from gate.
25317 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
25318 for functions not optimized.
25319 (ipa_fn_summary_read): Skip optimize check.
25320 (ipa_fn_summary_write): Likewise.
25321 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
25323 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
25325 (can_inline_edge_p): Check flag_pcc_struct_return for match.
25326 (check_callers): Give up on caller which is not optimized.
25327 (inline_small_functions): Likewise.
25328 (ipa_inline): Do not give up when not optimizing.
25329 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
25330 away unoptimizes cdtors.
25331 (whole_program_function_and_variable_visibility): Do
25332 ipa_discover_readonly_nonaddressable_vars in LTO mode.
25333 * ipa.c (process_references): Do not check optimize.
25334 (symbol_table::remove_unreachable_nodes): Update optimize check.
25335 (set_writeonly_bit): Update optimize check.
25336 (pass_ipa_cdtor_merge::gate): Do not check optimize.
25337 (pass_ipa_single_use::gate): Remove.
25339 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25341 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
25342 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
25343 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
25344 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
25345 permute_load, permute_store, adjust_extract, adjust_splat,
25346 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
25347 replace_swap_with_copy, dump_swap_insn_table,
25348 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
25349 recombine_lvx_pattern, recombine_stvx_pattern,
25350 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
25351 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
25352 to file rs6000-p8swap.c.
25353 * config/rs6000/rs6000-p8swap.c: New file.
25354 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
25355 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
25356 and rs6000*-*-* targets.
25358 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25360 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
25362 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25364 * lto-wrapper.c (merge_and_complain): Do not merge
25365 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
25366 fsigned_zeros, ftrapping_math, fwrapv.
25367 (append_compiler_options): Do not track these options.
25368 (append_linker_options): Likewie
25370 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25372 * cgraphunit.c (cgraph_node::finalize_function): When
25373 !flag_toplevel_reorde set no_reorder flag.
25374 (varpool_node::finalize_decl): Likewise.
25375 (symbol_table::compile): Drop no toplevel reorder path.
25377 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25379 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
25380 edges; zero probability is not better than uninitialized.
25382 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25384 * asan.h (asan_sanitize_allocas_p): Declare.
25385 * asan.c (asan_sanitize_allocas_p): New function.
25386 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
25387 (handle_builtin_alloca): Likewise.
25388 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
25389 if !asan_sanitize_allocas_p.
25390 * params.def (asan-instrument-allocas): Add new option.
25391 * params.h (ASAN_PROTECT_ALLOCAS): Define.
25392 * opts.c (common_handle_option): Disable allocas sanitization for
25395 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25397 * asan.c: Include gimple-fold.h.
25398 (get_last_alloca_addr): New function.
25399 (handle_builtin_stackrestore): Likewise.
25400 (handle_builtin_alloca): Likewise.
25401 (asan_emit_allocas_unpoison): Likewise.
25402 (get_mem_refs_of_builtin_call): Add new parameter, remove const
25403 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
25404 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
25405 (instrument_builtin_call): Pass gimple iterator to
25406 get_mem_refs_of_builtin_call.
25407 (last_alloca_addr): New global.
25408 * asan.h (asan_emit_allocas_unpoison): Declare.
25409 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
25410 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
25411 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
25412 if function calls alloca.
25413 * gimple-fold.c (replace_call_with_value): Remove static keyword.
25414 * gimple-fold.h (replace_call_with_value): Declare.
25415 * internal-fn.c: Include asan.h.
25416 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
25417 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
25419 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25421 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
25422 (C_SELFTEST_FLAGS): New.
25423 (CPP_SELFTEST_FLAGS): New.
25424 (SELFTEST_DEPS): New, from deps of s-selftest.
25425 (C_SELFTEST_DEPS): New, from deps of s-selftest.
25426 (CPP_SELFTEST_DEPS): New.
25427 (selftest): Add dependency on s-selftest-c++.
25428 (s-selftest): Rename to...
25429 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
25430 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
25431 than SELFTEST_FLAGS.
25432 (selftest-gdb): Rename to...
25433 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
25435 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
25436 (selftest-valgrind): Rename to...
25437 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
25439 (selftest-valgrind): Reintroduce as an alias for
25440 selftest-c-valgrind.
25441 (s-selftest-c++): New.
25442 (selftest-c++-gdb): New.
25443 (selftest-c++-valgrind): New.
25445 2017-07-06 Olivier Hainque <hainque@adacore.com>
25447 * gcc.c (process_command): When deciding if undefined variables
25448 should be ignored when processing specs, accept "gcc -v" as well.
25450 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25452 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
25453 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
25455 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25457 * config/arm/arm-cpus.in (armv8-r): Add new entry.
25458 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
25459 * config/arm/arm-tables.opt: Regenerate.
25460 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
25462 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
25464 2017-07-06 Carl Love <cel@us.ibm.com>
25466 * ChangeLog: Clean up from mid air collision
25468 2017-07-06 Carl Love <cel@us.ibm.com>
25470 * config/rs6000/rs6000-c.c: Add support for built-in functions
25471 vector signed int vec_subc (vector signed int, vector signed int);
25472 vector signed __int128 vec_subc (vector signed __int128,
25473 vector signed __int128);
25474 vector unsigned __int128 vec_subc (vector unsigned __int128,
25475 vector unsigned __int128);
25476 vector signed int vec_sube (vector signed int, vector signed int,
25477 vector signed int);
25478 vector unsigned int vec_sube (vector unsigned int,
25479 vector unsigned int,
25480 vector unsigned int);
25481 vector signed __int128 vec_sube (vector signed __int128,
25482 vector signed __int128,
25483 vector signed__int128);
25484 vector unsigned __int128 vec_sube (vector unsigned __int128,
25485 vector unsigned __int128,
25486 vector unsigned __int128);
25487 vector signed int vec_subec (vector signed int, vector signed int,
25488 vector signed int);
25489 vector unsigned int vec_subec (vector unsigned int,
25490 vector unsigned int,
25491 vector unsigned int);
25492 vector signed __int128 vec_subec (vector signed __int128,
25493 vector signed __int128,
25494 vector signed__int128);
25495 vector unsigned __int128 vec_subec (vector unsigned __int128,
25496 vector unsigned __int128,
25497 vector unsigned __int128);
25498 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
25499 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
25500 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
25501 BU_ALTIVEC_OVERLOAD_X definitions.
25502 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
25503 * doc/extend.texi: Update the built-in documentation file for the new
25504 built-in functions.
25506 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25509 * diagnostic-show-locus.c (layout::layout): Use start and finish
25510 spelling location for the start and finish of each range.
25511 * genmatch.c (linemap_client_expand_location_to_spelling_point):
25512 Add unused aspect param.
25513 * input.c (expand_location_1): Add "aspect" param, and use it
25514 to access the correct part of the location.
25515 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
25517 (expand_location_to_spelling_point): Likewise.
25518 (linemap_client_expand_location_to_spelling_point): Add "aspect"
25519 param, and pass it to expand_location_1.
25521 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
25523 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
25524 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
25525 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
25526 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
25527 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
25528 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
25529 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
25530 _mm_maskz_getmant_ss): New intrinsics.
25531 (__builtin_ia32_getexpss128_mask): Changed to ...
25532 __builtin_ia32_getexpss128_round ... this.
25533 (__builtin_ia32_getexpsd128_mask): Changed to ...
25534 __builtin_ia32_getexpsd128_round ... this.
25535 * config/i386/i386-builtin-types.def
25536 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
25537 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
25538 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
25539 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
25540 __builtin_ia32_getmantss_mask_round): New builtins.
25541 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
25542 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
25543 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
25544 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
25545 * config/i386/sse.md
25546 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
25547 avx512f_sgetexp<mode><mask_scalar_name>
25548 <round_saeonly_scalar_name> ... this.
25549 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
25550 %0, %1, %2<round_saeonly_op3>}): Changed to ...
25551 vgetexp<ssescalarmodesuffix>
25552 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25553 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
25554 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
25555 avx512f_vgetmant<mode><mask_scalar_name>
25556 <round_saeonly_scalar_name> ... this.
25557 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
25558 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
25559 vgetmant<ssescalarmodesuffix>
25560 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
25561 %0<mask_scalar_operand4>, %1, %2
25562 <round_saeonly_scalar_mask_op4>, %3} ... this.
25563 * config/i386/subst.md (mask_scalar_operand4,
25564 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
25565 round_saeonly_scalar_nimm_predicate): New subst attributes.
25567 2017-07-06 Julia Koval <julia.koval@intel.com>
25569 * config/i386/i386.c (ix86_erase_embedded_rounding):
25570 Remove code for old rounding pattern.
25572 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
25574 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
25576 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
25578 * doc/sourcebuild.texi (Test Directives, Variants of
25579 dg-require-support): Add documentation for dg-require-stack-check.
25581 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
25583 * config/i386/subst.md (mask_scalar, round_scalar,
25584 round_saeonly_scalar): New meta-templates.
25585 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
25586 round_scalar_mask_operand3, round_scalar_mask_op3,
25587 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
25588 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
25589 round_saeonly_scalar_constraint,
25590 round_saeonly_scalar_prefix): New subst attribute.
25591 * config/i386/sse.md
25592 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
25593 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
25594 <round_scalar_name> ... this.
25595 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
25596 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
25597 <round_scalar_name> ... this.
25598 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
25599 <sse>_vm<code><mode>3<mask_scalar_name>
25600 <round_saeonly_scalar_name> ... this.
25601 (v<plusminus_mnemonic><ssescalarmodesuffix>
25602 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25603 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25604 v<plusminus_mnemonic><ssescalarmodesuffix>
25605 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25606 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25607 (v<multdiv_mnemonic><ssescalarmodesuffix>
25608 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25609 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25610 v<multdiv_mnemonic><ssescalarmodesuffix>
25611 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25612 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25613 (v<maxmin_float><ssescalarmodesuffix>
25614 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
25615 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
25616 v<maxmin_float><ssescalarmodesuffix>
25617 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25618 %0<mask_scalar_operand3>, %1, %<iptr>2
25619 <round_saeonly_scalar_mask_op3>} ... this.
25621 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
25623 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
25624 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
25626 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
25627 Alan Hayward <alan.hayward@arm.com>
25628 David Sherwood <david.sherwood@arm.com>
25630 * combine.c (simplify_if_then_else): Remove "enum" before
25632 * compare-elim.c (can_eliminate_compare): Likewise.
25633 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
25635 (aarch64_lookup_simd_builtin_type): Likewise.
25636 (aarch64_simd_builtin_type): Likewise.
25637 (aarch64_init_simd_builtin_types): Likewise.
25638 (aarch64_simd_expand_args): Likewise.
25639 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
25641 (aarch64_reverse_mask): Likewise.
25642 (aarch64_simd_emit_reg_reg_move): Likewise.
25643 (aarch64_gen_adjusted_ldpstp): Likewise.
25644 (aarch64_ccmp_mode_to_code): Likewise.
25645 (aarch64_operands_ok_for_ldpstp): Likewise.
25646 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25647 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
25649 (aarch64_min_divisions_for_recip_mul): Likewise.
25650 (aarch64_reassociation_width): Likewise.
25651 (aarch64_get_condition_code_1): Likewise.
25652 (aarch64_simd_emit_reg_reg_move): Likewise.
25653 (aarch64_simd_attr_length_rglist): Likewise.
25654 (aarch64_reverse_mask): Likewise.
25655 (aarch64_operands_ok_for_ldpstp): Likewise.
25656 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25657 (aarch64_gen_adjusted_ldpstp): Likewise.
25658 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
25660 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
25661 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
25662 (arm_lookup_simd_builtin_type): Likewise.
25663 (arm_simd_builtin_type): Likewise.
25664 (arm_init_simd_builtin_types): Likewise.
25665 (arm_expand_builtin_args): Likewise.
25666 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
25667 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
25668 (ft32_setup_incoming_varargs): Likewise.
25669 (ft32_function_arg): Likewise.
25670 (ft32_function_arg_advance): Likewise.
25671 (ft32_pass_by_reference): Likewise.
25672 (ft32_arg_partial_bytes): Likewise.
25673 (ft32_valid_pointer_mode): Likewise.
25674 (ft32_addr_space_pointer_mode): Likewise.
25675 (ft32_addr_space_legitimate_address_p): Likewise.
25676 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
25678 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
25679 (ix86_emit_outlined_ms2sysv_restore): Likewise.
25680 (iamcu_alignment): Likewise.
25681 (canonicalize_vector_int_perm): Likewise.
25682 (ix86_noce_conversion_profitable_p): Likewise.
25683 (ix86_mpx_bound_mode): Likewise.
25684 (ix86_operands_ok_for_move_multiple): Likewise.
25685 * config/microblaze/microblaze-protos.h
25686 (microblaze_expand_conditional_branch_reg): Likewise.
25687 * config/microblaze/microblaze.c
25688 (microblaze_expand_conditional_branch_reg): Likewise.
25689 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
25691 (rs6000_reassociation_width): Likewise.
25692 (rs6000_invalid_binary_op): Likewise.
25693 (fusion_p9_p): Likewise.
25694 (emit_fusion_p9_load): Likewise.
25695 (emit_fusion_p9_store): Likewise.
25696 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
25698 (riscv_hard_regno_mode_ok_p): Likewise.
25699 (riscv_address_insns): Likewise.
25700 (riscv_split_symbol): Likewise.
25701 (riscv_legitimize_move): Likewise.
25702 (riscv_function_value): Likewise.
25703 (riscv_hard_regno_nregs): Likewise.
25704 (riscv_expand_builtin): Likewise.
25705 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
25706 (riscv_build_integer): Likewise.
25707 (riscv_split_integer): Likewise.
25708 (riscv_legitimate_constant_p): Likewise.
25709 (riscv_cannot_force_const_mem): Likewise.
25710 (riscv_regno_mode_ok_for_base_p): Likewise.
25711 (riscv_valid_base_register_p): Likewise.
25712 (riscv_valid_offset_p): Likewise.
25713 (riscv_valid_lo_sum_p): Likewise.
25714 (riscv_classify_address): Likewise.
25715 (riscv_legitimate_address_p): Likewise.
25716 (riscv_address_insns): Likewise.
25717 (riscv_load_store_insns): Likewise.
25718 (riscv_force_binary): Likewise.
25719 (riscv_split_symbol): Likewise.
25720 (riscv_force_address): Likewise.
25721 (riscv_legitimize_address): Likewise.
25722 (riscv_move_integer): Likewise.
25723 (riscv_legitimize_const_move): Likewise.
25724 (riscv_legitimize_move): Likewise.
25725 (riscv_address_cost): Likewise.
25726 (riscv_subword): Likewise.
25727 (riscv_output_move): Likewise.
25728 (riscv_canonicalize_int_order_test): Likewise.
25729 (riscv_emit_int_order_test): Likewise.
25730 (riscv_function_arg_boundary): Likewise.
25731 (riscv_pass_mode_in_fpr_p): Likewise.
25732 (riscv_pass_fpr_single): Likewise.
25733 (riscv_pass_fpr_pair): Likewise.
25734 (riscv_get_arg_info): Likewise.
25735 (riscv_function_arg): Likewise.
25736 (riscv_function_arg_advance): Likewise.
25737 (riscv_arg_partial_bytes): Likewise.
25738 (riscv_function_value): Likewise.
25739 (riscv_pass_by_reference): Likewise.
25740 (riscv_setup_incoming_varargs): Likewise.
25741 (riscv_print_operand): Likewise.
25742 (riscv_elf_select_rtx_section): Likewise.
25743 (riscv_save_restore_reg): Likewise.
25744 (riscv_for_each_saved_reg): Likewise.
25745 (riscv_register_move_cost): Likewise.
25746 (riscv_hard_regno_mode_ok_p): Likewise.
25747 (riscv_hard_regno_nregs): Likewise.
25748 (riscv_class_max_nregs): Likewise.
25749 (riscv_memory_move_cost): Likewise.
25750 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
25751 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
25752 (rl78_addr_space_address_mode): Likewise.
25753 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
25755 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
25756 (rs6000_reassociation_width): Likewise.
25757 (rs6000_invalid_binary_op): Likewise.
25758 (fusion_p9_p): Likewise.
25759 (emit_fusion_p9_load): Likewise.
25760 (emit_fusion_p9_store): Likewise.
25761 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
25762 (ok_for_simple_move_operands): Likewise.
25763 (ok_for_simple_move_strict_operands): Likewise.
25764 (ok_for_simple_arith_logic_operands): Likewise.
25765 (visium_legitimize_reload_address): Likewise.
25766 (visium_select_cc_mode): Likewise.
25767 (output_cbranch): Likewise.
25768 (visium_split_double_move): Likewise.
25769 (visium_expand_copysign): Likewise.
25770 (visium_expand_int_cstore): Likewise.
25771 (visium_expand_fp_cstore): Likewise.
25772 * config/visium/visium.c (visium_pass_by_reference): Likewise.
25773 (visium_function_arg): Likewise.
25774 (visium_function_arg_advance): Likewise.
25775 (visium_libcall_value): Likewise.
25776 (visium_setup_incoming_varargs): Likewise.
25777 (visium_legitimate_constant_p): Likewise.
25778 (visium_legitimate_address_p): Likewise.
25779 (visium_legitimize_address): Likewise.
25780 (visium_secondary_reload): Likewise.
25781 (visium_register_move_cost): Likewise.
25782 (visium_memory_move_cost): Likewise.
25783 (prepare_move_operands): Likewise.
25784 (ok_for_simple_move_operands): Likewise.
25785 (ok_for_simple_move_strict_operands): Likewise.
25786 (ok_for_simple_arith_logic_operands): Likewise.
25787 (visium_function_value_1): Likewise.
25788 (rtx_ok_for_offset_p): Likewise.
25789 (visium_legitimize_reload_address): Likewise.
25790 (visium_split_double_move): Likewise.
25791 (visium_expand_copysign): Likewise.
25792 (visium_expand_int_cstore): Likewise.
25793 (visium_expand_fp_cstore): Likewise.
25794 (visium_split_cstore): Likewise.
25795 (visium_select_cc_mode): Likewise.
25796 (visium_split_cbranch): Likewise.
25797 (output_cbranch): Likewise.
25798 (visium_print_operand_address): Likewise.
25799 * expmed.c (flip_storage_order): Likewise.
25800 * expmed.h (emit_cstore): Likewise.
25801 (flip_storage_order): Likewise.
25802 * genrecog.c (validate_pattern): Likewise.
25803 * hsa-gen.c (gen_hsa_addr): Likewise.
25804 * internal-fn.c (expand_arith_overflow): Likewise.
25805 * ira-color.c (allocno_copy_cost_saving): Likewise.
25806 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25807 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25808 (process_invariant_for_inheritance): Likewise.
25809 * lra-eliminations.c (move_plus_up): Likewise.
25810 * omp-low.c (lower_oacc_reductions): Likewise.
25811 * simplify-rtx.c (simplify_subreg): Likewise.
25812 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25813 (TARGET_CHKP_BOUND_MODE): Likewise..
25814 * targhooks.c (default_chkp_bound_mode): Likewise.
25815 (default_setup_incoming_vararg_bounds): Likewise.
25816 * targhooks.h (default_chkp_bound_mode): Likewise.
25817 (default_setup_incoming_vararg_bounds): Likewise.
25818 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25819 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25820 (have_whole_vector_shift): Likewise.
25821 * tree-vect-stmts.c (vectorizable_load): Likewise.
25822 * doc/tm.texi: Regenerate.
25824 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25826 Graceful degrade if Binutils PR21472 is not available.
25829 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25830 .rodata in flash test fails.
25831 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25832 * confgure: Regenerate.
25833 * config.in: Regenerate.
25834 * config/avr/avr.c (avr_asm_named_section)
25835 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25836 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25837 (avr_asm_init_sections): Same.
25839 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25841 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25842 (fma<VH:mode>4_intrinsic): Likewise.
25843 (*fmsub<VCVTF:mode>4): Likewise.
25844 (*fmsub<VH:mode>4_intrinsic): Likewise.
25846 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25849 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25850 Don't depend on "optimize > 0".
25851 (out_movhi_r_mr, out_movqi_mr_r): Same.
25852 (out_movhi_mr_r, out_movqi_r_mr): Same.
25853 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25854 io_address_operand on "optimize > 0".
25856 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25858 * tree-loop-distribution.c: Add general explanantion on the pass.
25859 (generate_loops_for_partition): Mark distributed loop.
25860 (pg_add_dependence_edges): New parameter. Handle alias data
25861 dependence specially and record it in the parameter if asked.
25862 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25863 (init_partition_graph_vertices, add_partition_graph_edge): New.
25864 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25865 (free_partition_graph_vdata, build_partition_graph): New.
25866 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25867 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25868 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25869 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25870 (version_for_distribution_p, finalize_partitions): New.
25871 (distribute_loop): Handle alias data dependence specially. Factor
25872 out loop fusion code as functions and call these functions.
25874 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25876 * tree-loop-distribution.c (classify_partition): New parameter and
25877 better handle reduction statement.
25878 (rdg_build_partitions): Revise comment.
25879 (distribute_loop): Compute statements in all partitions and pass it
25880 to classify_partition.
25882 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25884 * tree-loop-distribution.c (enum partition_type): New.
25885 (struct partition): New field type.
25886 (partition_merge_into): Add parameter. Update partition type.
25887 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25888 (build_rdg_partition_for_vertex): Compute partition type.
25889 (rdg_build_partitions): Dump partition type.
25890 (distribute_loop): Update calls to partition_merge_into.
25892 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25894 * tree-loop-distribution.c (struct ddr_hasher): New.
25895 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25897 (classify_partition): Call get_data_dependence.
25898 (pg_add_dependence_edges): Ditto.
25899 (distribute_loop): Release data dependence hash table.
25901 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25903 * tree-loop-distribution.c (ref_base_address): Delete.
25904 (similar_memory_accesses): Rename ...
25905 (share_memory_accesses): ... to this. Check if partitions access
25906 the same memory reference.
25907 (distribute_loop): Call share_memory_accesses.
25909 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25911 * tree-loop-distribution.c (struct partition): New field recording
25912 its data reference.
25913 (partition_alloc, partition_free): Init and release data refs.
25914 (partition_merge_into): Merge data refs.
25915 (build_rdg_partition_for_vertex): Collect data refs for partition.
25916 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25918 (distribute_loop): Remve data refs from vertice data of partition
25921 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25923 * tree-loop-distribution.c (params.h): Include header file.
25924 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25925 (datarefs_vec): New global var.
25926 (create_rdg_vertices): Use datarefs_vec directly.
25927 (free_rdg): Don't free data references.
25928 (build_rdg): Update use. Don't free data references.
25929 (distribute_loop): Compute global variable for data references.
25930 Bail out if there are too many data references.
25932 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25934 * tree-loop-distribution.c (loop_nest): New global var.
25935 (build_rdg): Use loop directly, rather than loop nest.
25936 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25938 (distribute_loop): Compute global variable loop nest. Update use.
25940 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25942 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25943 (partition_merge_into): New parameter. Dump reason for fusion.
25944 (distribute_loop): Update use of partition_merge_into.
25946 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25948 * tree-loop-distribution.c (bb_top_order_index): New.
25949 (bb_top_order_index_size, bb_top_order_cmp): New.
25950 (stmts_from_loop): Use topological order.
25951 (pass_loop_distribution::execute): Compute and release topological
25952 order for basic blocks.
25954 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25956 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25959 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25961 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25962 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25963 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25964 * internal-fn.def (LOOP_DIST_ALIAS): New.
25965 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25966 (fold_loop_internal_call): ... this.
25967 (vect_loop_dist_alias_call): New function.
25968 (set_uid_loop_bbs): Call fold_loop_internal_call.
25969 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25972 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25975 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25976 Require dead FLAGS_REG at the beginning of a peephole.
25978 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25981 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25982 arguments in the call to __builtin_ia32_sbb_u32.
25983 (_subborrow_u64): Swap _X and _Y arguments in the call to
25984 __builtin_ia32_sbb_u64.
25986 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25989 * tree-vrp.c (compare_assert_loc): Turn into a function template
25990 with stable template parameter. Only test if a->e is NULL,
25991 !a->e == !b->e has been verified already. Use e == NULL or
25992 e != NULL instead of e or ! e tests. If stable is true, don't use
25993 iterative_hash_expr, on the other side allow a or b or both NULL
25994 and sort the NULLs last.
25995 (process_assert_insertions): Sort using compare_assert_loc<false>
25996 instead of compare_assert_loc, later sort using
25997 compare_assert_loc<true> before calling process_assert_insertions_for
25998 in a loop. Use break instead of continue once seen NULL pointer.
26000 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26002 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
26003 Cortex-R7 and Cortex-R8 processors.
26005 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
26007 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
26008 uninitialized while src is not.
26010 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
26012 * common/config/arm/arm-common.c: Adjust include path for
26014 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
26015 (arm-cpu.h): Create in build directory. Adjust dependency rules.
26016 (arm-cpu-data.h): Likewise.
26017 (arm-cpu-cdata.h): Likewise.
26018 * config/arm/arm-cpu.h: Delete.
26019 * config/arm/arm-cpu-cdata.h: Delete.
26020 * config/arm/arm-cpu-data.h: Delete.
26022 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
26024 * config/arm/arm-cpus.in (cortex-a55): New.
26025 (cortex-a75): Likewise.
26026 (cortex-a75.cortex-a55): Likewise.
26027 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
26029 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
26030 * config/arm/arm-cpu-cdata.h: Regenerate.
26031 * config/arm/arm-cpu-data.h: Regenerate.
26032 * config/arm/arm-cpu.h: Regenerate.
26033 * config/arm/arm-tables.opt: Regenerate.
26034 * config/arm/arm-tune.md: Regenerate.
26036 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
26038 * haifa-sched.c (sched_create_recovery_edges): Update profile.
26040 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
26042 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
26045 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
26047 PR tree-optimization/81292
26048 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
26049 full_string_p, also call adjust_related_strinfos if the adjustment
26050 is simple, otherwise invalidate related strinfos.
26052 2017-07-04 Martin Liska <mliska@suse.cz>
26055 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
26056 newly created variable as DECL_IGNORED_P.
26058 2017-07-04 Martin Liska <mliska@suse.cz>
26061 * ipa-inline.c (inline_small_functions):
26062 Use xstrdup_for_dump.
26064 2017-07-04 Tom de Vries <tom@codesourcery.com>
26066 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
26068 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
26071 * config/darwin.c (darwin_function_switched_text_sections):
26074 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
26076 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
26078 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
26080 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
26082 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26084 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
26085 min_profitable_iters, and th as inclusive lower bounds.
26086 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
26087 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
26088 for min_profitable_iters and min_profitable_estimate.
26089 (vect_transform_loop): Treat th as an inclusive lower bound.
26090 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
26092 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
26095 * config/darwin.c (darwin_function_switched_text_sections):
26096 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
26097 in two pieces, and suppress the use of buf.
26099 2017-07-03 Nathan Sidwell <nathan@acm.org>
26101 * hash-table.h (hash_table_mod1): Fix indentation.
26103 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26105 PR middle-end/81290
26106 * predict.c (force_edge_cold): Be more careful about propagation
26108 * profile-count.h (profile_probability::guessed,
26109 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
26111 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
26113 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
26115 * doc/invoke.texi (rcpc architecture extension): Document it.
26117 2017-07-03 Richard Biener <rguenther@suse.de>
26119 PR tree-optimization/60510
26120 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
26121 the scalar reduction PHI and use it.
26122 (vectorizable_reduction): Properly guard the single_defuse_cycle
26123 path for non-SLP reduction chains where we cannot use it.
26124 Rework reduc_def/index and vector type deduction. Rework
26125 vector operand gathering during reduction op code-gen.
26126 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
26127 chains dissolve the chain and leave it to non-SLP reduction
26130 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26132 * tree-data-ref.h (dr_alignment): Declare.
26133 * tree-data-ref.c (dr_alignment): New function.
26134 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
26135 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
26137 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
26139 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26141 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
26142 and base_misalignment fields.
26143 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
26144 * tree-data-ref.c: Include builtins.h.
26145 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
26146 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
26147 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
26148 * tree-vect-data-refs.c: Include tree-cfg.h.
26149 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
26150 fields instead of calculating an alignment here.
26151 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
26152 innermost_loop_behavior fields.
26154 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26156 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
26158 (DR_STEP_ALIGNMENT): New macro.
26159 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
26160 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
26161 (create_data_ref): Print it.
26162 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
26163 to tell whether the step preserves vector (mis)alignment.
26164 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26165 Move the check for an integer step and generalise to all INTEGER_CST.
26166 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
26167 Print the outer step alignment.
26169 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26171 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
26172 with offset_alignment.
26173 (DR_ALIGNED_TO): Delete.
26174 (DR_OFFSET_ALIGNMENT): New macro.
26175 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
26176 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
26177 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
26178 (create_data_ref): Likewise.
26179 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26180 (vect_analyze_data_refs): Likewise.
26181 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
26182 creating dummy innermost behavior.
26184 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26186 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
26187 with a "innermost_loop_behavior *" and refeence tree.
26188 * tree-data-ref.c (dr_analyze_innermost): Likewise.
26189 (create_data_ref): Update call accordingly.
26190 * tree-predcom.c (find_looparound_phi): Likewise.
26192 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26194 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
26195 fields with dr_wrt_vec_loop.
26196 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
26197 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
26198 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
26199 (vect_dr_behavior): New function.
26200 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26201 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
26202 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
26203 track whether the step preserves the misalignment.
26204 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26205 Use vect_dr_behavior.
26206 (vect_setup_realignment): Update call accordingly.
26207 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
26208 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
26209 call to vect_create_addr_base_for_vector_ref.
26210 (vect_create_cond_for_align_checks): Likewise.
26211 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
26212 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
26213 (vect_recog_mask_conversion_pattern): Likewise.
26214 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
26215 (new_stmt_vec_info): Remove redundant zeroing.
26217 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
26219 * common/config/arm/arm-common.c (arm_be8_option): New function.
26220 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
26221 (ISA_ARMv6): Add isa_bit_be8.
26222 * config/arm/arm.h (arm_be8_option): Add prototype.
26223 (BE8_SPEC_FUNCTION): New define.
26224 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
26225 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
26226 (mlittle-endian): Similarly.
26227 (mbe8, mbe32): New options.
26228 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
26229 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
26231 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26233 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
26235 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26237 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
26238 (cleanup_tree_cfg_bb): Use it.
26239 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
26241 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
26243 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26246 * loop-doloop.c (add_test): Update profile.
26248 2017-07-03 Martin Liska <mliska@suse.cz>
26251 * sanopt.c (rewrite_usage_of_param): New function.
26252 (sanitize_rewrite_addressable_params): Likewise.
26253 (pass_sanopt::execute): Call rewrite_usage_of_param.
26255 2017-07-03 Richard Biener <rguenther@suse.de>
26257 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
26258 back to using VIEW_CONVERT_EXPR.
26260 2017-07-03 Martin Liska <mliska@suse.cz>
26263 * doc/extend.texi: Document when a resolver function is
26264 generated for target_clones.
26266 2017-07-03 Martin Liska <mliska@suse.cz>
26268 * asan.c (asan_emit_stack_protection): Unpoison just red zones
26269 and shadow memory of auto variables which are subject of
26270 use-after-scope sanitization.
26271 (asan_expand_mark_ifn): Add do set only when is_poison.
26273 2016-07-03 Richard Biener <rguenther@suse.de>
26275 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
26277 (vect_force_simple_reduction): Record reduction def -> phi mapping.
26278 (vectorizable_reduction): Perform reduction PHI creation when
26279 visiting a reduction PHI and adjust and simplify code generation
26280 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
26281 (vect_transform_loop): Visit reduction PHIs.
26282 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
26283 defs into the SLP tree.
26284 (vect_build_slp_tree): Reduction defs terminate the recursion.
26285 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
26287 (vect_get_vec_defs_for_stmt_copy): Export.
26288 (vect_get_vec_defs): Likewise.
26289 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
26291 (vect_get_vec_defs_for_stmt_copy): Declare.
26292 (vect_get_vec_defs): Likewise.
26294 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26296 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
26297 parameter with a "loop" parameter and use it instead of the
26298 loop containing DR_STMT. Don't check simple_iv when doing
26299 BB analysis. Describe the two analysis modes in the comment.
26301 2017-07-03 Tom de Vries <tom@codesourcery.com>
26303 PR tree-optimization/69468
26304 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
26305 (find_same_succ_bb): Handle ignore_edge_flags.
26307 2017-07-03 Tom de Vries <tom@codesourcery.com>
26309 PR tree-optimization/81192
26310 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
26312 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
26314 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
26316 2017-07-03 Tom de Vries <tom@codesourcery.com>
26318 PR tree-optimization/81192
26319 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
26320 BB_SAME_SUCC (bb) == NULL.
26322 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26324 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
26327 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26329 * dumpfile.c: Include profile-count.h
26330 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
26332 (insert_cond_bb): Update profile.
26333 * tree-cfg.h (insert_cond_bb): Update prototype.
26334 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
26335 * tree-dump.c: Do not include tree-cfg.
26337 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26339 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
26341 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26343 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
26346 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26348 * tree-complex.c (expand_complex_div_wide): update profile.
26350 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26351 Alan Hayward <alan.hayward@arm.com>
26352 David Sherwood <david.sherwood@arm.com>
26354 * Makefile.in (MACHMODE_H): Remove insn-modes.h
26355 (CORETYPES_H): New define.
26356 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
26357 (insn-modes-inline.h, s-modes-inline-h): New rules.
26358 (generated_files): Add insn-modes-inline.h.
26359 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
26360 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
26361 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
26362 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
26363 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
26364 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
26365 (build/gencodes.o, build/genconditions.o): Likewise.
26366 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
26367 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
26368 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
26369 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
26370 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
26371 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
26372 * coretypes.h: Include everything up to real.h for generators.
26373 Include insn-modes.h first. Include wide-int-print.h after
26374 wide-int.h. Include insn-modes-inline.h and then machmode.h.
26375 * machmode.h: Don't include insn-modes.h here.
26376 * function-tests.c: Remove includes of signop.h, machmode.h,
26377 double-int.h and wide-int.h.
26379 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
26381 * optc-save-gen.awk: Likewise.
26382 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
26383 * godump.c: Remove include of wide-int-print.h.
26384 * pretty-print.h: Likewise.
26385 * wide-int-print.cc: Likewise.
26386 * wide-int.cc: Likewise.
26387 * hash-map-tests.c: Remove include of signop.h.
26388 * hash-set-tests.c: Likewise.
26389 * rtl-tests.c: Likewise.
26390 * mkconfig.sh: Remove include of machmode.h.
26391 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
26393 (emit_insn_modes_inline_h): ...this new function. Emit the code
26394 into an insn-modes-inline.h header file, adding appropriate
26395 include guards and end comments.
26396 (emit_insn_modes_c_header): Remove include of machmode.h.
26397 (emit_min_insn_modes_c_header): Include coretypes.h rather than
26399 (main): Handle -i flag and call emit_insn_modes_inline_h when
26402 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26404 * tree-ssa-strlen.c (strinfo): Rename the length field to
26405 nonzero_chars. Add a full_string_p field.
26406 (compare_nonzero_chars, zero_length_string_p): New functions.
26407 (get_addr_stridx): Add an offset_out parameter.
26408 Use compare_nonzero_chars.
26409 (get_stridx): Update accordingly. Use compare_nonzero_chars.
26410 (new_strinfo): Update after above changes to strinfo.
26411 (set_endptr_and_length): Set full_string_p.
26412 (get_string_length): Update after above changes to strinfo.
26413 (unshare_strinfo): Update call to new_strinfo.
26414 (maybe_invalidate): Likewise.
26415 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
26416 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
26417 as a uhwi instead of an shwi. Update after above changes to
26418 strinfo and new_strinfo.
26419 (zero_length_string): Assert that chainsi contains full strings.
26420 Use zero_length_string_p. Update call to new_strinfo.
26421 (adjust_related_strinfos): Update after above changes to strinfo.
26422 Copy full_string_p from origsi.
26423 (adjust_last_stmt): Use zero_length_string_p.
26424 (handle_builtin_strlen): Update after above changes to strinfo and
26425 new_strinfo. Install the lhs as the string length if the previous
26426 entry didn't describe a full string.
26427 (handle_builtin_strchr): Update after above changes to strinfo
26429 (handle_builtin_strcpy): Likewise.
26430 (handle_builtin_strcat): Likewise.
26431 (handle_builtin_malloc): Likewise.
26432 (handle_pointer_plus): Likewise.
26433 (handle_builtin_memcpy): Likewise. Track nonzero characters
26434 that aren't necessarily followed by a nul terminator.
26435 (handle_char_store): Likewise.
26437 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26439 PR tree-optimization/80769
26440 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
26441 for malloc and calloc. Document the new invariant that all related
26442 strinfos have delayed lengths or none do.
26443 (verify_related_strinfos): Move earlier in file.
26444 (set_endptr_and_length): New function, split out from...
26445 (get_string_length): ...here. Also set the lengths of related
26447 (zero_length_string): Assert that chainsi has known (rather than
26449 (adjust_related_strinfos): Likewise.
26451 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26453 PR tree-optimization/81136
26454 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
26455 assert that two references with the same misalignment have the same
26456 compile-time misalignment if those compile-time misalignments
26459 2017-07-01 Andi Kleen <ak@linux.intel.com>
26461 * print-tree.c (print_node): Print all attributes.
26463 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26465 * cfg.c (scale_bbs_frequencies): New function.
26466 * cfg.h (scale_bbs_frequencies): Declare it.
26467 * cfgloopanal.c (single_likely_exit): Cleanup.
26468 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
26470 (scale_loop_profile): Likewise.
26471 (loop_version): Likewise.
26472 (create_empty_loop_on_edge): Update.
26473 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
26474 scale_loop_frequencies, scale_loop_profile, loopify,
26475 loop_version): Update prototypes.
26476 * modulo-sched.c (sms_schedule): Update.
26477 * predict.c (unlikely_executed_edge_p): Also check probability.
26478 (probably_never_executed_edge_p): Fix typo.
26479 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26480 * tree-parloops.c (gen_parallel_loop): Update.
26481 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
26482 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26483 * tree-ssa-loop-split.c (split_loop): Update.
26484 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26485 * tree-vect-loop-manip.c (vect_do_peeling): Update.
26486 (vect_loop_versioning): Update.
26487 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26489 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26491 * trans-mem.c (split_bb_make_tm_edge): Update profile.
26493 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26495 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
26496 to keep profile consistent.
26498 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26500 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
26501 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
26502 * profile-count.h (max_safe_multiplier): Make unsigned.
26503 (profile_count::guessed_zero): New.
26505 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26507 * bb-reorder.c (fix_up_crossing_landing_pad,
26508 fix_crossing_conditional_branches): Use make_single_succ_edge
26509 to keep profile consistent.
26511 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26513 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
26516 2017-07-01 Jakub Jelinek <jakub@redhat.com>
26519 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
26520 the right scopes, make sure cond_jump isn't preserved between multiple
26521 iterations. Search for fallthru edge whenever there are 3+ edges and
26522 use find_fallthru_edge for it.
26524 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26526 Patch by Alexander Monakov <amonakov@ispras.ru>
26527 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
26528 probabilities consistently.
26530 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26532 * pa.c (pa_expand_compare_and_swap_loop): Update call of
26533 emit_cmp_and_jump_insns.
26535 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26538 * tree-inline.c (expand_call_inline): Combine profile statuses.
26540 2017-06-30 Andrew Pinski <apinski@cavium.com>
26542 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
26543 fold_stmt returned true.
26545 2017-06-30 Nathan Sidwell <nathan@acm.org>
26547 * ggc.h (empty_string): Delete.
26548 * cfgexpand.c (expand_asm_stmt): Use plain "".
26549 * optabs.c (expand_asm_memory_barrier): Likewise.
26550 * stringpool.c (empty_string): Delete.
26551 (digit_vector, digit_string): Delete.
26552 (ggc_alloc_string): Use plain "", don't optimize single digit
26553 strings. Use ggc_alloc_atomic.
26555 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
26557 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
26558 comparison set and one other set, use the cost of the non-comparison
26561 2017-06-30 Nathan Sidwell <nathan@acm.org>
26563 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
26566 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
26568 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
26569 loops. Remove now unneeded calls to gimple_switch_set_label() that
26570 just set removed labels to NULL_TREE.
26572 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
26574 * tree-ssanames.c (set_range_info_raw): Abstract from ...
26575 (set_range_info): ...here. Only call set_range_info_raw if domain
26577 (set_nonzero_bits): Call set_range_info_raw.
26578 * tree-ssanames.h (set_range_info_raw): New.
26580 2017-06-30 Jakub Jelinek <jakub@redhat.com>
26583 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
26584 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
26585 of nonimmediate_operand and <store_mask_constraint> instead of m
26586 for the input operand. For V8FI iterator, always split if input
26587 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
26588 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
26589 <store_mask_predicate> instead of register_operand and
26590 <store_mask_constraint> instead of v for the input operand. Make
26591 sure both operands aren't MEMs for if not <mask_applied>.
26593 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
26595 * lto-wrapper.c (copy_file) Close both file descriptors before
26598 2017-06-30 Martin Liska <mliska@suse.cz>
26601 * multiple_target.c (create_dispatcher_calls): Make ifunc
26602 also for function that don't have calls or are not referenced.
26604 2017-06-30 Richard Biener <rguenther@suse.de>
26606 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
26607 analyze the first scalar stmt. Move vector type computation
26608 for the BB case here from ...
26609 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
26610 live operation processing in the SLP case properly.
26612 2017-06-30 Richard Biener <rguenther@suse.de>
26614 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
26616 2017-06-30 Martin Liska <mliska@suse.cz>
26619 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
26620 before BUILT_IN_UNWIND_RESUME when ASAN is used.
26622 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
26624 * doc/invoke.texi (AArch64): Add missing options and remove redundant
26627 2017-06-30 Richard Biener <rguenther@suse.de>
26629 PR tree-optimization/81249
26630 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
26631 condition reduction result to original scalar type.
26633 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26635 * profile-count.h (enum profile_quality): Fix typos and whitespace
26638 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26640 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
26641 type for branch probabilities.
26643 2017-06-29 Julian Brown <julian@codesourcery.com>
26644 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26646 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
26647 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
26648 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
26649 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
26651 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26653 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
26654 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
26655 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
26656 CC usage from generic code to here.
26657 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
26658 CC usage into the target macros.
26660 2017-06-29 Maya Rashish <coypu@sdf.org>
26662 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
26665 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26667 * arm/arm-builtins.c: Include profile-count.h
26668 * except.c (sjlj_emit_function_enter): Use
26669 profile_probability::unlikely.
26671 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26673 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
26674 and tocrel_offset be pointer args rather than implicitly using
26676 (legitimate_constant_pool_address_p, rs6000_emit_move,
26677 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
26678 tocrel_offset and use in toc_relative_expr_p call.
26679 (print_operand, print_operand_address): Use static tocrel_base_oac
26680 and tocrel_offset_oac.
26681 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
26684 2017-06-29 Maya Rashish <coypu@sdf.org>
26686 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
26688 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
26690 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
26691 objects, take into account only the alignment of 'op0' and 'mode1' if
26694 2017-06-29 Steve Ellcey <sellcey@cavium.com>
26696 * ccmp.c (ccmp_tree_comparison_p): New function.
26697 (ccmp_candidate_p): Update to use above function.
26698 (get_compare_parts): New function.
26699 (expand_ccmp_next): Update to use new functions.
26700 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
26702 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
26703 take mode as argument.
26704 * ccmp.h (expand_ccmp_expr): Add mode as argument.
26705 * expr.c (expand_expr_real_1): Pass mode as argument.
26707 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
26709 * combine.c (combine_instructions): Print insns to dump_file, together
26712 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26714 * asan.c (asan_emit_stack_protection): Update.
26715 (create_cond_insert_point): Update.
26716 * auto-profile.c (afdo_propagate_circuit): Update.
26717 * basic-block.h (struct edge_def): Turn probability to
26718 profile_probability.
26719 (EDGE_FREQUENCY): Update.
26720 * bb-reorder.c (find_traces_1_round): Update.
26721 (better_edge_p): Update.
26722 (sanitize_hot_paths): Update.
26723 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
26724 (make_single_succ_edge): Update.
26725 (check_bb_profile): Update.
26726 (dump_edge_info): Update.
26727 (update_bb_profile_for_threading): Update.
26728 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
26729 probabilitycount to 0.
26730 * cfgbuild.c (compute_outgoing_frequencies): Update.
26731 * cfgcleanup.c (try_forward_edges): Update.
26732 (outgoing_edges_match): Update.
26733 (try_crossjump_to_edge): Update.
26734 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
26735 (expand_gimple_tailcall): Update.
26736 (construct_init_block): Use make_single_succ_edge.
26737 (construct_exit_block): Use make_single_succ_edge.
26738 * cfghooks.c (verify_flow_info): Update.
26739 (redirect_edge_succ_nodup): Update.
26740 (split_edge): Update.
26741 (account_profile_record): Update.
26742 * cfgloopanal.c (single_likely_exit): Update.
26743 * cfgloopmanip.c (scale_loop_profile): Update.
26744 (set_zero_probability): Remove.
26745 (duplicate_loop_to_header_edge): Update.
26746 * cfgloopmanip.h (loop_version): Update prototype.
26747 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
26748 (force_nonfallthru_and_redirect): Update.
26749 (update_br_prob_note): Update.
26750 (rtl_verify_edges): Update.
26751 (purge_dead_edges): Update.
26752 (rtl_lv_add_condition_to_bb): Update.
26753 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
26754 * cgraphunit.c (init_lowered_empty_function): Update.
26755 (cgraph_node::expand_thunk): Update.
26756 * cilk-common.c: Include profile-count.h
26757 * dojump.c (inv): Remove.
26758 (jumpifnot): Update.
26759 (jumpifnot_1): Update.
26760 (do_jump_1): Update.
26762 (do_jump_by_parts_greater_rtx): Update.
26763 (do_compare_rtx_and_jump): Update.
26764 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
26765 do_jump_1. do_compare_rtx_and_jump): Update prototype.
26766 * dwarf2cfi.c: Include profile-count.h
26767 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
26768 (sjlj_emit_dispatch_table): Likewise.
26769 * explow.c: Include profile-count.h
26770 * expmed.c (emit_store_flag_force): Update.
26771 (do_cmp_and_jump): Update.
26772 * expr.c (compare_by_pieces_d::generate): Update.
26773 (compare_by_pieces_d::finish_mode): Update.
26774 (emit_block_move_via_loop): Update.
26775 (store_expr_with_bounds): Update.
26776 (store_constructor): Update.
26777 (expand_expr_real_2): Update.
26778 (expand_expr_real_1): Update.
26779 * expr.h (try_casesi, try_tablejump): Update prototypes.
26780 * gimple-pretty-print.c (dump_probability): Update.
26781 (dump_profile): New.
26782 (dump_gimple_label): Update.
26783 (dump_gimple_bb_header): Update.
26784 * graph.c (draw_cfg_node_succ_edges): Update.
26785 * hsa-gen.c (convert_switch_statements): Update.
26786 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
26787 (find_if_case_1): Update.
26788 (find_if_case_2): Update.
26789 * internal-fn.c (expand_arith_overflow_result_store): Update.
26790 (expand_addsub_overflow): Update.
26791 (expand_neg_overflow): Update.
26792 (expand_mul_overflow): Update.
26793 (expand_vector_ubsan_overflow): Update.
26794 * ipa-cp.c (good_cloning_opportunity_p): Update.
26795 * ipa-split.c (split_function): Use make_single_succ_edge.
26796 * ipa-utils.c (ipa_merge_profiles): Update.
26797 * loop-doloop.c (add_test): Update.
26798 (doloop_modify): Update.
26799 * loop-unroll.c (compare_and_jump_seq): Update.
26800 (unroll_loop_runtime_iterations): Update.
26801 * lra-constraints.c (lra_inheritance): Update.
26802 * lto-streamer-in.c (input_cfg): Update.
26803 * lto-streamer-out.c (output_cfg): Update.
26804 * mcf.c (adjust_cfg_counts): Update.
26805 * modulo-sched.c (sms_schedule): Update.
26806 * omp-expand.c (expand_omp_for_init_counts): Update.
26807 (extract_omp_for_update_vars): Update.
26808 (expand_omp_ordered_sink): Update.
26809 (expand_omp_for_ordered_loops): Update.
26810 (expand_omp_for_generic): Update.
26811 (expand_omp_for_static_nochunk): Update.
26812 (expand_omp_for_static_chunk): Update.
26813 (expand_cilk_for): Update.
26814 (expand_omp_simd): Update.
26815 (expand_omp_taskloop_for_outer): Update.
26816 (expand_omp_taskloop_for_inner): Update.
26817 * omp-simd-clone.c (simd_clone_adjust): Update.
26818 * optabs.c (expand_doubleword_shift): Update.
26819 (expand_abs): Update.
26820 (emit_cmp_and_jump_insn_1): Update.
26821 (expand_compare_and_swap_loop): Update.
26822 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26823 * predict.c (predictable_edge_p): Update.
26824 (edge_probability_reliable_p): Update.
26825 (set_even_probabilities): Update.
26826 (combine_predictions_for_insn): Update.
26827 (combine_predictions_for_bb): Update.
26828 (propagate_freq): Update.
26829 (estimate_bb_frequencies): Update.
26830 (force_edge_cold): Update.
26831 * profile-count.c (profile_count::dump): Add missing space into dump.
26832 (profile_count::debug): Add newline.
26833 (profile_count::differs_from_p): Explicitly convert to unsigned.
26834 (profile_count::stream_in): Update.
26835 (profile_probability::dump): New member function.
26836 (profile_probability::debug): New member function.
26837 (profile_probability::differs_from_p): New member function.
26838 (profile_probability::differs_lot_from_p): New member function.
26839 (profile_probability::stream_in): New member function.
26840 (profile_probability::stream_out): New member function.
26841 * profile-count.h (profile_count_quality): Rename to ...
26842 (profile_quality): ... this one.
26843 (profile_probability): New.
26844 (profile_count): Update.
26845 * profile.c (compute_branch_probabilities): Update.
26846 * recog.c (peep2_attempt): Update.
26847 * sched-ebb.c (schedule_ebbs): Update.
26848 * sched-rgn.c (find_single_block_region): Update.
26849 (compute_dom_prob_ps): Update.
26850 (schedule_region): Update.
26851 * sel-sched-ir.c (compute_succs_info): Update.
26852 * stmt.c (struct case_node): Update.
26853 (do_jump_if_equal): Update.
26854 (get_outgoing_edge_probs): Update.
26855 (conditional_probability): Update.
26856 (emit_case_dispatch_table): Update.
26857 (expand_case): Update.
26858 (expand_sjlj_dispatch_table): Update.
26859 (emit_case_nodes): Update.
26860 * targhooks.c: Update.
26861 * tracer.c (better_p): Update.
26862 (find_best_successor): Update.
26863 * trans-mem.c (expand_transaction): Update.
26864 * tree-call-cdce.c: Update.
26865 * tree-cfg.c (gimple_split_edge): Upate.
26866 (move_sese_region_to_fn): Upate.
26867 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26868 * tree-eh.c (lower_resx): Upate.
26869 (cleanup_empty_eh_move_lp): Upate.
26870 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26871 * tree-inline.c (copy_edges_for_bb): Update.
26872 (copy_cfg_body): Update.
26873 * tree-parloops.c (gen_parallel_loop): Update.
26874 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26875 (gimple_gen_time_profiler): Update.
26876 * tree-ssa-dce.c (remove_dead_stmt): Update.
26877 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26878 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26879 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26880 (unloop_loops): Update.
26881 (try_peel_loop): Update.
26882 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26883 * tree-ssa-loop-split.c (connect_loops): Update.
26884 (split_loop): Update.
26885 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26886 (hoist_guard): Update.
26887 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26888 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26889 (value_replacement): Update.
26890 * tree-ssa-reassoc.c (branch_fixup): Update.
26891 * tree-ssa-tail-merge.c (replace_block_by): Update.
26892 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26893 (create_edge_and_update_destination_phis): Update.
26894 (compute_path_counts): Update.
26895 (recompute_probabilities): Update.
26896 (update_joiner_offpath_counts): Update.
26897 (freqs_to_counts_path): Update.
26898 (duplicate_thread_path): Update.
26899 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26900 (struct switch_conv_info): Update.
26901 (gen_inbound_check): Update.
26902 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26903 (vect_do_peeling): Update.
26904 (vect_loop_versioning): Update.
26905 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26906 (optimize_mask_stores): Update.
26907 * ubsan.c (ubsan_expand_null_ifn): Update.
26908 * value-prof.c (gimple_divmod_fixed_value): Update.
26909 (gimple_divmod_fixed_value_transform): Update.
26910 (gimple_mod_pow2): Update.
26911 (gimple_mod_pow2_value_transform): Update.
26912 (gimple_mod_subtract): Update.
26913 (gimple_mod_subtract_transform): Update.
26914 (gimple_ic): Update.
26915 (gimple_stringop_fixed_value): Update.
26916 (gimple_stringops_transform): Update.
26917 * value-prof.h: Update.
26919 2017-06-29 Carl Love <cel@us.ibm.com>
26921 * config/rs6000/rs6000-c.c: Add support for built-in functions
26922 vector signed int vec_signed (vector float);
26923 vector signed long long vec_signed (vector double);
26924 vector signed int vec_signed2 (vector double, vector double);
26925 vector signed int vec_signede (vector double);
26926 vector signed int vec_signedo (vector double);
26927 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26928 instruction generator.
26929 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26930 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26931 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26933 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26934 vunsignede_v2df): Add define_expands.
26935 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26936 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26937 VEC_UNSIGNEDO): Add definitions.
26938 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26939 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26940 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26941 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26942 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26943 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26944 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26945 vec_unsignede, vec_unsignedo): Add builtin defines.
26946 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26948 * doc/extend.texi: Update the built-in documentation file for the
26949 new built-in functions.
26951 2017-06-29 Richard Biener <rguenther@suse.de>
26953 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26954 reduction chains to LOOP_VINFO_REDUCTIONS.
26955 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26956 SLP reductions after processing reduction chains.
26958 2017-06-29 Nathan Sidwell <nathan@acm.org>
26960 * builtins.c (fold_builtin_FUNCTION): Use
26961 lang_hooks.decl_printable_name.
26963 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26965 PR middle-end/81194
26966 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26967 with only one label.
26968 * stmt.c (expand_case): Assert NCASES is greater than one.
26970 2017-06-29 Richard Biener <rguenther@suse.de>
26972 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26974 (group_case_labels): Likewise.
26975 (find_taken_edge): Push sanity checking on val to workers...
26976 (find_taken_edge_cond_expr): ... here
26977 (find_taken_edge_switch_expr): ... and here, handle cases
26978 with just a default label.
26979 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26980 (group_case_labels): Likewise.
26981 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26982 group_case_labels does anything cleanup the CFG again.
26984 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26986 PR tree-optimization/81196
26987 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26988 exit condition comparing two IVs.
26990 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26992 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26993 profile to the dummy entry at the end of the list of architectures.
26994 * config/arm/arm-cpu-cdata.h: Regenerated.
26996 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26997 Michael Collison <michael.collison@arm.com>
27000 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
27002 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
27003 (*aarch64_reg_<mode>3_minus_mask): New pattern.
27004 (*aarch64_<optab>_reg_di3_mask2): New pattern.
27005 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
27006 of shift when the shift amount is masked with constant equal to
27007 the size of the mode.
27008 * config/aarch64/predicates.md (subreg_lowpart_operator): New
27011 2017-06-29 Martin Liska <mliska@suse.cz>
27013 * config/i386/i386.opt: Change range from [1,5] to [0,5].
27015 2017-06-29 Yury Gribov <tetra2005@gmail.com>
27018 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
27021 (edge_growth_cache_entry::edge_growth_cache_entry): New
27023 (reset_edge_growth_cache): Update to use constructor.
27025 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27027 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
27028 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
27029 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
27031 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
27033 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
27034 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
27036 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
27038 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
27039 (*-linux-uclibc*): Add t-uclibc tmake_file.
27040 * config/t-musl: New.
27041 * config/t-uclibc: New.
27043 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
27045 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
27047 (gen_comm_data): Emit architectural setting of arch_prof.
27048 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
27050 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
27051 (armv8-m.base, armv8-m.main): Likewise.
27052 * arm-protos.h (arm_build_target): Add profile field.
27053 (arch_option): Likewise.
27054 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
27056 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
27057 arm_active_target.profile.
27059 2017-06-28 Richard Biener <rguenther@suse.de>
27061 PR middle-end/81227
27062 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
27063 TYPE_OVERFLOW_WRAPS.
27064 * match.pd (negate_expr_p): Likewise.
27065 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
27066 fold_build2, not fold_binary.
27068 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
27070 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
27071 Convert memory address to Pmode.
27072 (aarch64_print_operand): Assert MEM operands are always Pmode.
27074 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
27077 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
27078 Remove redundant if.
27079 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
27080 * config/arm/aarch-common-protos.h
27081 (aarch_forward_to_shift_is_not_shifted_re): Remove.
27082 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
27084 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
27087 * multiple_target.c (create_dispatcher_calls): Set the default
27088 clone to be static, not public.
27090 2017-06-28 Richard Biener <rguenther@suse.de>
27092 * tree-vect-loop.c (vectorizable_reduction): Move special
27093 cond reduction IV var creation ...
27094 (vect_create_epilog_for_reduction): ... here. Remove induction_index
27095 parameter. Use STMT_VINFO_VECTYPE.
27096 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
27099 2017-06-28 Martin Liska <mliska@suse.cz>
27102 * ipa-visibility.c (non_local_p): Handle visibility.
27104 2017-06-28 Martin Liska <mliska@suse.cz>
27107 * common.opt: Add IntegerRange to various options.
27108 * opt-functions.awk (integer_range_info): New function.
27109 * optc-gen.awk: Add integer_range_info to cl_options struct.
27110 * opts-common.c (decode_cmdline_option): Handle
27111 CL_ERR_INT_RANGE_ARG.
27112 (cmdline_handle_error): Likewise.
27113 * opts.c (print_filtered_help): Show valid interval in
27114 when --help is provided.
27115 * opts.h (struct cl_option): Add range_min and range_max fields.
27116 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
27118 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
27120 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
27121 (x * C EQ/NE y * C): New transformation.
27123 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
27125 * genmultilib (combination_space): Accept '+' in option names.
27127 2017-06-28 Martin Liska <mliska@suse.cz>
27130 * asan.c (instrument_derefs): Bail out inner references
27131 that are hard register variables.
27133 2017-06-28 Jakub Jelinek <jakub@redhat.com>
27136 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
27137 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
27139 2017-06-28 Richard Biener <rguenther@suse.de>
27141 * tree-vectorizer.h (vect_get_vec_defs): Remove.
27142 (vect_get_slp_defs): Adjust.
27143 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
27145 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
27147 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
27148 get_initial_defs_for_reduction instead of vect_get_vec_defs.
27149 (vectorizable_reduction): Adjust.
27150 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
27152 (vect_get_slp_defs): Likewise.
27153 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
27154 (vectorizable_bswap): Adjust.
27155 (vectorizable_call): Likewise.
27156 (vectorizable_conversion): Likewise.
27157 (vectorizable_assignment): Likewise.
27158 (vectorizable_shift): Likewise.
27159 (vectorizable_operation): Likewise.
27160 (vectorizable_store): Likewise.
27161 (vectorizable_condition): Likewise.
27162 (vectorizable_comparison): Likewise.
27164 2017-06-28 Michael Collison <michael.collison@arm.com>
27167 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
27169 (arm_gen_movmemqi): Removed unused variable 'i'.
27170 Convert 'for' loop into 'while' loop.
27171 (arm_expand_prologue): Remove last unnecessary set of insn.
27172 (thumb_pop): Remove unused variable 'pushed_words'.
27173 (thumb_exit): Remove last unnecessary set of regs_to_pop.
27175 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27177 * config/s390/predicates.md: Use s390_rel_address_ok_p.
27178 * config/s390/s390-protos.h: Add prototype of
27179 s390_rel_address_ok_p.
27180 * config/s390/s390.c (s390_got_symbol): New function.
27181 (s390_rel_address_ok_p): New function.
27182 (legitimize_pic_address): Use s390_rel_address_ok_p.
27183 (s390_load_got): Use s390_got_symbol.
27184 (s390_option_override): Issue error if
27185 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
27186 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
27188 * config/s390/s390.opt: New option mpic-data-is-text-relative.
27190 2017-06-27 Andrew Pinski <apinski@cavium.com>
27192 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
27193 (X * copysign (1.0, X)): New pattern.
27194 (X * copysign (1.0, -X)): New pattern.
27195 (copysign (-1.0, CST)): New pattern.
27197 2017-06-27 Joseph Myers <joseph@codesourcery.com>
27199 * genmultilib (combination_space): Remove variable.
27200 Validate reuse rules against regular expression for any sequence
27201 of multilib options in any order.
27203 2017-06-27 Michael Collison <michael.collison@arm.com>
27205 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
27206 call aarch64_split_simd_combine.
27207 * (aarch64_combine_internal<mode>): Delete pattern.
27208 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
27209 Allow register and subreg operands.
27211 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27213 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
27214 specific need, just fallback on defaults.
27215 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
27217 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27218 Olivier Hainque <hainque@adacore.com>
27220 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
27222 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
27223 targets. Pick a default if no particular attempt applied.
27224 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
27227 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27229 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
27230 (x86_64-wrs-vxworks7): Likewise.
27232 2017-06-27 Marek Polacek <polacek@redhat.com>
27235 * ubsan.c (instrument_null): Check get_base_address's result for null.
27237 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27239 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
27241 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27243 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
27244 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
27245 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
27246 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
27247 New function types.
27248 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
27249 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
27250 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
27251 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
27252 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
27253 BUILT_IN_FEUPDATEENV): New builtins.
27254 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
27255 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
27256 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
27257 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
27259 (builtin_structptr_types): Adjust size.
27260 * tree.c (builtin_structptr_types): Add four entries.
27262 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27263 Olivier Hainque <hainque@adacore.com>
27265 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
27266 (TLS_SYM): New local macro, forcing reference to __tls__ on
27267 link command lines for VxWorks 7 RTPs, triggering initialization
27269 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
27270 OS features TLS support, true for RTPs on VxWorks 7.
27271 * config/vxworks.c (vxworks_override_options): Setup emutls
27274 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27276 * predict.c (test_prediction_value_range): Use -1U instead of -1
27277 to avoid narrowing conversion warning.
27278 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
27279 to avoid narrowing conversion warning.
27280 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
27282 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
27284 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27286 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
27287 64bit configurations.
27288 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
27289 (SIZE_TYPE): Likewise.
27290 * config/vxworks.c (vxworks_emutls_var_fields): Use
27291 long_unsigned_type_node instead of unsigned_type_node as the offset
27292 field type, which is "pointer" mode in emutls.c.
27294 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27297 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
27299 PR middle-end/81207
27300 * gimple-fold.c (replace_call_with_call_and_fold): Handle
27301 gimple_vuse copying separately from gimple_vdef copying.
27303 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27305 * value-prof.c (free_hist): Remove call to memset and the enclosing if
27308 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27309 Olivier Hainque <hainque@adacore.com>
27311 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
27312 for all vxworks7 targets.
27313 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
27314 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
27315 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
27316 variations for VX6/VX7 and 32/64bits later on in ...
27317 (VXWORKS_LIB_SPEC): Leverage new macros.
27318 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
27319 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
27321 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27323 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
27324 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
27326 2017-06-26 Carl Love <cel@us.ibm.com>
27328 * config/rs6000/rs6000-c.c: Add support for built-in functions
27329 vector bool char vec_reve (vector bool char);
27330 vector signed char vec_reve (vector signed char);
27331 vector unsigned char vec_reve (vector unsigned char);
27332 vector bool int vec_reve (vector bool int);
27333 vector signed int vec_reve (vector signed int);
27334 vector unsigned int vec_reve (vector unsigned int);
27335 vector bool long long vec_reve (vector bool long long);
27336 vector signed long long vec_reve (vector signed long long);
27337 vector unsigned long long vec_reve (vector unsigned long long);
27338 vector bool short vec_reve (vector bool short);
27339 vector signed short vec_reve (vector signed short);
27340 vector double vec_reve (vector double);
27341 vector float vec_reve (vector float);
27342 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
27343 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
27344 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
27345 (altivec_vreve): New pattern.
27346 * config/rs6000/altivec.h (vec_reve): New define.
27347 * doc/extend.texi (vec_rev): Update the built-in documentation file
27348 for the new built-in functions.
27350 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27352 PR tree-optimization/71815
27353 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
27355 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
27357 (slsr_process_phi): Likewise.
27358 (replace_uncond_cands_and_profitable_phis): Don't replace a
27359 multiply candidate with a stride of 1 (copy or cast).
27360 (phi_incr_cost): Call uses_consumed_by_stmt rather than
27362 (lowest_cost_path): Likewise.
27363 (total_savings): Likewise.
27365 2017-06-26 Richard Biener <rguenther@suse.de>
27368 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
27369 Use def_builtin_pure for all gather builtins.
27371 2017-06-26 Richard Biener <rguenther@suse.de>
27373 PR tree-optimization/81203
27374 * tree-tailcall.c (find_tail_calls): Do not move stmts into
27375 non-dominating BBs.
27377 2017-06-26 Marek Polacek <polacek@redhat.com>
27380 * doc/invoke.texi: Document -Wmultistatement-macros.
27382 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
27384 * doc/sourcebuild.texi (ARM-specific attributes): Document new
27385 arm_neon_ok_no_float_abi effective target.
27387 2017-06-26 Richard Biener <rguenther@suse.de>
27389 PR tree-optimization/80928
27390 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
27391 (copy_bbs): Set BB_DUPLICATED flag early.
27392 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
27394 (execute_on_shrinking_pred): Likewise.
27395 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
27396 BB_DUPLICATED blocks.
27397 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
27398 iterate over all PHIs considering removal of *gsi.
27400 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
27402 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
27405 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27407 * config/rs6000/rs6000-string.c: (expand_block_clear,
27408 do_load_for_compare, select_block_compare_mode,
27409 compute_current_alignment, expand_block_compare,
27410 expand_strncmp_align_check, expand_strn_compare,
27411 expand_block_move, rs6000_output_load_multiple)
27412 Move functions related to string/block move/compare
27413 to a separate file.
27414 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
27415 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
27416 for this function which is now used in two files.
27417 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
27418 * config.gcc: Add rs6000-string.o to extra_objs for
27419 targets powerpc*-*-* and rs6000*-*-*.
27421 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
27424 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
27425 32-bit, since indexed is not valid for DImode.
27426 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
27427 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
27428 (define_peephole2 for Altivec d-form load): Add 32-bit support.
27429 (define_peephole2 for Altivec d-form store): Likewise.
27432 * multiple_target.c (create_dispatcher_calls): Only create the
27433 dispatcher call if the function is the default clone of a
27434 versioned function.
27436 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
27438 PR middle-end/80902
27439 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
27440 a call, force the call to not be a tail call.
27442 2017-06-23 Jeff Law <law@redhat.com>
27444 * doc/contrib.texi: Add entry for Steven Pemberton's work on
27447 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
27449 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
27450 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
27451 handling for early expansion of vector shifts (sl,sr,sra,rl).
27452 (builtin_function_type): Add vector shift right instructions
27453 to the unsigned argument list.
27455 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
27457 rtl-optimizatoin/79286
27458 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
27459 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
27460 trap. PIC register plus a const unspec without offset can never trap.
27462 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
27464 * tree.h (builtin_structptr_type): New type.
27465 (builtin_structptr_types): Declare new array.
27466 * tree.c (builtin_structptr_types): New array.
27467 (free_lang_data, build_common_tree_nodes): Use it.
27469 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
27472 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
27475 2017-06-22 Matt Turner <mattst88@gmail.com>
27477 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
27478 Lake models to skylake case. Assume skylake for unknown
27479 models with clflushopt.
27481 2017-06-22 Jeff Law <law@redhat.com>
27483 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
27484 frame sizes that do not satisfy aarch64_uimm12_shift.
27486 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
27488 * profile-count.h (apply_probability,
27489 apply_scale, probability_in): Fix checks for zero.
27491 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27493 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
27494 * doc/cppdiropts.texi (-I @var{dir}): Document it.
27496 2016-06-22 Richard Biener <rguenther@suse.de>
27498 * tree-vect-loop.c (vect_model_reduction_cost): Handle
27499 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
27500 REDUC_MAX_EXPR support.
27501 (vectorizable_reduction): Likewise.
27502 (vect_create_epilog_for_reduction): Likewise.
27504 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
27506 * match.pd (A / (1 << B) -> A >> B): New.
27507 * generic-match-head.c: Include optabs-tree.h.
27508 * gimple-match-head.c: Likewise.
27509 * optabs-tree.h (target_supports_op_p): New.
27510 * optabs-tree.c (target_supports_op_p): New.
27512 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27514 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
27515 $gcc_cv_ld --help output.
27516 (gcc_cv_ld_demangle): Likewise.
27517 (gcc_cv_ld_eh_frame_hdr): Likewise.
27518 (gcc_cv_ld_pie): Likewise.
27519 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
27520 (gcc_cv_ld_buildid): Likewise.
27521 (gcc_cv_ld_sysroot): Likewise.
27522 (ld_bndplt_support): Likewise.
27523 (ld_pushpopstate_support): Likewise.
27524 * configure: Regenerate.
27525 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
27527 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27530 * config/i386/sse.md (round<mode>2): Renumber match_dup and
27531 operands indexes to avoid gap between operands and match_dups.
27533 2017-06-21 Andrew Pinski <apinski@cavium.com>
27535 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
27536 Increment Arith_shift and Arith_shift_reg by 1.
27537 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
27539 * config/aarch64/aarch64.c (thunderx_tunings): Enable
27540 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
27541 (aarch64_strip_extend): Add new argument and test for it.
27542 (aarch64_cheap_mult_shift_p): New function.
27543 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
27544 add a cost if it is true.
27545 Update calls to aarch64_strip_extend.
27546 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
27548 2017-06-21 Andrew Pinski <apinski@cavium.com>
27550 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
27552 (thunderxt88): Likewise.
27553 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
27554 (thunderx_prefetch_tune): New variable.
27555 (thunderx2t99_prefetch_tune): Update for the correct values.
27556 (thunderxt88_tunings): New variable.
27557 (thunderx_tunings): Use thunderx_prefetch_tune instead of
27558 generic_prefetch_tune.
27559 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
27561 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27563 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
27564 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
27565 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
27566 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
27567 (aarch64_atomic_cas<mode>, GPI): Likewise.
27569 2017-06-21 Martin Liska <mliska@suse.cz>
27571 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
27572 statements on cold and hot labels.
27573 * predict.c (tree_estimate_probability_bb): Remove the
27574 prediction from this place.
27576 2017-06-21 Martin Liska <mliska@suse.cz>
27578 PR tree-optimization/79489
27579 * gimplify.c (maybe_add_early_return_predict_stmt): New
27581 (gimplify_return_expr): Call the function.
27582 * predict.c (tree_estimate_probability_bb): Remove handling
27584 * predict.def: Update comment about early return predictor.
27585 * gimple-predict.h (is_gimple_predict): New function.
27586 * predict.def: Change default value of early return to 66.
27587 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
27589 * passes.def: Put pass_strip_predict_hints to the beginning of
27592 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
27594 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
27595 FUNCTION_DECL declarations.
27596 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
27598 (dwaf2out_decl): Likewise.
27599 * godump.c (go_early_global_decl): Skip call to the real debug hook
27600 for FUNCTION_DECL declarations.
27601 * passes.c (rest_of_decl_compilation): Skip call to the
27602 early_global_decl debug hook for FUNCTION_DECL declarations, unless
27603 -fdump-go-spec is passed.
27605 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27607 * config/i386/i386.c (struct builtin_isa): New field pure_p.
27608 Reorder for compactness.
27609 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
27610 (def_builtin_pure, def_builtin_pure2): New functions.
27611 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
27613 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27615 * match.pd (nop_convert): New predicate.
27616 ((A +- CST1) +- CST2): Allow some NOP conversions.
27618 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27621 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
27622 with ctors/dtors if GOVD_SHARED is set.
27624 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27626 * config/aarch64/aarch64.md (movti_aarch64):
27627 Emit mov rather than orr.
27628 (movtf_aarch64): Likewise.
27629 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
27630 Emit mov rather than orr.
27632 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27634 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
27635 Swap alternatives, make integer dup more expensive.
27637 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27639 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
27640 Return true for non-tls symbols.
27642 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
27644 * config/aarch64/aarch64-cores.def (cortex-a55): New.
27645 (cortex-a75): Likewise.
27646 (cortex-a75.cortex-a55): Likewise.
27647 * config/aarch64/aarch64-tune.md: Regenerate.
27648 * doc/invoke.texi (-mtune): Document new values for -mtune.
27650 2017-06-21 Tom de Vries <tom@codesourcery.com>
27652 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
27653 stack_size feature.
27654 (Effective-Target Keywords, Other attributes): Suggest using
27655 dg-add-options stack_size feature to get stack limit in stack_size
27656 effective target documentation.
27658 2017-06-21 Julian Brown <julian@codesourcery.com>
27659 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27661 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
27662 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
27663 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
27665 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
27666 attribute type list for neon_multiply.
27667 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
27668 attribute type list for neon_multiply.
27669 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
27670 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
27671 attribute type list for neon_multiply.
27672 * config/arm/types.md (crypto_pmull): Add.
27673 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
27674 attribute type list.
27676 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
27678 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
27681 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27683 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
27684 to make sure not to dereference a NULL cost_classes_ptr pointer.
27686 2017-06-20 Carl Love <cel@us.ibm.com>
27688 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27689 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
27690 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
27691 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
27692 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
27693 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
27694 VMULOSW): New enum "unspec" values.
27695 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
27696 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
27697 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
27698 altivec_vmulosw): New patterns.
27699 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
27700 VMULOSW): Add definitions.
27702 2017-06-20 Julia Koval <julia.koval@intel.com>
27704 * config/i386/i386.c: Fix rounding expand for new pattern.
27705 * config/i386/subst.md: Fix pattern (parallel -> unspec).
27707 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27709 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
27710 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
27712 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27714 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
27717 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27719 * config/aarch64/aarch64-cores.def: Rearrange to sort by
27720 architecture, then by implementer ID.
27721 * config/aarch64/aarch64-tune.md: Regenerate.
27723 2017-06-20 Richard Biener <rguenther@suse.de>
27725 PR middle-end/81097
27726 * fold-const.c (split_tree): Fold to type before negating.
27728 2017-06-20 David Malcolm <dmalcolm@redhat.com>
27730 * diagnostic-show-locus.c
27731 (selftest::test_fixit_deletion_affecting_newline): New function.
27732 (selftest::diagnostic_show_locus_c_tests): Call it.
27734 2017-06-20 Andreas Schwab <schwab@suse.de>
27737 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
27740 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27742 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
27743 __ARM_FEATURE_COPROC according to support.
27745 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27747 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27748 Rewritten to avoid overflow for > 32-bit pointers.
27751 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
27752 by removing enum keyword.
27753 (ubsan_type_descriptor): Likewise. Formatting fix.
27756 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
27757 splitter): Require TARGET_SSE2 in the condition.
27759 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
27762 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
27763 for doing vector set of SFmode on ISA 3.0.
27764 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
27765 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
27767 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
27768 SFmode value into a V4SF variable that was extracted from another
27769 V4SF variable without converting the element to double precision
27770 and back to single precision vector format.
27771 (vsx_insert_extract_v4sf_p9_2): Likewise.
27773 2017-06-19 Jakub Jelinek <jakub@redhat.com>
27775 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
27776 in UWHI to avoid undefined overflow.
27779 * ubsan.h (enum ubsan_encode_value_phase): New.
27780 (ubsan_encode_value): Change second argument to
27781 enum ubsan_encode_value_phase with default value of
27782 UBSAN_ENCODE_VALUE_GENERIC.
27783 * ubsan.c (ubsan_encode_value): Change second argument to
27784 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
27785 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
27786 create_tmp_var_raw instead of create_tmp_var and use a
27788 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
27789 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27790 ubsan_encode_value callers.
27793 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27794 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27795 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27797 2017-06-19 Richard Biener <rguenther@suse.de>
27799 PR middle-end/81118
27800 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27801 estimates if we changed anything.
27803 2017-06-19 Richard Biener <rguenther@suse.de>
27805 PR tree-optimization/80887
27806 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27807 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27808 simplified lookups, then reset mprts_hook.
27809 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27811 (try_to_simplify): Likewise.
27813 2017-06-19 Martin Liska <mliska@suse.cz>
27816 * gimplify.c (gimplify_switch_expr):
27817 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27819 2017-06-19 Martin Liska <mliska@suse.cz>
27821 * doc/install.texi: Document that PGO runs in 4 stages.
27823 2017-06-19 Martin Liska <mliska@suse.cz>
27826 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27827 to dispatcher function name.
27828 * multiple_target.c (replace_function_decl): New function.
27829 (create_dispatcher_calls): Redirect both edges and references.
27831 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27833 * profile-count.c (profile_count::dump): Dump quality.
27834 (profile_count::differs_from_p): Update for unsigned val.
27835 * profile-count.h (profile_count_quality): New enum.
27836 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27838 2017-06-19 Richard Biener <rguenther@suse.de>
27840 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27841 struct function as arg.
27842 (estimate_numbers_of_iterations): Export overload with loop arg.
27843 (free_numbers_of_iterations_estimates_loop): Use an overload of
27844 free_numbers_of_iterations_estimates instead.
27845 * tree-cfg.c (remove_bb): Adjust.
27846 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27847 * tree-parloops.c (gen_parallel_loop): Likewise.
27848 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27850 (tree_unroll_loops_completely): Likewise.
27851 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27852 Use an overload instead and export.
27853 (estimated_loop_iterations): Adjust.
27854 (max_loop_iterations): Likewise.
27855 (likely_max_loop_iterations): Likewise.
27856 (estimate_numbers_of_iterations): Take struct function as arg
27858 (loop_exits_before_overflow): Adjust.
27859 (free_numbers_of_iterations_estimates_loop): Use an overload.
27860 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27861 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27863 2017-06-19 Richard Biener <rguenther@suse.de>
27866 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27867 RANGE_EXPR conservatively.
27869 2017-06-16 Carl Love <cel@us.ibm.com>
27871 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27872 definitions for vec_float, vec_float2, vec_floato,
27873 vec_floate built-ins.
27874 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27875 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27877 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27878 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27879 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27880 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27881 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27882 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27883 vec_floato): Add builtin defines.
27884 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27885 Update the built-in documentation file for the new built-in
27888 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27890 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27891 (mthumb): Mark as the negative of -marm.
27893 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27895 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27897 (ARM Options, -mtune): Document that this accepts the same
27898 extension options as -mcpu.
27899 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27901 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27903 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27904 permitted extensions.
27906 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27908 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27909 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27910 (armv8-m.main): Add option +nodsp.
27911 * config/arm/arm-cpu-cdata.h: Regenerated.
27913 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27915 * config/arm/t-fuchsia: New file.
27916 * config.gcc (arm*-*-fuchsia*): Use it.
27918 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27920 * config/arm/t-symbian: Rewrite for new option infrastructure.
27922 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27924 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27925 (MULTILIB_REQUIRED): Likewise.
27927 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27929 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27930 (MULTILIB_RESUE): Likewise.
27931 (MULTILIB_MATCHES): Likewise.
27932 (MULTLIB_REQUIRED): Likewise.
27934 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27936 * config/arm/t-rtems: Rewrite for new option framework.
27938 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27940 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27941 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27942 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27943 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27944 * config/arm/t-multilib: ... here.
27945 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27946 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27947 armv7-a and armv8*-a when A-profile libraries have not been built.
27948 * config/arm/t-rmprofile: Rewrite.
27950 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27952 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27953 with a backslash. Remove the backslash after substituting unescaped
27955 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27957 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27959 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27960 the driver to pass through the default setting of -mfloat-abi.
27961 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27963 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27964 (all_feat_combs): New rule.
27965 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27967 * config/arm/t-aprofile: Rewrite.
27969 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27971 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27972 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27973 fpu is not specified by the user/command-line.
27974 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27975 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27976 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27977 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27978 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27979 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27981 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27983 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27984 * config/arm/t-arm-elf: Rewritten.
27986 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27988 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27989 have some floating-point instructions.
27990 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27991 (TARGET_MAYBE_HARD_FLOAT): New macro.
27992 * config/arm/arm-builtins.c (arm_init_builtins): Use
27993 TARGET_MAYBE_HARD_FLOAT.
27994 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27996 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27998 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27999 (configargs.h): Include it.
28000 (arm_print_hint_for_fpu_option): New function.
28001 (arm_parse_fpu_option): New function.
28002 (candidate_extension): New class.
28003 (arm_canon_for_multilib): New function.
28004 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
28005 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
28006 (ARCH_CANONICAL_SPECS): New macro.
28007 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
28009 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28011 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
28012 are set after handling multilib fragments. Set target_cpu_default2
28015 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28017 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
28019 (arm*-*-*): Set target_cpu_default2 to a quoted string.
28020 * config/arm/parsecpu.awk (check_cpu): Validate any extension
28022 (check_arch): Likewise.
28023 * config/arm/arm.c (arm_configure_build_target): Handle
28024 TARGET_CPU_DEFAULT being a string constant. Scan any feature
28025 options in the default.
28027 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28029 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
28030 when an option is an alias of another.
28031 * config/arm/parsecpu.awk (optalias): New parser token.
28032 (gen_comm_data): Mark non-alias options as such. Emit entries
28033 for extension aliases.
28034 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
28035 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
28036 (armv6kz, armv6zk, armv6t2): Likewise.
28037 (armv7): Make vfpv3-d16 an alias.
28038 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
28040 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
28041 Sort in canonical order.
28042 (armv8-a): Sort in canonical order.
28043 (armv8.1-a, armv8.2-a): Likewise.
28044 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
28046 (cortex-a9): Sort in canonical order.
28047 * config/arm/arm.c (selftests.h): Include it.
28048 (arm_test_cpu_arch_data): New function.
28049 (arm_run_self_tests): New function.
28050 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
28051 (targetm): Move declaration to the end of the file.
28052 * arm-cpu-cdata.h: Regenerated.
28054 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28056 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
28057 call to target_mode_check describing the type of option passed.
28058 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
28059 (arm_target_thumb_only): Use arm_parse_arch_option_name or
28060 arm_parse_cpu_option_name to match parameters against list of
28062 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
28063 arm_arch_core_flags data structure.
28064 * config/arm/arm-cpu_cdata.h: Regenerated.
28066 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28068 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
28070 (arm_print_hint_for_cpu_option): Likewise.
28071 (arm_print_hint_for_arch_option): Likewise.
28072 (arm_parse_cpu_option_name): Likewise.
28073 (arm_parse_arch_option_name): Likewise.
28074 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
28075 of entries in the all_fpus list.
28076 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
28077 (arm_parse_cpu_option_name): Declare.
28078 (arm_parse_arch_option_name): Declare.
28079 (arm_parse_option_features): Declare.
28080 (arm_intialize_isa): Declare.
28081 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
28083 (gen_comm_data): ... here. Make definitions non-static.
28084 * config/arm/arm-cpu-data.h: Regenerated.
28085 * config/arm/arm-cpu-cdata.h: Regenerated.
28087 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28089 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
28090 (cpu_arch_extension): New structure.
28091 (cpu_arch_option, arch_option, cpu_option): New structures.
28092 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
28093 architecture types.
28094 (gen_data): Generate new format data tables.
28095 * config/arm/arm.c (cpu_tune): New structure.
28096 (cpu_option, processors): Delete.
28097 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
28098 (arm_print_hint_for_cpu_option): ... this and ...
28099 (arm_print_hint_for_arch_option): ... this.
28100 (arm_parse_arch_cpu_name): Delete. Replace with ...
28101 (arm_parse_cpu_option_name): ... this and ...
28102 (arm_parse_arch_option_name): ... this.
28103 (arm_unrecognized_feature): Change type of target parameter to
28105 (arm_parse_arch_cpu_features): Delete. Replace with ...
28106 (arm_parse_option_features): ... this.
28107 (arm_configure_build_target): Rework to use new configuration data
28109 (arm_print_tune_info): Rework for new configuration data tables.
28110 * config/arm/arm-cpu-data.h: Regenerated.
28111 * config/arm/arm-cpu.h: Regenerated.
28113 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28115 * Makefile.in (OBJS): Move sbitmap.o from here ...
28116 (OBJS-libcommon): ... to here.
28118 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28120 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
28121 (ISA_ALL_CRYPTO): New macro.
28122 (ISA_ALL_SIMD): New macro
28123 (ISA_ALL_FP): New macro.
28124 * config/arm/arm.c (fpu_bitlist): Update initializer.
28125 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
28127 (arm9e): Add fpu. Add option for nofp
28128 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
28129 (arm926ej-s, arm1026ej-s): Likewise.
28130 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
28131 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
28132 neon-fp16, neon-vfpv4, nofp and nosimd.
28133 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
28134 (cortex-a8): Add fpu. Add option for nofp.
28135 (cortex-a9): Add fpu. Add options for nosimd and nofp.
28136 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
28137 (cortex-r4f): Add fpu.
28138 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
28139 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
28141 (cortex-r8): Likewise.
28142 (cortex-m4): Add fpu. Add option for nofp.
28143 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
28144 (cortex-a17.cortex-a7): Likewise.
28145 (cortex-a32): Add fpu. Add options for crypto and nofp.
28146 (cortex-a35, cortex-a53): Likewise.
28147 (cortex-a57): Add fpu. Add option for crypto.
28148 (cortex-a72, cortex-a73): Likewise.
28149 (exynos-m1): Likewise.
28150 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
28151 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
28152 (cortex-m33): Add fpu. Add option for nofp.
28153 * config/arm/arm-cpu-cdata.h: Regenerated
28154 * config/arm/arm-cpu-data.h: Regenerated.
28156 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28158 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
28159 (armv5te, armv5tej): Likewise.
28160 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
28161 (armv7): Add options fp and vfpv3-d16.
28162 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
28163 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
28165 (armv7ve): Likewise.
28166 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
28167 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
28168 (armv8-a): Add nocrypto option.
28169 (armv8.1-a, armv8.2-a): Likewise.
28170 (armv8-m.main): add options fp, fp.dp and nofp.
28172 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28174 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
28176 (armv8-a+crc): Delete.
28177 (armv8.1-a): Add options simd, crypto and nofp.
28178 (armv8.2-a): Add options fp16, simd, crypto and nofp.
28179 (armv8.2-a+fp16): Delete.
28180 (armv8-m.main): Add option dsp.
28181 (armv8-m.main+dsp): Delete.
28182 (cortex-a8): Add fpu. Add nofp option.
28183 (cortex-a9): Add fpu. Add nofp and nosimd options.
28184 * config/arm/parsecpu.awk (gen_data): Generate option tables and
28185 link to main cpu and architecture data structures.
28186 (gen_comm_data): Only put isa attributes from the main architecture
28188 (option): New statement for architecture and CPU entries.
28189 * arm.c (struct cpu_option): New structure.
28190 (struct processors): Add entry for options.
28191 (arm_unrecognized_feature): New function.
28192 (arm_parse_arch_cpu_name): Ignore any characters after the first
28194 (arm_parse_arch_cpu_feature): New function.
28195 (arm_configure_build_target): Separate out any CPU and architecture
28196 features and parse separately. Don't error out if -mfpu=auto is
28197 used with only an architecture string.
28198 (arm_print_asm_arch_directives): New function.
28199 (arm_file_start): Call it.
28200 * config/arm/arm-cpu-cdata.h: Regenerated.
28201 * config/arm/arm-cpu-data.h: Likewise.
28202 * config/arm/arm-tables.opt: Likewise.
28204 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28206 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
28207 assembler when it is not -mfpu=auto.
28209 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28211 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
28212 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
28213 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
28214 (ASM_CPU_SPEC): Rewrite.
28215 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
28216 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
28217 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
28218 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
28219 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
28220 copied string is NUL-terminated. Also strip any characters prefixed
28222 (arm_rewrite_selected_arch): New function.
28223 (arm_rewrite_march): New function.
28225 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28227 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
28228 (x_arm_cpu_string, x_arm_tune_string): Likewise.
28229 (march, mcpu, mtune): Convert to string-based options.
28230 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
28231 (arm_parse_arch_cpu_name): New function.
28232 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
28233 identify selected architecture or CPU.
28234 (arm_option_save): New function.
28235 (TARGET_OPTION_SAVE): Redefine.
28236 (arm_option_restore): Restore string options.
28237 (arm_option_print): Print string options.
28239 2017-06-16 Martin Sebor <msebor@redhat.com>
28241 PR tree-optimization/80933
28242 PR tree-optimization/80934
28243 * builtins.c (fold_builtin_3): Do not handle bcmp here.
28244 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
28245 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
28246 (gimple_fold_builtin): Call them.
28248 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28250 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
28251 as unlikely; update profile.
28253 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28255 * predict.c (force_edge_cold): Handle declaring edges impossible
28258 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28260 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
28262 (try_unroll_loop_completely): Fix reporting.
28264 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28266 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
28268 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
28271 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
28272 if given a non-constant argument for an intrinsic which requires a
28275 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28277 * profile.c (compare_freqs): New function.
28278 (branch_prob): Sort edge list.
28279 (find_spanning_tree): Assume that the list is priority sorted.
28281 2017-06-16 Richard Biener <rguenther@suse.de>
28283 PR tree-optimization/81090
28284 * passes.def (pass_record_bounds): Remove.
28285 * tree-pass.h (make_pass_record_bounds): Likewise.
28286 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
28287 make_pass_record_bounds): Likewise.
28288 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
28289 not free niter estimates at the beginning but at the end.
28290 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
28292 2017-06-16 Richard Biener <rguenther@suse.de>
28294 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
28295 initializer to workaround ICE in host GCC 4.8.
28297 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28299 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
28301 (clone_inlined_nodes): Update.
28303 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28305 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
28306 prefetch settings, and enable prefetching by default at -O3.
28308 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28310 * config/aarch64/aarch64.c (aarch64_override_options_internal):
28311 Set flag_prefetch_loop_arrays according to tuning data.
28313 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28315 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
28316 New tune structure.
28317 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
28318 [Unrelated to main purpose of the patch] Place the pointer field last
28319 to enable type checking errors when tune structure are wrongly merged.
28320 * config/aarch64/aarch64.c (generic_prefetch_tune,)
28321 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
28322 (thunderx2t99_prefetch_tune): New tune constants.
28323 (tune_params *_tunings): Update all tunings (no functional change).
28324 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
28325 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
28326 from tunings structures.
28328 2017-06-16 Jakub Jelinek <jakub@redhat.com>
28331 * ubsan.c (instrument_null): Add T argument, use it instead
28332 of computing it based on IS_LHS.
28333 (instrument_object_size): Likewise.
28334 (pass_ubsan::execute): Adjust instrument_null and
28335 instrument_object_size callers to pass gimple_get_lhs or
28336 gimple_assign_rhs1 result to it. Use instrument_null instead of
28337 calling get_base_address and instrument_mem_ref. Handle
28338 aggregate call arguments for object-size sanitization.
28340 2017-06-16 Yury Gribov <tetra2005@gmail.com>
28342 PR tree-optimization/81089
28343 * tree-vrp.c (is_masked_range_test): Validate operands of
28346 2017-06-15 Martin Sebor <msebor@redhat.com>
28349 * dumpfile.c (dump_register): Avoid calling memset to initialize
28350 a class with a default ctor.
28351 * gcc.c (struct compiler): Remove const qualification.
28352 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
28353 * hash-table.h: Ditto.
28354 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
28356 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
28357 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
28359 * params.h (struct param_info): Make struct members non-const.
28360 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
28361 with default initialization.
28362 * vec.h (vec_copy_construct, vec_default_construct): New helper
28364 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
28365 with vec_copy_construct.
28366 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
28367 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
28368 * doc/invoke.texi (-Wclass-memaccess): Document.
28370 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28372 * emit-rtl.h (is_leaf): Update comment about local
28373 register allocator.
28375 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28378 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
28379 for a variable to have a section before checking if the section has a
28381 Set section to.persistent if persistent attribute is set.
28382 Warn if .persistent attribute is used on an automatic variable.
28384 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
28386 PR rtl-optimization/80474
28387 * reorg.c (update_block): Do not ignore instructions in a delay slot.
28389 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
28391 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
28394 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
28396 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
28397 (casesi): Emit bounds checking as RTL.
28398 (casesi_internal_mips16_<mode>): Remove bounds checking.
28400 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
28402 * config/xtensa/xtensa.c (xtensa_option_override): Append
28403 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
28404 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
28405 xtensa_doloop_hooks): Define unconditionally.
28406 (xtensa_reorg_loops): Only call reorg_loops in the presence of
28408 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
28409 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
28410 for it in xtensa_option_override.
28411 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
28412 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
28414 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
28416 * doc/cppopts.texi: Document '-' special value to -MF.
28418 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
28420 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
28421 (cortex_a53_fconst): Likewise.
28422 (cortex_a53_fpmul): Likewise.
28423 (cortex_a53_f_load_64): Likewise.
28424 (cortex_a53_f_load_many): Likewise.
28425 (cortex_a53_advsimd_alu): Likewise.
28426 (cortex_a53_advsimd_alu_q): Likewise.
28427 (cortex_a53_advsimd_mul): Likewise.
28428 (cortex_a53_advsimd_mul_q): Likewise.
28429 (fpmac bypass): Add new bypass for fpmac-fpmac case.
28430 Add missing fmul, r2f_cvt and fconst cases.
28432 2017-06-14 Richard Biener <rguenther@suse.de>
28434 PR middle-end/81088
28435 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
28437 (fold_binary_loc): When associating do not treat pre-existing
28438 TREE_OVERFLOW on literal constants as a reason to allow
28439 TREE_OVERFLOW on associated literal constants.
28441 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28443 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
28444 (MASK_FEATURES): New macro.
28445 * config/sparc/sparc.c (sparc_option_override): Remove the special
28446 handling of -mfpu and generalize it to all MASK_FEATURES switches.
28448 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28450 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
28451 a division of 0 if non-call exceptions are enabled.
28453 2017-06-14 Andrew Pinski <apinski@cavium.com>
28454 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28457 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
28458 Improve vector initialization code gen for only variable case.
28460 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28462 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
28464 2017-06-14 Richard Biener <rguenther@suse.de>
28466 PR tree-optimization/81083
28467 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
28470 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28472 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
28473 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
28474 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
28475 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
28476 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
28477 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
28479 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28481 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
28482 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
28484 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28486 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
28488 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28490 * config/rs6000/t-rtems: Don't handle SPE.
28492 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28494 * config/rs6000/t-linux: Don't handle SPE.
28496 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28498 * config/rs6000/eabispe.h: Delete file.
28500 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28502 * config/rs6000/t-spe: Delete file.
28504 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28506 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
28507 (rs6000_legitimate_offset_address_p): Return false for anything in
28508 V2SImode or V2SFmode.
28510 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28512 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
28513 except V2SF and V2SI. Rearrange the vector modes, and add comments.
28514 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
28516 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
28517 (rs6000_legitimate_offset_address_p): Ditto.
28518 (rs6000_emit_move): Ditto.
28519 (rs6000_init_builtins): Remove V4HI_type_node.
28521 2017-06-13 Martin Liska <mliska@suse.cz>
28524 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
28525 (gate_asan): Likewise.
28526 * asan.h (asan_no_sanitize_address_p): Remove the function.
28527 (sanitize_flags_p): New function.
28528 * builtins.def: Fix coding style.
28529 * common.opt: Use renamed enum value.
28530 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
28531 * doc/extend.texi: Document no_sanitize attribute.
28532 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
28533 to SANITIZE_UNDEFINED_NONDEFAULT.
28534 * gcc.c (sanitize_spec_function): Use the renamed enum value.
28535 * gimple-fold.c (optimize_atomic_compare_exchange_p):
28536 Use sanitize_flags_p.
28537 * gimplify.c (gimplify_function_tree): Likewise.
28538 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
28539 * opts.c (parse_no_sanitize_attribute): New function.
28540 (common_handle_option): Use renamed enum value.
28541 * opts.h (parse_no_sanitize_attribute): Declare.
28542 * tree.c (sanitize_flags_p): New function.
28543 * tree.h: Declared here.
28544 * tsan.c: Use sanitize_flags_p.
28545 * ubsan.c (ubsan_expand_null_ifn): Likewise.
28546 (instrument_mem_ref): Likewise.
28547 (instrument_bool_enum_load): Likewise.
28548 (do_ubsan_in_current_function): Remove the function.
28549 (pass_ubsan::execute): Use sanitize_flags_p.
28550 * ubsan.h: Remove do_ubsan_in_current_function
28551 * tree-cfg.c (print_no_sanitize_attr_value): New function.
28552 (dump_function_to_file): Use it here.
28554 2017-06-13 Martin Jambor <mjambor@suse.cz>
28556 PR tree-optimization/80803
28557 PR tree-optimization/81063
28558 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
28559 (propagate_subaccesses_across_link): Enqueue subtree whenever
28560 necessary instead of relying on the caller.
28562 2017-06-13 Martin Jambor <mjambor@suse.cz>
28564 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
28565 that have a first_link.
28566 (sort_and_splice_var_accesses): Do not check first_link before
28568 (subtree_mark_written_and_enqueue): Likewise.
28569 (propagate_all_subaccesses): Likewise and do not stop at first
28570 parent with a first_link.
28572 2017-06-13 Martin Jambor <mjambor@suse.cz>
28574 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
28577 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28579 * match.pd: New pattern.
28581 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28583 * tree-vrp.c (is_masked_range_test): New function.
28584 (register_edge_assert_for): Determine ranges for
28587 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28589 PR tree-optimization/67328
28590 * fold-const.c (maskable_range_p): New function.
28591 (build_range_check): Generate bittests if possible.
28593 2017-06-13 Martin Liska <mliska@suse.cz>
28595 * gimple-pretty-print.c (dump_probability): Add new argument.
28596 (dump_edge_probability): Dump both probability and count.
28597 (dump_gimple_label): Likewise.
28598 (dump_gimple_bb_header): Likewise.
28600 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
28603 * config/avr/avr-devices.c: Fix indentation.
28604 * config/avr/gen-avr-mmcu-specs.c: Dito.
28606 2017-06-13 Richard Biener <rguenther@suse.de>
28608 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
28609 instead get vector type from stmt_info.
28610 (vectorizable_reduction): Adjust. Remove dead code.
28612 2017-06-13 Richard Biener <rguenther@suse.de>
28614 PR middle-end/81065
28615 * fold-const.c (extract_muldiv_1): Remove bogus distribution
28616 case of C * (x * C2 + C3).
28617 (fold_addr_of_array_ref_difference): Properly fold index difference.
28619 2017-06-12 David S. Miller <davem@davemloft.net>
28622 * config/sparc/sparc.md (return expander): Emit frame blockage if
28623 function uses alloca.
28625 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
28627 * combine.c (make_field_assignment): Check len rather than the mode
28628 precision when calling force_to_mode.
28630 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
28632 Support multilibs and devices that see flash in RAM address range.
28635 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
28636 (avr_mcu_t) <flash_pm_offset>: New field.
28637 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
28638 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
28639 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
28640 (AVR_TINY_PM_OFFSET): Remove macro.
28641 * config/avr/avr.opt (-mshort-calls): New option.
28642 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
28643 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
28644 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
28645 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
28646 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
28647 instead of avr_arch->have_jmp_call.
28648 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
28649 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
28650 avr_arch->flash_pm_offset to define.
28651 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
28652 new field flash_pm_offset. Add entry for avrxmega3.
28653 (avr_texinfo): Add entry for avrxmega3.
28654 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
28655 attiny212, attiny214,
28656 attiny412, attiny414, attiny416, attiny417,
28657 attiny814, attiny816, attiny817,
28658 attiny1614, attiny1616, attiny1617,
28659 attiny3214, attiny3216, attiny3217.
28660 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
28661 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
28662 (avr_print_operand_address) [AVR_TINY]: Same.
28663 (avr_asm_init_sections) <readonly_data_section>: Only patch
28664 callback if avr_arch->flash_pm_offset = 0.
28665 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
28666 for rodata if avr_arch->flash_pm_offset != 0.
28667 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
28668 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
28669 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
28670 (m_options): Append opt_rcall.
28671 (m_dirnames): Append dir_rcall.
28672 * config/avr/t-multilib: Regenerate.
28674 * configure.ac [target=avr]: Check whether avrxmega3 default
28675 linker description file works as needed.
28676 * configure: Regenerate.
28677 * doc/avr-mmcu.texi: Regenerate.
28678 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
28679 <__AVR_ARCH__>: Document avrxmega3 and 103.
28680 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
28681 <__AVR_SHORT_CALLS__>: Document it.
28682 <__AVR_PM_BASE_ADDRESS__>: Document it.
28683 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
28684 (AVR Variable Attributes) <progmem>: Document this is
28685 not needed for avrxmega3.
28686 (AVR Named Address Spaces) <__flash>: Dito.
28688 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
28690 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
28692 2017-06-12 Doug Rupp <rupp@adacore.com>
28694 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
28695 Append vxworks-stdint.h to the tm_file list.
28696 * config/vxworks-stdint.h: New file.
28698 2017-06-12 Martin Liska <mliska@suse.cz>
28700 PR tree-optimization/81041
28701 * tree-profile.c (gimple_gen_ic_func_profiler):
28702 Create an extra BB in profile-generate
28703 (gimple_gen_time_profiler): Likewise.
28705 2017-06-12 Jakub Jelinek <jakub@redhat.com>
28707 PR tree-optimization/81003
28708 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
28709 (update_range_test): Use it instead of force_gimple_operand_gsi.
28711 2017-06-12 Richard Biener <rguenther@suse.de>
28713 PR tree-optimization/81053
28714 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
28715 with backedge value not defined in loop. Simplify def stmt
28718 2017-06-11 Tom de Vries <tom@codesourcery.com>
28721 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
28723 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
28724 nvptx_cannot_force_const_mem.
28726 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28728 * opts.c (finish_options): Move test for flag_split_stack after
28729 it has been initialized.
28731 2017-06-11 Jason Merrill <jason@redhat.com>
28733 * tree.h (id_equal): New.
28734 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
28735 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
28736 instead of strcmp of IDENTIFIER_POINTER.
28738 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28740 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
28741 (mark_all_inlined_calls_cdtor): Fix formating.
28742 (inline_transform): Rescale profile before inlining.
28744 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28746 * cgraph.h (cgraph_edge::clone): Update prototype.
28747 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
28748 (cgraph_node::create_clone): Update.
28749 (cgraph_node::create_version_clone): Update.
28750 * tree-inline.c (copy_bb): Update.
28751 (expand_call_inline): Update.
28753 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28755 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
28756 factored out from ...
28757 (rs6000_emit_prologue): ... here.
28759 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28761 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
28762 factored out from ...
28763 (rs6000_emit_prologue): ... here.
28765 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28767 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
28769 (handle_missing_profiles): Fix computation of tp_first_run.
28770 (counts_to_freqs): Do not touch freqs when count is 0.
28772 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28774 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
28777 2017-06-10 Tom de Vries <tom@codesourcery.com>
28779 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28780 attributes): Document signal effective target.
28782 2017-06-10 Tom de Vries <tom@codesourcery.com>
28784 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28785 Document effective target stack_size.
28787 2017-06-09 David Malcolm <dmalcolm@redhat.com>
28789 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28790 to the edit_context if they can be auto-applied.
28792 2017-06-9 Ian Lance Taylor <iant@golang.org>
28794 * opts.c (finish_options): If -fsplit-stack, disable implicit
28795 -forder-blocks-and-partition.
28796 * doc/invoke.texi (Optimize Options): Document that when using
28797 -fsplit-stack -forder-blocks-and-partition is not implicitly
28800 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28802 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28803 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28804 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28805 * builtins.def (abort, trap, unreachable): Declare cold.
28806 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28807 * tree-core.h (ECF_COLD): New.
28808 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28809 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28811 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28813 * predict.c (unlikely_executed_stmt_p): Cleanup.
28815 2017-06-09 Richard Biener <rguenther@suse.de>
28817 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28818 model if the ref is always written to.
28820 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28822 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28824 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28826 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28829 2017-06-09 Tom de Vries <tom@codesourcery.com>
28832 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28833 "target cannot support label values" when encountering LABEL_REF.
28835 2017-06-09 Martin Liska <mliska@suse.cz>
28837 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28838 (gimple_gen_ic_func_profiler): Emit direct comparison
28839 of __gcov_indirect_call_callee with NULL.
28840 (gimple_gen_time_profiler): Change probability from
28841 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28843 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28845 * profile.c (edge_gcov_counts): Turn to pointer.
28846 (compute_branch_probabilities, compute_branch_probabilities): Update.
28847 (branch_prob): Do not clear edge_gcov_count.
28848 * profile.h (edge_gcov_counts): Turn to pointer.
28849 (edge_gcov_count): Update.
28851 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28853 * gimple.h (gimple_check_failed): Mark cold.
28855 2017-06-09 Richard Biener <rguenther@suse.de>
28857 PR tree-optimization/66623
28858 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28859 refactor check_reduction into two parts, properly computing
28860 whether we have to check reduction validity for outer loop
28863 2017-06-09 Richard Biener <rguenther@suse.de>
28865 PR tree-optimization/79483
28866 * graphite-scop-detection.c (order): New global.
28867 (get_order): Compute bb to order mapping that satisfies code
28868 generation constraints.
28869 (cmp_pbbs): New helper.
28870 (build_scops): Start domwalk at entry block, sort generated
28873 2017-06-09 Richard Biener <rguenther@suse.de>
28875 PR middle-end/81007
28876 * ipa-polymorphic-call.c
28877 (ipa_polymorphic_call_context::restrict_to_inner_class):
28878 Skip FIELD_DECLs with error_mark_node type.
28879 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28882 2017-06-09 Martin Liska <mliska@suse.cz>
28884 * predict.c (struct branch_predictor): New struct.
28885 (test_prediction_value_range): New test.
28886 (predict_c_tests): New function.
28887 * selftest-run-tests.c (selftest::run_tests): Run the function.
28888 * selftest.h: Declare new tests.
28890 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28893 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28894 gen_add3_insn did not fail.
28895 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28896 r0, construct that number in a temporary reg and add that reg to r0.
28897 If asked to put the result in r0 as well, fail.
28899 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28901 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28902 for early expansion of vec_eqv.
28904 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28906 PR middle-end/81005
28907 * ubsan.c (instrument_null): Avoid pointless code temporary.
28908 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28910 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28914 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28917 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28920 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28922 * predict.c (unlikely_executed_edge_p): Move ahead.
28923 (probably_never_executed_edge_p): Use it.
28925 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28927 PR middle-end/79988
28928 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28929 gimple_call_builtin_p call.
28931 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28933 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28934 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28935 rtl_check_failed_type2, rtl_check_failed_code1,
28936 rtl_check_failed_code2, rtl_check_failed_code_mode,
28937 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28938 rtvec_check_failed_bounds, rtl_check_failed_flag,
28939 _fatal_insn_not_found, _fatal_insn): Likewise.
28940 * tree.h (tree_contains_struct_check_failed,
28941 tree_check_failed, tree_not_check_failed,
28942 tree_class_check_failed, tree_range_check_failed,
28943 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28944 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28945 tree_operand_check_failed, omp_clause_check_failed,
28946 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28949 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28951 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28952 flag_branch_probabilities.
28953 * ipa-inline.c (edge_badness): Likewise.
28954 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28955 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28956 * predict.c (maybe_hot_frequency_p): Likewise.
28957 (probably_never_executed): Likewise.
28958 * sched-ebb.c (schedule_ebbs): Likewise.
28959 * sched-rgn.c (find_single_block_region): Likewise.
28960 * tracer.c (tail_duplicate): Likewise.
28962 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28964 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28965 longer requires x_flag_profile_use.
28967 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28969 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28970 instead of flag_reorder_blocks_and_partition.
28971 * dbxout.c (dbxout_function_end): Likewise.
28972 * dwarf2out.c (gen_subprogram_die): Likewise.
28973 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28974 * hw-doloop.c (reorg_loops): Likewise.
28975 * varasm.c (assemble_start_function,
28976 assemble_end_function): Likewise.
28977 (decide_function_section): Do not check for
28978 flag_reorder_blocks_and_partition.
28980 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28982 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28984 (chkp_get_hard_register_fake_addr_expr): Ditto.
28985 (chkp_build_addr_expr): Add check for hard reg case.
28986 (chkp_parse_array_and_component_ref): Ditto.
28987 (chkp_find_bounds_1): Ditto.
28988 (chkp_process_stmt): Don't generate bounds store for
28991 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28993 * predict.c (maybe_hot_bb_p): Do not check profile status.
28994 (maybe_hot_edge_p): Likewise.
28995 (probably_never_executed): Check for zero counts even if profile
28997 (unlikely_executed_edge_p): New function.
28998 (unlikely_executed_stmt_p): New function.
28999 (unlikely_executed_bb_p): New function.
29000 (set_even_probabilities): Use unlikely predicates.
29001 (combine_predictions_for_bb): Likewise.
29002 (predict_paths_for_bb): Likewise.
29003 (predict_paths_leading_to_edge): Likewise.
29004 (determine_unlikely_bbs): New function.
29005 (estimate_bb_frequencies): Use it.
29006 (compute_function_frequency): Use zero counts even if profile is
29008 * profile-count.h: Fix typo.
29010 2017-08-08 Julia Koval <julia.koval@intel.com>
29012 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
29013 _mm512_mask_cvtsepi16_storeu_epi8,
29014 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
29015 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
29016 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
29017 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
29018 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
29019 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
29020 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
29021 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
29022 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
29023 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
29024 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
29025 __builtin_ia32_pmovuswb256mem_mask,
29026 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
29027 __builtin_ia32_pmovwb512mem_mask): New builtins.
29029 2017-08-08 Julia Koval <julia.koval@intel.com>
29031 PR target/73350,80862
29032 * config/i386/subst.md (round): Fix round pattern.
29033 * config/i386/i386.c (ix86_erase_embedded_rounding):
29034 Fix erasing rounding for the fixed pattern.
29036 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
29038 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
29040 2017-06-08 Martin Liska <mliska@suse.cz>
29042 PR gcov-profile/80911
29043 * gcov.c (block_info::block_info): New constructor.
29045 2017-06-07 Carl Love <cel@us.ibm.com>
29047 * config/rs6000/rs6000-c: The return type of the following
29048 built-in functions was implemented as int not long long. Fix sign
29049 of return value for the unsigned version of vec_mulo and vec_mule.
29050 vector unsigned long long vec_bperm (vector unsigned long long,
29051 vector unsigned char)
29052 vector signed long long vec_mule (vector signed int,
29054 vector unsigned long long vec_mule (vector unsigned int,
29055 vector unsigned int)
29056 vector signed long long vec_mulo (vector signed int,
29058 vector unsigned long long vec_mulo (vector unsigned int,
29059 vector unsigned int)
29060 * doc/extend.texi: Fix the documentation for the built-in
29063 2017-06-07 Carl Love <cel@us.ibm.com>
29066 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
29069 2017-06-07 Carl Love <cel@us.ibm.com>
29071 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
29072 support, Generate doublehv for signed int/float for BE case only.
29074 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
29076 * doc/invoke.texi (mcx16): Rewrite.
29078 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29080 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
29081 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
29082 *mov<mode>_softfloat, and an anonymous splitter): Use
29083 nonimmediate_operand instead of rs6000_nonimmediate_operand.
29085 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29087 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
29089 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
29090 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
29091 (rs6000_debug_reg_global): Adjust.
29092 (rs6000_init_hard_regno_mode_ok): Adjust.
29093 (rs6000_dbx_register_number): Adjust.
29094 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
29095 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
29096 Remove SPE_ACC and SPEFSCR.
29097 (REG_ALLOC_ORDER): Ditto.
29098 (FRAME_POINTER_REGNUM): Change to 111.
29099 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
29100 (REG_CLASS_NAMES): Ditto.
29101 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
29102 (REGISTER_NAMES): Ditto.
29103 (ADDITIONAL_REG_NAMES): Ditto.
29104 (rs6000_reg_names): Ditto.
29105 * config/rs6000/rs6000.md: Renumber some register number
29108 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29110 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
29112 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
29113 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
29115 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
29116 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
29117 Delete the SPE high registers.
29118 (REG_ALLOC_ORDER): Ditto.
29119 (enum reg_class): Remove SPE_HIGH_REGS.
29120 (REG_CLASS_NAMES): Ditto.
29121 (REG_CLASS_CONTENTS): Delete the SPE high registers.
29122 (REGISTER_NAMES): Ditto.
29123 (rs6000_reg_names): Ditto.
29124 * doc/tm.texi.in: Remove SPE as example.
29125 * doc/tm.texi: Regenerate.
29127 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29129 * config/rs6000/8540.md (ppc8540_brinc): Delete.
29130 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
29131 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
29132 * config/rs6000/rs6000.md (type): Remove "brinc".
29134 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29136 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
29137 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
29138 * config/rs6000/linuxspe.h: Delete file.
29139 * config/rs6000/rs6000.md: Don't include spe.md.
29140 * config/rs6000/spe.h: Delete file.
29141 * config/rs6000/spe.md: Delete file.
29142 * config/rs6000/t-rs6000: Remove spe.md.
29144 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29146 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
29147 (reg_or_none500mem_operand): Delete.
29148 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
29149 instead of reg_or_none500mem_operand.
29151 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29153 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29154 handling of SPE flags.
29155 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
29157 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29159 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
29161 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
29162 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
29163 paired_divv2sf3): Similar.
29164 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
29165 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
29166 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
29168 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
29169 Rename the paired_* instruction patterns.
29170 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
29172 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
29173 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
29174 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
29175 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
29176 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
29177 PAIRED_VECTOR_MODE.
29178 (struct machine_function): Delete field spe_insn_chain_scanned_p.
29179 (spe_func_has_64bit_regs_p): Delete.
29180 (spe_expand_predicate_builtin): Delete.
29181 (spe_expand_evsel_builtin): Delete.
29182 (TARGET_DWARF_REGISTER_SPAN): Do not define.
29183 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
29184 (invalid_e500_subreg): Delete.
29185 (rs6000_legitimize_address): Always force_reg op2 as well, for
29186 paired single memory accesses.
29187 (rs6000_member_type_forces_blk): Delete.
29188 (rs6000_spe_function_arg): Delete.
29189 (rs6000_expand_unop_builtin): Delete SPE handling.
29190 (rs6000_expand_binop_builtin): Ditto.
29191 (spe_expand_stv_builtin): Delete.
29192 (bdesc_2arg_spe): Delete.
29193 (spe_expand_builtin): Delete.
29194 (spe_expand_predicate_builtin): Delete.
29195 (spe_expand_evsel_builtin): Delete.
29196 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
29197 (spe_init_builtins): Delete.
29198 (spe_func_has_64bit_regs_p): Delete.
29199 (savres_routine_name): Delete "info" parameter. Adjust callers.
29200 (rs6000_emit_stack_reset): Ditto.
29201 (rs6000_dwarf_register_span): Delete.
29202 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
29203 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
29204 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
29206 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
29208 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
29209 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
29210 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
29211 mulv2sf3, divv2sf3): Delete expanders.
29213 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29215 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
29217 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29219 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
29220 * config/rs6000/rs6000.c: Ditto.
29222 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29224 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
29225 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
29226 comparison_operator.
29228 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29230 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
29231 * config/rs6000/rs6000.opt: Ditto.
29232 * config/rs6000/t-rtems: Ditto.
29234 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29236 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
29237 TARGET_E500_SINGLE by 0, simplify.
29238 * config/rs6000/rs6000.c: Ditto.
29239 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
29240 (spe_build_register_parallel): Delete.
29241 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
29242 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
29243 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
29244 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
29245 (E500_CONVERT): Delete.
29246 * config/rs6000/spe.md: Remove many patterns and all define_constants.
29248 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29250 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
29251 * config/rs6000/dfp.md: Ditto.
29252 (negdd2, *negdd2_fpr): Merge.
29253 (absdd2, *absdd2_fpr): Merge.
29254 (negtd2, *negtd2_fpr): Merge.
29255 (abstd2, *abstd2_fpr): Merge.
29256 * config/rs6000/e500.h: Delete file.
29257 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
29258 TARGET_FPRS by 1 and simplify.
29259 * config/rs6000/rs6000-c.c: Ditto.
29260 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
29261 TARGET_DF_SPE by 0.
29262 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
29264 * config/rs6000/rs6000.md: Ditto.
29265 (floatdidf2, *floatdidf2_fpr): Merge.
29266 (move_from_CR_gt_bit): Delete.
29267 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
29268 (E500_CR_IOR_COMPARE): Delete.
29269 (All patterns that require !TARGET_FPRS): Delete.
29270 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
29272 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29274 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
29276 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29278 * graphds.c (add_edge): Intitialize edge's attached data.
29279 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
29280 pointer parameter. Call pointed function on each edge during
29281 graph traversing. Skip traversing the edge when the function
29283 (graphds_dfs, graphds_scc): Ditto.
29284 (for_each_edge): New parameter. Pass the new parameter to callback
29286 * graphds.h (skip_edge_callback): New function pointer type.
29287 (graphds_dfs, graphds_scc): New function pointer parameter.
29288 (graphds_edge_callback, for_each_edge): New parameter.
29290 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29292 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
29293 out code checking if runtime alias check is possible to below ...
29294 Call the new function.
29295 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
29296 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
29298 2017-06-07 Marek Polacek <polacek@redhat.com>
29301 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
29302 TYPE_OVERFLOW_WRAPS check.
29304 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29306 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
29307 if versioning is required.
29308 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
29309 peeling with the check for versioning.
29311 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29313 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
29314 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
29315 Set true to new parameter if new ssa variable is defined.
29316 (vect_gen_vector_loop_niters): Refactor. Set range information
29317 for the new vector loop bound variable.
29318 (vect_do_peeling): Ditto.
29320 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29322 * tree-affine.c (ssa.h): Include header file.
29323 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
29324 has wrapping overflow behavior.
29326 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29328 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
29330 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29332 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
29333 (tree_to_aff_combination): ... here.
29335 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29337 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
29338 reg_pressure model function.
29339 (ivopts_global_cost_for_size): Delete.
29340 (determine_set_costs, iv_ca_recount_cost): Call new model function
29341 ivopts_estimate_reg_pressure.
29343 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29345 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
29346 expensive than udiv. Remove floating point cases from mod.
29348 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29350 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
29351 Increase idiv cost.
29353 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29355 * config/aarch64/aarch64.md
29356 (copysignsf3): Fix mask generation.
29358 2017-06-07 Jakub Jelinek <jakub@redhat.com>
29360 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
29362 (class dump_manager): Add register_dumps method.
29363 * dumpfile.c: Include langhooks.h.
29364 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
29365 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
29366 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
29367 (dump_manager::dump_register): Start with 512 entries instead of 32.
29368 (dump_manager::register_dumps): New method.
29369 * toplev.c (general_init): Instead of invoking register_dumps
29370 langhook, invoke register_dumps method on the dump manager.
29371 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
29374 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
29376 * doc/md.texi: Clarify the restrictions on a define_insn condition.
29377 Say that # requires an associated define_split to exist, and that
29378 the define_split must be suitable for use after register allocation.
29380 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29382 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
29383 (compute_outgoing_frequencies): Also initialize zero counts.
29384 (find_many_sub_basic_blocks): Do not produce uninitialized profile
29385 around loops; preserve more of profile when nothing changes.
29387 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
29389 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
29391 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
29392 * config/arm/arm-cpu-cdata.h: Regenerate.
29393 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
29394 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
29395 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
29396 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
29397 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
29399 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
29401 * config/arm/t-rmprofile: Likewise.
29402 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
29404 2017-06-06 David S. Miller <davem@davemloft.net>
29407 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
29408 blockage if function uses alloca.
29410 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29412 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
29413 New "uid" fields to hold pretty-print IDs of group and ref.
29414 Memory references are now identified as <group_id>:<ref_id>
29415 instead of using [random] addresses.
29416 (dump_mem_details): Simplify, no functional change.
29417 (dump_mem_ref): Simplify and make output more concise.
29418 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
29419 (find_or_create_group): Initialize group uid.
29420 (record_ref): Initialize ref uid. Improve debug output.
29421 (prune_group_by_reuse, should_issue_prefetch_p,)
29422 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
29423 (mark_nontemporal_store, determine_loop_nest_reuse):
29424 Improve debug output.
29426 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29428 * dbgcnt.def (prefetch): New debug counter.
29429 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
29430 (schedule_prefetches): Stop issueing prefetches if debug counter
29433 2017-06-06 Tom de Vries <tom@codesourcery.com>
29435 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
29436 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
29438 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29440 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
29441 Use aarch64_reg_or_zero predicate for operand 4.
29442 (aarch64_compare_and_swap<mode> define_insn_and_split):
29443 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
29444 (aarch64_store_exclusive<mode>): Likewise for operand 2.
29446 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
29448 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
29449 (arm_compute_save_core_reg_mask): This.
29450 (thumb1_compute_save_reg_mask): Rename into ...
29451 (thumb1_compute_save_core_reg_mask): This.
29452 (arm_compute_save_reg0_reg12_mask): Adapt comment.
29453 (arm_compute_frame_layout): Likewise.
29455 2017-06-06 Richard Biener <rguenther@suse.de>
29457 PR tree-optimization/80974
29458 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
29459 keep or clear leaders SSA info.
29461 2017-06-06 Tom de Vries <tom@codesourcery.com>
29463 * config/nvptx/nvptx.c (split_mode_p): New function.
29464 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
29466 2017-06-06 Tom de Vries <tom@codesourcery.com>
29468 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
29470 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29473 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
29476 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29478 * shrink-wrap.c (handle_simple_exit): Update profile.
29479 (try_shrink_wrapping): Upate profile.
29481 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29483 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
29484 (tree_guess_outgoing_edge_probabilities): New.
29485 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
29486 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
29488 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29490 * ipa-split.c (split_function): Initialize return bb profile.
29492 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29494 * profile.c (compute_branch_probabilities): Also initialize
29495 EXIT_BLOCK profile.
29497 2017-06-06 Richard Biener <rguenther@suse.de>
29499 PR tree-optimization/80928
29500 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
29501 (vect_analyze_loop_operations): Properly guard analysis for
29503 (vect_transform_loop): Likewise.
29504 (vect_analyze_loop_2): Also reset SLP type on PHIs.
29505 (vect_model_induction_cost): Do not cost for pure SLP.
29506 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
29507 of induction in inner loop vectorization.
29508 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
29509 (vect_get_and_check_slp_defs): Handle vect_induction_def.
29510 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
29512 (vect_analyze_slp_cost_1): Cost induction.
29513 (vect_detect_hybrid_slp_stmts): Handle PHIs.
29514 (vect_get_slp_vect_defs): Likewise.
29515 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
29516 (vect_transform_stmt): Handle SLP reductions.
29517 * tree-vectorizer.h (vectorizable_induction): Adjust.
29519 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29521 * config/rs6000/rs6000.c (make_resolver_func): Update
29522 init_lowered_empty_function call.
29524 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
29526 * doc/invoke.texi: Document the -fprofile-abs-path option.
29527 * common.opt (fprofile-abs-path): New option.
29528 * gcov-io.h (gcov_write_filename): Declare.
29529 * gcov-io.c (gcov_write_filename): New function.
29530 * coverage.c (coverage_begin_function): Use gcov_write_filename.
29531 * profile.c (output_location): Likewise.
29533 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29535 * shring-wrap.c: Revert accidental commit.
29537 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
29539 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
29541 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29543 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
29545 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
29546 profile in callgraph edge.
29547 * profile-count.h (apply_probability): If THIS is 0, then result is 0
29548 (apply_scale): Likewise.
29549 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
29550 Also scale profile when inlining function with zero profile.
29551 (initialize_cfun): Update exit block profile even when it is zero.
29552 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
29553 when profile is read.
29555 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29557 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
29558 (CLONE_*): New constants to define the processors we can generate
29559 code for with the target_clone attribute.
29560 (rs6000_clone_map): New array to identify which clone processors
29561 the current program is running on.
29562 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
29563 target_clone attribute.
29564 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
29565 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
29566 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
29567 (cpu_expand_builtin): Add support for target_clone attribute.
29568 (rs6000_valid_attribute_p): Allow "default" attribute.
29569 (get_decl_name): New debug function to simplify printing the
29570 current function name in debugging statements.
29571 (rs6000_clone_priority): New functions to support the target_clone
29572 attribute, and be able to generate code to switch between ISA 2.05
29573 through ISA 3.0 (power6 through power9).
29574 (rs6000_compare_version_priority): Likewise.
29575 (rs6000_get_function_versions_dispatcher): Likewise.
29576 (make_resolver_func): Likewise.
29577 (add_condition_to_bb): Likewise.
29578 (dispatch_function_versions): Likewise.
29579 (rs6000_generate_version_dispatcher_body): Likewise.
29580 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
29581 (fusion_gpr_load_p): Fix a spacing issue.
29582 * doc/extend.texi (Common Function Attributes): Document that the
29583 PowerPC supports the target_clone attribute.
29585 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29587 * config/arm/arm.h: explain F symbol found in description of ARM
29588 register allocation in its legend.
29590 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29592 * config/mips/frame-header-opt.c: Include profile-count.h.
29593 * config/riscv/riscv.c: Include profile-count.h
29595 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29597 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
29599 (sm_set_flag_if_changed): Add bbs field.
29600 (execute_sm_if_changed_flag_set): Pass BBS.
29601 (execute_sm): Update.
29603 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29605 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
29608 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29610 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
29611 (peephole2): New peephole2 to emit the above.
29612 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
29614 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29616 * config/aarch64/aarch64.c (define_peephole2 above
29617 *sub_<shift>_<mode>): New peephole.
29619 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29621 * config/i386/i386.c (make_resolver_func): Update.
29622 * Makefile.in: Add profile-count.h and profile-count.o
29623 * auto-profile.c (afdo_indirect_call): Update to new API.
29624 (afdo_set_bb_count): Update.
29625 (afdo_propagate_edge): Update.
29626 (afdo_propagate_circuit): Update.
29627 (afdo_calculate_branch_prob): Update.
29628 (afdo_annotate_cfg): Update.
29629 * basic-block.h: Include profile-count.h
29630 (struct edge_def): Turn count to profile_count.
29631 (struct basic_block_def): Likewie.
29632 (REG_BR_PROB_BASE): Move to profile-count.h
29633 (RDIV): Move to profile-count.h
29634 * bb-reorder.c (max_entry_count): Turn to profile_count.
29635 (find_traces): Update.
29636 (rotate_loop):Update.
29637 (connect_traces):Update.
29638 (sanitize_hot_paths):Update.
29639 * bt-load.c (migrate_btr_defs): Update.
29640 * cfg.c (RDIV): Remove.
29641 (init_flow): Use alloc_block.
29642 (alloc_block): Uninitialize count.
29643 (unchecked_make_edge): Uninitialize count.
29644 (check_bb_profile): Update.
29645 (dump_edge_info): Update.
29646 (dump_bb_info): Update.
29647 (update_bb_profile_for_threading): Update.
29648 (scale_bbs_frequencies_int): Update.
29649 (scale_bbs_frequencies_gcov_type): Update.
29650 (scale_bbs_frequencies_profile_count): New.
29651 * cfg.h (update_bb_profile_for_threading): Update.
29652 (scale_bbs_frequencies_profile_count): Declare.
29653 * cfgbuild.c (compute_outgoing_frequencies): Update.
29654 (find_many_sub_basic_blocks): Update.
29655 * cfgcleanup.c (try_forward_edges): Update.
29656 (try_crossjump_to_edge): Update.
29657 * cfgexpand.c (expand_gimple_tailcall): Update.
29658 (construct_exit_block): Update.
29659 * cfghooks.c (verify_flow_info): Update.
29660 (dump_bb_for_graph): Update.
29661 (split_edge): Update.
29662 (make_forwarder_block): Update.
29663 (duplicate_block): Update.
29664 (account_profile_record): Update.
29665 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
29666 (get_estimated_loop_iterations): Update.
29667 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
29668 (single_likely_exit): Update.
29669 * cfgloopmanip.c (scale_loop_profile): Update.
29671 (set_zero_probability): Update.
29672 (lv_adjust_loop_entry_edge): Update.
29673 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
29674 (purge_dead_edges): Update.
29675 (rtl_account_profile_record): Update.
29676 * cgraph.c (cgraph_node::create): Uninitialize count.
29677 (symbol_table::create_edge): Uninitialize count.
29678 (cgraph_update_edges_for_call_stmt_node): Update.
29679 (cgraph_edge::dump_edge_flags): Update.
29680 (cgraph_node::dump): Update.
29681 (cgraph_edge::maybe_hot_p): Update.
29682 * cgraph.h: Include profile-count.h
29683 (create_clone), create_edge, create_indirect_edge): Update.
29684 (cgraph_node): Turn count to profile_count.
29685 (cgraph_edge0: Likewise.
29686 (make_speculative, clone): Update.
29687 (create_edge): Update.
29688 (init_lowered_empty_function): Update.
29689 * cgraphclones.c (cgraph_edge::clone): Update.
29690 (duplicate_thunk_for_node): Update.
29691 (cgraph_node::create_clone): Update.
29692 * cgraphunit.c (cgraph_node::analyze): Update.
29693 (cgraph_node::expand_thunk): Update.
29694 * final.c (dump_basic_block_info): Update.
29695 * gimple-streamer-in.c (input_bb): Update.
29696 * gimple-streamer-out.c (output_bb): Update.
29697 * graphite.c (print_global_statistics): Update.
29698 (print_graphite_scop_statistics): Update.
29699 * hsa-brig.c: Include basic-block.h.
29700 * hsa-dump.c: Include basic-block.h.
29701 * hsa-gen.c (T sum_slice): Update.
29702 (convert_switch_statements):Update.
29703 * hsa-regalloc.c: Include basic-block.h.
29704 * ipa-chkp.c (chkp_produce_thunks): Update.
29705 * ipa-cp.c (struct caller_statistics): Update.
29706 (init_caller_stats): Update.
29707 (gather_caller_stats): Update.
29708 (ipcp_cloning_candidate_p): Update.
29709 (good_cloning_opportunity_p): Update.
29710 (get_info_about_necessary_edges): Update.
29711 (dump_profile_updates): Update.
29712 (update_profiling_info): Update.
29713 (update_specialized_profile): Update.
29714 (perhaps_add_new_callers): Update.
29715 (decide_about_value): Update.
29716 (ipa_cp_c_finalize): Update.
29717 * ipa-devirt.c (struct odr_type_warn_count): Update.
29718 (struct decl_warn_count): Update.
29719 (struct final_warning_record): Update.
29720 (possible_polymorphic_call_targets): Update.
29721 (ipa_devirt): Update.
29722 * ipa-fnsummary.c (redirect_to_unreachable): Update.
29723 * ipa-icf.c (sem_function::merge): Update.
29724 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
29725 * ipa-inline.c (compute_uninlined_call_time): Update.
29726 (compute_inlined_call_time): Update.
29727 (want_inline_small_function_p): Update.
29728 (want_inline_self_recursive_call_p): Update.
29729 (edge_badness): Update.
29730 (lookup_recursive_calls): Update.
29731 (recursive_inlining): Update.
29732 (inline_small_functions): Update.
29733 (dump_overall_stats): Update.
29734 (dump_inline_stats): Update.
29735 * ipa-profile.c (ipa_profile_generate_summary): Update.
29736 (ipa_propagate_frequency): Update.
29737 (ipa_profile): Update.
29738 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29739 * ipa-utils.c (ipa_merge_profiles): Update.
29740 * loop-doloop.c (doloop_modify): Update.
29741 * loop-unroll.c (report_unroll): Update.
29742 (unroll_loop_runtime_iterations): Update.
29743 * lto-cgraph.c (lto_output_edge): Update.
29744 (lto_output_node): Update.
29745 (input_node): Update.
29746 (input_edge): Update.
29747 (merge_profile_summaries): Update.
29748 * lto-streamer-in.c (input_cfg): Update.
29749 * lto-streamer-out.c (output_cfg): Update.
29750 * mcf.c (create_fixup_graph): Update.
29751 (adjust_cfg_counts): Update.
29752 (sum_edge_counts): Update.
29753 * modulo-sched.c (sms_schedule): Update.
29754 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
29755 * predict.c (maybe_hot_count_p): Update.
29756 (probably_never_executed): Update.
29757 (dump_prediction): Update.
29758 (combine_predictions_for_bb): Update.
29759 (propagate_freq): Update.
29760 (handle_missing_profiles): Update.
29761 (counts_to_freqs): Update.
29762 (rebuild_frequencies): Update.
29763 (force_edge_cold): Update.
29764 * predict.h: Include profile-count.h
29765 (maybe_hot_count_p, counts_to_freqs): UPdate.
29766 * print-rtl-function.c: Do not include cfg.h
29767 * print-rtl.c: Include basic-block.h
29768 * profile-count.c: New file.
29769 * profile-count.h: New file.
29770 * profile.c (is_edge_inconsistent): Update.
29771 (correct_negative_edge_counts): Update.
29772 (is_inconsistent): Update.
29773 (set_bb_counts): Update.
29774 (read_profile_edge_counts): Update.
29775 (compute_frequency_overlap): Update.
29776 (compute_branch_probabilities): Update; Initialize and deinitialize
29778 (branch_prob): Update.
29779 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
29780 (edge_gcov_count): New.
29781 (bb_gcov_count): New.
29782 * shrink-wrap.c (try_shrink_wrapping): Update.
29783 * tracer.c (better_p): Update.
29784 * trans-mem.c (expand_transaction): Update.
29785 (ipa_tm_insert_irr_call): Update.
29786 (ipa_tm_insert_gettmclone_call): Update.
29787 * tree-call-cdce.c: Update.
29788 * tree-cfg.c (gimple_duplicate_sese_region): Update.
29789 (gimple_duplicate_sese_tail): Update.
29790 (gimple_account_profile_record): Update.
29791 (execute_fixup_cfg): Update.
29792 * tree-inline.c (copy_bb): Update.
29793 (copy_edges_for_bb): Update.
29794 (initialize_cfun): Update.
29795 (freqs_to_counts): Update.
29796 (copy_cfg_body): Update.
29797 (expand_call_inline): Update.
29798 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29799 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29800 (try_unroll_loop_completely): Update.
29801 (try_peel_loop): Update.
29802 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29803 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29804 * tree-ssa-loop-split.c (connect_loops): Update.
29805 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29806 * tree-ssa-reassoc.c (branch_fixup): Update.
29807 * tree-ssa-tail-merge.c (replace_block_by): Update.
29808 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29809 (compute_path_counts): Update.
29810 (update_profile): Update.
29811 (recompute_probabilities): Update.
29812 (update_joiner_offpath_counts): Update.
29813 (estimated_freqs_path): Update.
29814 (freqs_to_counts_path): Update.
29815 (clear_counts_path): Update.
29816 (ssa_fix_duplicate_block_edges): Update.
29817 (duplicate_thread_path): Update.
29818 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29819 (struct switch_conv_info): Update.
29820 * tree-tailcall.c (decrease_profile): Update.
29821 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29822 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29823 * value-prof.c (check_counter): Update.
29824 (gimple_divmod_fixed_value): Update.
29825 (gimple_mod_pow2): Update.
29826 (gimple_mod_subtract): Update.
29827 (gimple_ic_transform): Update.
29828 (gimple_stringop_fixed_value): Update.
29829 * value-prof.h (gimple_ic): Update.
29831 2017-06-02 Carl Love <cel@us.ibm.com>
29833 * config/rs6000/rs6000-c: Add support for built-in functions
29834 vector double vec_doublee (vector signed int);
29835 vector double vec_doublee (vector unsigned int);
29836 vector double vec_doublee (vector float);
29837 vector double vec_doubleh (vector signed int);
29838 vector double vec_doubleh (vector unsigned int);
29839 vector double vec_doubleh (vector float);
29840 vector double vec_doublel (vector signed int);
29841 vector double vec_doublel (vector unsigned int);
29842 vector double vec_doublel (vector float);
29843 vector double vec_doubleo (vector signed int);
29844 vector double vec_doubleo (vector unsigned int);
29845 vector double vec_doubleo (vector float);.
29846 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29847 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29849 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29850 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29851 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29853 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29854 vec_doublel, vec_doubleh.
29855 * doc/extend.texi: Update the built-in documentation file for the
29856 new built-in functions.
29858 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29861 * ipa-inline-analysis.c (free_growth_caches): Set
29862 edge_removal_hook_holder to NULL after removing it.
29864 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29866 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29867 comparision with zero.
29869 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29870 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29871 for early expansion of vec_min and vec_max builtins.
29872 (builtin_function_type): Add min/max unsigned variants to those
29873 identified as having unsigned arguments.
29875 2017-06-02 Olivier Hainque <hainque@adacore.com>
29877 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29879 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29881 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29882 Use VALL_F16 iterator rather than VALL.
29884 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29886 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29887 Emit CBNZ inside loop when doing a strong exchange and comparing
29888 against zero. Generate the CC flags after the loop.
29890 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29892 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29893 (dl_section_ref): New.
29894 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29895 On AIX, append an expression to subtract the size of the
29896 section length to dl_section_ref.
29898 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29900 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29901 for early expansion of vector absolute builtins.
29903 2017-06-02 Richard Biener <rguenther@suse.de>
29905 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29906 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29908 2017-06-02 Richard Biener <rguenther@suse.de>
29910 PR tree-optimization/80948
29911 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29912 stmt order as well.
29914 2017-06-02 Richard Biener <rguenther@suse.de>
29916 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29918 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29919 for inductions as relevant.
29921 2017-06-02 Richard Biener <rguenther@suse.de>
29923 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29924 (vectorizable_induction): ... this. Remove dead code.
29926 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29928 * builtins. (expand_builtin_alloca): Remove second parameter and
29929 infer its value from the first parameter instead.
29930 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29932 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29934 PR rtl-optimization/80903
29935 * loop-doloop.c (add_test): Unshare sequence.
29937 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29939 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29941 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29943 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29945 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29946 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29947 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29948 (xlogue_layout::get_stub_rtx): Make static.
29949 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29950 (xlogue_layout::compute_stub_managed_regs): Rename to...
29951 (xlogue_layout::count_stub_managed_regs): ...this.
29952 (xlogue_layout::is_stub_managed_reg): New function.
29953 (xlogue_layout::m_stub_names): Rename to...
29954 (xlogue_layout::s_stub_names): ...this, make static.
29955 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29956 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29957 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29958 xlogue_layout::s_stub_names): Instantiate statics.
29959 (stub_managed_regs): Remove.
29960 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29961 (disable_call_ms2sysv_xlogues): Rename to...
29962 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29963 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29965 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29966 change after reload_completed.
29967 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29969 (ix86_expand_prologue): Likewise.
29970 (ix86_expand_epilogue): Likewise.
29971 (ix86_expand_split_stack_prologue): Likewise.
29972 (ix86_compute_frame_layout): Remove frame parameter ...
29973 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29974 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29976 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29977 (ix86_frame): Move from here ...
29978 * config/i386/i386.h (ix86_frame): ... to here.
29979 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29980 complete ix86_frame data structure instead. Remove some_ld_name.
29982 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29984 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29985 symbols that hold a DECL_VALUE_EXPR.
29987 2017-06-01 Martin Jambor <mjambor@suse.cz>
29989 PR tree-optimization/80898
29990 * tree-sra.c (process_subtree_disqualification): Removed.
29991 (disqualify_candidate): Do not acll
29992 process_subtree_disqualification.
29993 (subtree_mark_written_and_enqueue): New function.
29994 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29995 RHS has been disqualified and re-queue LHS if necessary. Apart
29996 from that, ignore disqualified RHS.
29998 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30000 * config/s390/s390.c (s390_emit_epilogue): Disable early return
30001 address fetch for z10 or later.
30003 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30005 * config/arc/arc.md (tst_movb): Add guard when splitting.
30007 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30009 * config/arc/arc.c (arc_can_eliminate): Test against
30010 arc_frame_pointer_needed.
30012 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30014 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
30015 to prevent store reordering.
30016 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
30017 (type): Add block type.
30018 (stack_tie): Define special instruction to be used in
30021 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30023 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
30024 constraint. It is not valid for the pattern.
30025 (noncommutative_binary_comparison): Likewise.
30027 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30029 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
30032 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30034 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
30035 be used by the reg-alloc.
30037 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30039 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
30040 reg-alloc when having mul64 or mul32x16 instructions.
30041 (mulsidi3): Likewise.
30042 (umulsidi3): Likewise.
30043 (mulsi32x16): New pattern.
30044 (mulsi64): Likewise.
30045 (mulsidi64): Likewise.
30046 (umulsidi64): Likewise.
30047 (MUL32x16_REG): Define.
30048 (mul64_600): Use MUL32x16_REG.
30049 (mac64_600): Likewise.
30050 (umul64_600): Likewise.
30051 (umac64_600): Likewise.
30053 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30055 * config/arc/arc.md (mulsi3_700): Make it commutative.
30057 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
30059 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
30061 (*sign_extendsidi2_insn): Likewise for movstosw.
30063 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
30065 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
30066 the type of the input discriminant value. Convert the
30067 discriminant value of signedness vary.
30069 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
30071 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
30072 Add to -Wall section.
30074 2017-06-01 Richard Biener <rguenther@suse.de>
30076 PR middle-end/66313
30077 * fold-const.c (fold_plusminus_mult_expr): If the factored
30078 factor may be zero use a wrapping type for the inner operation.
30079 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
30080 and handle moved defs.
30081 (process_assignment): Properly guard the unary op case. Return a
30082 tri-state indicating that moving the stmt before the call may allow
30083 to continue. Pass through to_move.
30084 (find_tail_calls): Handle moving unrelated defs before
30087 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
30090 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
30091 splitter result in the canonical way.
30093 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
30095 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
30096 also for 32bit target. Update insn attributes.
30097 (zero-extendsidi2 splitter): Allow all registers for operand 1.
30099 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
30101 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
30102 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
30103 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
30104 (_mm_maskz_min_ss): New intrinsics.
30106 2017-05-31 Martin Liska <mliska@suse.cz>
30108 * tree-vect-loop.c (vect_create_epilog_for_reduction):
30109 Change comment style to one we normally use.
30110 (vectorizable_reduction): Likewise.
30111 (vectorizable_induction): Likewise.
30112 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
30113 (vectorizable_call): Likewise.
30114 (vectorizable_simd_clone_call): Likewise.
30115 (vectorizable_conversion): Likewise.
30116 (vectorizable_assignment): Likewise.
30117 (vectorizable_shift): Likewise.
30118 (vectorizable_operation): Likewise.
30119 (vectorizable_store): Likewise.
30120 (vectorizable_load): Likewise.
30121 * tree-vectorizer.h: Likewise.
30123 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
30125 * passes.c (emergency_dump_function): New.
30126 * tree-pass.h (emergency_dump_function): Declare.
30127 * plugin.c (plugins_internal_error_function): Remove.
30128 * plugin.h (plugins_internal_error_function): Remove declaration.
30129 * toplev.c (internal_error_function): New static function. Use it...
30130 (general_init): ...here.
30132 2017-05-31 Graham Markall <graham.markall@embecosm.com>
30134 * config/arc/arc.c (arc_print_operand): Handle constant operands.
30135 (arc_rtx_costs): Add costs for new patterns.
30136 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
30137 * config/arc/predicates.md: Add _1_2_3_operand predicate.
30139 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
30141 * tree-ssa-strlen.c (get_next_strinfo): New function.
30142 (get_stridx_plus_constant): Use it.
30143 (zero_length_string): Likewise.
30144 (adjust_related_strinfos): Likewise.
30145 (adjust_last_stmt): Likewise.
30147 2017-05-31 Richard Biener <rguenther@suse.de>
30150 * config/i386/i386.c (ix86_expand_builtin): Remove assert
30151 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
30153 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
30155 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
30156 loop_vinfo argument and use of dependence distance vectors.
30157 Check instead whether the two references differ only in their
30158 initial value and assume that they have the same alignment if the
30159 difference is a multiple of the vector alignment.
30160 (vect_analyze_data_refs_alignment): Update call accordingly.
30162 2017-05-31 Martin Liska <mliska@suse.cz>
30165 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
30167 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30169 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
30170 (create_intersect_range_checks): Move from ...
30171 * tree-data-ref.c (create_intersect_range_checks_index)
30172 (create_intersect_range_checks): ... to here.
30173 (create_runtime_alias_checks): New function factored from ...
30174 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
30175 here. Call above function.
30176 * tree-data-ref.h (create_runtime_alias_checks): New function.
30178 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30180 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
30181 segment length for dr_b and compute it in wide_int.
30183 2017-05-31 Richard Biener <rguenther@suse.de>
30185 PR tree-optimization/80906
30186 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
30187 and pass through iv_map.
30188 (copy_bb_and_scalar_dependences): Adjust.
30189 (translate_pending_phi_nodes): Likewise.
30190 (copy_loop_close_phi_args): Handle code-generating IVs instead
30193 2017-05-30 David Malcolm <dmalcolm@redhat.com>
30195 * diagnostic-color.c (color_dict): Add "type-diff".
30196 (parse_gcc_colors): Update comment.
30197 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
30198 -fdiagnostics-show-template-tree and -fno-elide-type.
30199 (GCC_COLORS): Add type-diff to example.
30201 (-fdiagnostics-show-template-tree): New.
30202 (-fno-elide-type): New.
30203 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
30204 the pp_format_decoder callback. Call any m_format_postprocessor's
30206 (pretty_printer::pretty_printer): Initialize
30207 m_format_postprocessor.
30208 (pretty_printer::~pretty_printer): Delete any
30209 m_format_postprocessor.
30210 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
30211 (class format_postprocessor): New class.
30212 (struct pretty_printer::format_decoder): Document the new parameters.
30213 (struct pretty_printer::m_format_postprocessor): New field.
30214 * tree-diagnostic.c (default_tree_printer): Update for new
30215 bool and const char ** params.
30216 * tree-diagnostic.h (default_tree_printer): Likewise.
30218 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
30220 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
30221 (lwa_operand): Delete rs6000_gen_cell_microcode test.
30222 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
30223 rs6000_gen_cell_microcode code.
30224 (rs6000_final_prescan_insn): Delete.
30225 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
30226 "warn-cell-microcode" entries.
30227 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
30228 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
30229 throughout. Change cc_reg_not_micro_cr0_operand to
30230 cc_reg_not_cr0_operand throughout.
30231 (*extendhi<mode>2_noload): Delete.
30232 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
30233 (mwarn-cell-microcode): Delete.
30234 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
30235 -mgen-cell-microcode and -mwarn-cell-microcode.
30237 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
30240 * config/i386/constraints.md (Yd): New constraint.
30242 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
30243 and (?Yd, r) alternatives. Update insn attributes.
30244 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
30245 and (?*Yd, r) alternatives. Update insn attributes.
30246 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
30248 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
30250 * gimplify.c (gimplify_modify_expr): Don't create a
30251 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
30254 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
30256 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
30258 2017-05-30 Richard Biener <rguenther@suse.de>
30260 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
30261 and reduc_def fields.
30262 (STMT_VINFO_REDUC_TYPE): New define.
30263 (STMT_VINFO_REDUC_DEF): Likewise.
30264 (vect_force_simple_reduction): Adjust prototype.
30265 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
30266 (vect_is_simple_reduction): Remove check_reduction argument.
30267 (vect_force_simple_reduction): Adjust and set
30268 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30269 (vectorizable_reduction): Do not re-do reduction analysis
30270 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30271 * tree-parloops.c (gather_scalar_reductions): Adjust.
30273 2017-05-30 Richard Biener <rguenther@suse.de>
30275 PR middle-end/80901
30276 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
30279 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
30281 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
30282 Introduce unknown_misalignment parameter and remove vf.
30283 (vect_peeling_hash_get_lowest_cost):
30284 Pass unknown_misalignment parameter.
30285 (vect_enhance_data_refs_alignment):
30286 Fix unsupportable data ref treatment.
30288 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30290 * tree-vect-data-refs.c (vect_get_data_access_cost):
30291 Workaround for SLP handling.
30292 (vect_enhance_data_refs_alignment):
30293 Compute costs for doing no peeling at all, compare to the best
30294 peeling costs so far and avoid peeling if cheaper.
30296 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30298 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
30299 Return peeling info and set costs to zero for unlimited cost
30301 (vect_enhance_data_refs_alignment): Also inspect all datarefs
30302 with unknown misalignment. Compute and costs for unknown
30303 misalignment, compare them to the costs for known misalignment
30304 and choose the cheapest for peeling.
30306 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30308 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
30309 (vect_get_peeling_costs_all_drs): Create function.
30310 (vect_peeling_hash_get_lowest_cost):
30311 Use vect_get_peeling_costs_all_drs.
30312 (vect_peeling_supportable): Create function.
30313 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
30315 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30317 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
30318 DR_HAS_NEGATIVE_STEP.
30319 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
30320 (vect_enhance_data_refs_alignment): Use.
30321 (vect_duplicate_ssa_name_ptr_info): Use.
30322 * tree-vectorizer.h (dr_misalignment): Use.
30323 (known_alignment_for_access_p): Use.
30325 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
30328 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
30330 (has_section_name): New function.
30332 2017-05-30 Martin Liska <mliska@suse.cz>
30335 * auto-profile.c (get_function_decl_from_block): Fix
30338 2017-05-30 Richard Biener <rguenther@suse.de>
30340 PR middle-end/80876
30341 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
30343 2017-05-30 Martin Liska <mliska@suse.cz>
30345 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
30346 * dumpfile.h (struct dump_file_info): Remove ctors.
30348 2017-05-30 Martin Liska <mliska@suse.cz>
30350 * predict.def: Fix GNU coding style.
30352 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30354 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30355 Mark 'to' argument with ATTRIBUTE_UNUSED.
30357 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30359 * config/xtensa/xtensa.c (xtensa_emit_call): Use
30360 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
30361 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
30364 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
30366 * doc/install.texi (Options specification): Restore entry of
30367 --enable-sjlj-exceptions.
30369 2017-05-27 Michael Eager <eager@eagercon.com>
30372 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
30374 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
30376 * config/microblaze/microblaze.h
30377 (FIXED_REGISTERS): Update in macro.
30378 (CALL_USED_REGISTERS): Update in macro.
30380 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
30382 * doc/install.texi: Add links to macOS binary distributions.
30384 2017-05-27 Jakub Jelinek <jakub@redhat.com>
30388 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30390 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30392 2017-05-26 Martin Liska <mliska@suse.cz>
30394 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
30396 2017-05-26 Martin Liska <mliska@suse.cz>
30398 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
30399 always leading ';; '.
30400 (dump_bb_info): Likewise.
30401 (brief_dump_cfg): Likewise.
30402 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
30403 * dumpfile.c: Remove usage of TDF_VERBOSE.
30404 * dumpfile.h (enum dump_kind): Likewise.
30405 (dump_gimple_bb_header): Do not use TDF_COMMENT.
30406 * print-tree.c (debug_verbose): Remove.
30407 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
30408 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
30409 * tree-diagnostic.c (default_tree_printer): Replace
30410 TDF_DIAGNOSTIC with TDF_SLIM.
30412 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30414 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
30415 in parameter loop, rather than loop_vinfo.
30416 (create_intersect_range_checks): Ditto.
30417 (vect_create_cond_for_alias_checks): Update call to above functions.
30419 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30421 PR tree-optimization/80815
30422 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
30423 for merging runtime alias checks. Handle negative DR_STEPs.
30425 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30427 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
30429 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
30430 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
30431 out code pruning runtime alias checks.
30432 * tree-data-ref.c (prune_runtime_alias_test_list): New function
30433 factored out from above.
30434 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30436 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30438 (prune_runtime_alias_test_list): New decalaration.
30440 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30442 * tree-vect-data-refs.c (compare_tree): Rename and move ...
30443 * tree-data-ref.c (data_ref_compare_tree): ... to here.
30444 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
30445 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
30446 (operator==, comp_dr_with_seg_len_pair): Ditto.
30447 (vect_prune_runtime_alias_test_list): Ditto.
30449 2017-05-26 Martin Liska <mliska@suse.cz>
30452 * params.def: Bound partial-inlining-entry-probability param.
30454 2017-05-26 Marek Polacek <polacek@redhat.com>
30457 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
30460 2017-05-26 Richard Biener <rguenther@suse.de>
30462 PR tree-optimization/80842
30463 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
30466 2017-05-26 Richard Biener <rguenther@suse.de>
30468 PR tree-optimization/80844
30469 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
30471 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
30473 * doc/md.texi (Machine Constraints): Update x86 family
30474 machine constraints section to match 'config/i386/constraints.md'.
30476 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
30478 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
30480 2017-05-25 Nathan Sidwell <nathan@acm.org>
30482 * doc/invoke.texi (--enable-languages): Update documentation.
30484 2017-05-25 Martin Liska <mliska@suse.cz>
30486 * dumpfile.c: Add TDF_FOLDING.
30487 * dumpfile.h (enum dump_kind): Likewise.
30488 * genmatch.c (dt_simplify::gen_1): Use it.
30490 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30492 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
30494 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30496 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30497 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
30499 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30501 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
30502 * match.pd (X == C): Rewrite it here.
30503 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
30504 with_certain_nonzero_bits2): New predicates.
30505 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
30507 2017-05-24 Nathan Sidwell <nathan@acm.org>
30509 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
30512 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
30515 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30517 * config/powerpcspe: New port. Files are copied from the rs6000
30518 port, with "rs6000" in filenames replaced by "powerpcspe".
30520 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
30522 PR rtl-optimization/80754
30523 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
30525 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
30527 * config/sparc/sparc.md (length): Return the correct value for -mflat
30528 sibcalls to match output_sibcall.
30530 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30534 * config/rs6000/rs6000.c (struct machine_function): Add new field
30536 (rs6000_get_separate_components): Init that field, use it.
30537 (rs6000_components_for_bb): Use the field.
30539 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
30541 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
30543 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
30545 PR middle-end/80823
30546 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
30548 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30551 * config/s390/s390.c (s390_check_qrst_address): Check incoming
30552 address against address_operand predicate.
30553 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
30555 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
30557 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
30558 parameters passed indirectly.
30560 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
30562 * config/i386/i386.md (*movdi_internal): Remove SSE4
30563 alternative 18 (?r, *v). Update insn attributes.
30564 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
30565 Update insn attributes.
30566 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
30567 Update insn attributes.
30568 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
30569 alternative 1 (r, v). Remove isa attribute.
30570 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
30571 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
30572 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
30574 2017-05-23 Tom de Vries <tom@codesourcery.com>
30576 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
30579 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30581 * cgraphunit.c (symbol_table::process_new_functions): Update.
30582 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
30583 (inline_generate_summary): Rename to ...
30584 (ipa_fn_summary_generate): ... this one.
30585 (inline_read_summary): Rename to ...
30586 (ipa_fn_summary_read): ... this one.
30587 (inline_write_summary): Rename to ...
30588 (ipa_fn_summary_write): ... this one.
30589 (inline_free_summary): Rename to ...
30590 (ipa_free_fn_summary): ... this one.
30591 (pass_data_local_fn_summary, pass_local_fn_summary,
30592 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
30593 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
30594 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
30595 make_pass_ipa_fn_summary): New.
30596 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
30597 inline_write_summary, inline_free_summary): Remove.
30598 (ipa_free_fn_summary) : New.
30599 * ipa-inline.c (ipa_inline): Update.
30600 (pass_ipa_inline): Do not generate summaries.
30601 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
30603 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
30604 and add pass_ipa_fn_summary.
30605 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
30607 (make_pass_inline_parameters): Remove.
30609 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
30611 * omp-low.c (struct omp_context): Remove "default_kind" member.
30614 * omp-offload.c (execute_oacc_device_lower): Remove the
30615 parallelism dimensions function attributes for unparallelized
30616 OpenACC kernels constructs.
30618 2017-05-23 Martin Liska <mliska@suse.cz>
30620 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
30622 (cgraph_edge::make_speculative): Likewise.
30623 (cgraph_edge::resolve_speculation): Likewise.
30624 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30625 (cgraph_node::dump): Likewise.
30626 * cgraph.h: Likewise.
30627 * cgraphunit.c (analyze_functions): Likewise.
30628 (symbol_table::compile): Likewise.
30629 * ipa-cp.c (print_all_lattices): Likewise.
30630 (determine_versionability): Likewise.
30631 (initialize_node_lattices): Likewise.
30632 (ipcp_verify_propagated_values): Likewise.
30633 (estimate_local_effects): Likewise.
30634 (update_profiling_info): Likewise.
30635 (create_specialized_node): Likewise.
30636 (perhaps_add_new_callers): Likewise.
30637 (decide_about_value): Likewise.
30638 (decide_whether_version_node): Likewise.
30639 (identify_dead_nodes): Likewise.
30640 (ipcp_store_bits_results): Likewise.
30641 * ipa-devirt.c (dump_targets): Likewise.
30642 (ipa_devirt): Likewise.
30643 * ipa-icf.c (sem_item::dump): Likewise.
30644 (sem_function::equals): Likewise.
30645 (sem_variable::equals): Likewise.
30646 (sem_item_optimizer::read_section): Likewise.
30647 (sem_item_optimizer::execute): Likewise.
30648 (congruence_class::dump): Likewise.
30649 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
30650 (dump_inline_summary): Likewise.
30651 (estimate_node_size_and_time): Likewise.
30652 (inline_analyze_function): Likewise.
30653 * ipa-inline-transform.c (inline_call): Likewise.
30654 * ipa-inline.c (report_inline_failed_reason): Likewise.
30655 (want_early_inline_function_p): Likewise.
30656 (edge_badness): Likewise.
30657 (update_edge_key): Likewise.
30658 (inline_small_functions): Likewise.
30659 * ipa-profile.c (ipa_profile): Likewise.
30660 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
30661 (ipa_make_edge_direct_to_target): Likewise.
30662 (remove_described_reference): Likewise.
30663 (ipa_impossible_devirt_target): Likewise.
30664 (propagate_controlled_uses): Likewise.
30665 (ipa_print_node_params): Likewise.
30666 (ipcp_transform_function): Likewise.
30667 * ipa-pure-const.c (pure_const_read_summary): Likewise.
30668 (propagate_pure_const): Likewise.
30669 * ipa-reference.c (generate_summary): Likewise.
30670 (read_write_all_from_decl): Likewise.
30671 (propagate): Likewise.
30672 (ipa_reference_read_optimization_summary): Likewise.
30673 * ipa-utils.c (ipa_merge_profiles): Likewise.
30674 * ipa.c (walk_polymorphic_call_targets): Likewise.
30675 (symbol_table::remove_unreachable_nodes): Likewise.
30676 (ipa_single_use): Likewise.
30677 * passes.c (execute_todo): Likewise.
30678 * predict.c (drop_profile): Likewise.
30679 * symtab.c (symtab_node::get_dump_name): New function.
30680 (symtab_node::dump_name): Likewise.
30681 (symtab_node::dump_asm_name): Likewise.
30682 (symtab_node::dump_references): Likewise.
30683 (symtab_node::dump_referring): Likewise.
30684 (symtab_node::dump_base): Likewise.
30685 (symtab_node::debug_symtab): Likewise.
30686 * tree-sra.c (convert_callers_for_node): Likewise.
30687 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
30688 * value-prof.c (init_node_map): Likewise.
30690 2017-05-23 Martin Liska <mliska@suse.cz>
30692 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
30693 and symtab_node::debug_symtab to symbol_table::debug.
30694 * cgraphunit.c (analyze_functions): Use the renamed function.
30695 (symbol_table::compile): Likewise.
30696 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
30697 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
30698 * passes.c (execute_todo): Likewise.
30699 * symtab.c (symbol_table::dump): New function.
30700 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
30702 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30704 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
30705 that nonconst implies exec.
30707 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30709 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30710 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30711 (inline_edge_summary_vec): Turn into ...
30712 (ipa_call_summaries): ... this one.
30713 (redirect_to_unreachable, edge_set_predicate,
30714 evaluate_properties_for_edge, inline_summary_alloc,
30715 reset_ipa_call_summary, reset_inline_summary,
30716 inline_summary_t::duplicate): Update.
30717 (inline_edge_duplication_hook): Turn to ...
30718 (ipa_call_summary_t::duplicate): ... this one.
30719 (inline_edge_removal_hook): Turn to ...
30720 (ipa_call_summary_t::remove): ... this one.
30721 (dump_inline_edge_summary): Turn to ...
30722 (dump_ipa_call_summary): ... this one.
30723 (estimate_function_body_sizes): Update.
30724 (inline_update_callee_summaries): Update.
30725 (remap_edge_change_prob): Update.
30726 (remap_edge_summaries): Update.
30727 (inline_merge_summary): Update.
30728 (do_estimate_edge_time): Update.
30729 (inline_generate_summary): Update.
30730 (inline_read_section): Update.
30731 (inline_read_summary): Update.
30732 (inline_free_summary): Update.
30733 * ipa-inline.c (can_inline_edge_p): Update.
30734 (compute_inlined_call_time): Update.
30735 (want_inline_small_function_p): Update.
30736 (edge_badness): Update.
30737 (early_inliner): Update.
30738 * ipa-inline.h (inline_edge_summary): Turn to ...
30739 (ipa_call_summary): ... this one.
30740 (ipa_call_summary_t): New class.
30741 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30742 (ipa_call_summaries): New.
30743 (inline_edge_summary): Remove.
30744 (estimate_edge_growth): Update.
30745 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30746 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30747 * ipa-split.c (execute_split_functions): Update.
30748 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30750 2017-05-23 Tom de Vries <tom@codesourcery.com>
30752 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30753 attributes): Document rdrand effective target.
30755 2017-05-23 Tom de Vries <tom@codesourcery.com>
30757 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30758 attributes): Sort alphabetically.
30760 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
30762 * config/avr/genmultilib.awk: Use gsub instead of gensub.
30764 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
30767 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
30768 V2DF/V2DI splat into two separate patterns, one that handles
30769 registers, and the other that only handles memory. Drop support
30770 for splatting from a GPR on ISA 2.07 and then splitting the
30771 splat into direct move and splat.
30772 (vsx_splat_<mode>_reg): Likewise.
30773 (vsx_splat_<mode>_mem): Likewise.
30775 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
30777 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
30779 2017-05-22 Jakub Jelinek <jakub@redhat.com>
30781 PR middle-end/80809
30782 * omp-low.c (finish_taskreg_remap): New function.
30783 (finish_taskreg_scan): If unit size of ctx->record_type
30784 is non-constant, unshare the size expression and replace
30785 decls in it with possible outer var refs.
30787 PR middle-end/80809
30788 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
30789 GOVD_SHARED rather than GOVD_PRIVATE with it.
30790 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30791 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30793 PR middle-end/80853
30794 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30795 as last argument to build_outer_var_ref for pointer bases of array
30796 section reductions.
30798 2017-05-19 Martin Sebor <msebor@redhat.com>
30800 * print-tree.c (print_node): Print DECL_READ_P flag.
30802 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30804 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30805 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30806 * cgraph.c: Likewise.
30807 * cgraphunit.c: Likewise.
30808 * gengtype.c: Likewise.
30809 * ipa-cp.c: Likewise.
30810 * ipa-devirt.c: Likewise.
30811 * ipa-icf.c: Likewise.
30812 * ipa-predicate.c: Likewise.
30813 * ipa-profile.c: Likewise.
30814 * ipa-prop.c: Likewise.
30815 * ipa-split.c: Likewise.
30817 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30818 edge_predicate_pool, dump_inline_hints,
30819 inline_summary::account_size_time, redirect_to_unreachable,
30820 edge_set_predicate, set_hint_predicate,
30821 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30822 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30823 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30824 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30825 ipa_call_summary_t::remove, initialize_growth_caches,
30826 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30827 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30828 mark_modified, unmodified_parm_1, unmodified_parm,
30829 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30830 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30831 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30832 will_be_nonconstant_predicate, record_modified_bb_info,
30833 get_minimal_bb, record_modified, param_change_prob,
30834 phi_result_unknown_predicate, predicate_for_phi_result,
30835 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30836 estimate_function_body_sizes, compute_inline_parameters,
30837 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30838 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30839 inline_update_callee_summaries, remap_edge_change_prob,
30840 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30841 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30842 inline_analyze_function, inline_summary_t::insert,
30843 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30844 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30845 inline_free_summary): Move to ipa-fnsummary.h
30846 (predicate_t): Remove.
30847 * ipa-fnsummary.c: New file.
30848 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30849 (enum inline_hints_vals, inline_hints, agg_position_info,
30850 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30851 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30852 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30853 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30854 inline_read_summary, inline_write_summary, inline_free_summary,
30855 inline_analyze_function, initialize_inline_failed,
30856 inline_merge_summary, inline_update_overall_summary,
30857 compute_inline_parameters): Move to ipa-fnsummary.h
30858 * ipa-fnsummary.h: New file.
30859 * ipa-inline-transform.h: Include ipa-inline.h.
30860 * ipa-inline.c: LIkewise.
30862 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30864 * ipa-inline.c (edge_badness): Use inlined_time instead of
30865 inline_summaries->get.
30867 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30869 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30871 2017-05-22 Nathan Sidwell <nathan@acm.org>
30873 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30874 (fdump-lang): Document 'raw' option.
30875 * dumpfile.h (TDI_tu): Delete.
30876 * dumpfile.c (dump_files): Remove translation-unit.
30877 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30879 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30881 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30882 command option from $(AWK) call.
30883 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30884 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30885 [FORMAT]: Remove handling of variable.
30886 * config/avr/t-multilib: Regenerate.
30888 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30890 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30892 (dump_inline_summary): Do not print self_time.
30893 (estimate_function_body_sizes): Do not set self_time.
30894 (compute_inline_parameters): Likewise.
30895 (inline_read_section, inline_write_summary): Do not stream self_time.
30896 * ipa-inline.h (inline_summary): Drop self_time.
30898 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30900 * ipa-inline-analysis.c (account_size_time): Rename to ...
30901 (inline_summary::account_size_time): ... this one.
30902 (reset_ipa_call_summary): Turn to ...
30903 (ipa_call_summary::reset): ... this one.
30904 (reset_inline_summary): Turn to ...
30905 (inline_summary::reset): ... this one.
30906 (inline_summary_t::remove): Update.
30907 (inline_summary_t::duplicate): Update.
30908 (ipa_call_summary_t::remove): Update.
30909 (dump_inline_summary): Update.
30910 (estimate_function_body_sizes): Update.
30911 (compute_inline_parameters): Update.
30912 (estimate_node_size_and_time): Update.
30913 (inline_merge_summary): Update.
30914 (inline_update_overall_summary): Update.
30915 (inline_read_section): Update.
30916 (inline_write_summary): Update.
30917 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30918 add account_size_time and reset member functions.
30919 (ipa_call_summary): Add reset function.
30920 * ipa-predicate.h (predicate::operator &): Constify.
30922 2017-05-22 Richard Biener <rguenther@suse.de>
30924 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30926 2017-05-19 Jason Merrill <jason@redhat.com>
30928 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30930 2017-05-19 Marek Polacek <polacek@redhat.com>
30933 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30934 TYPE_OVERFLOW_WRAPS checks.
30936 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30938 * tree-core.h (enum omp_clause_default_kind): Add
30939 "OMP_CLAUSE_DEFAULT_PRESENT".
30940 * tree-pretty-print.c (dump_omp_clause): Handle it.
30941 * gimplify.c (enum gimplify_omp_var_data): Add
30942 "GOVD_MAP_FORCE_PRESENT".
30943 (gimplify_adjust_omp_clauses_1): Map it to
30944 "GOMP_MAP_FORCE_PRESENT".
30945 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30947 * gimplify.c (oacc_default_clause): Clarify.
30949 2017-05-19 Nathan Sidwell <nathan@acm.org>
30951 LANG_HOOK_REGISTER_DUMPS
30952 * toplev.c (general_init): Call register dump lang hook.
30953 * doc/invoke.texi: Document -fdump-lang option family.
30954 * dumpfile.c (dump_files): Remove class dump here.
30955 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30956 * dumpfile.h (tree_dump_index): Remove TDI_class.
30957 * langhooks-def.h (lhd_register_dumps): Declare.
30958 (LANG_HOOKS_REGISTER_DUMPS): Define.
30959 (LANG_HOOKS_INITIALIZER): Add it.
30960 * langhooks.c (lhd_register_dumps): Define.
30961 * langhooks.h (struct lang_hooks): Add register_dumps.
30963 2017-05-19 Nathan Sidwell <nathan@acm.org>
30965 * context.h (context::set_passes): New.
30966 * context.c (context::context): Do not create pass manager.
30967 * toplev.c (general_init): Create pass manager here.
30969 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30971 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30972 use this splitter if two add or or instructions would also work for
30973 the constant we want to generate.
30975 2017-05-19 Richard Biener <rguenther@suse.de>
30978 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30979 predicate evaluation.
30981 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30983 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30985 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30988 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30990 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30991 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30992 (inline_edge_summary_vec): Turn into ...
30993 (ipa_call_summaries): ... this one.
30994 (redirect_to_unreachable, edge_set_predicate,
30995 evaluate_properties_for_edge, inline_summary_alloc,
30996 reset_ipa_call_summary, reset_inline_summary,
30997 inline_summary_t::duplicate): Update.
30998 (inline_edge_duplication_hook): Turn to ...
30999 (ipa_call_summary_t::duplicate): ... this one.
31000 (inline_edge_removal_hook): Turn to ...
31001 (ipa_call_summary_t::remove): ... this one.
31002 (dump_inline_edge_summary): Turn to ...
31003 (dump_ipa_call_summary): ... this one.
31004 (estimate_function_body_sizes): Update.
31005 (inline_update_callee_summaries): Update.
31006 (remap_edge_change_prob): Update.
31007 (remap_edge_summaries): Update.
31008 (inline_merge_summary): Update.
31009 (do_estimate_edge_time): Update.
31010 (inline_generate_summary): Update.
31011 (inline_read_section): Update.
31012 (inline_read_summary): Update.
31013 (inline_free_summary): Update.
31014 * ipa-inline.c (can_inline_edge_p): Update.
31015 (compute_inlined_call_time): Update.
31016 (want_inline_small_function_p): Update.
31017 (edge_badness): Update.
31018 (early_inliner): Update.
31019 * ipa-inline.h (inline_edge_summary): Turn to ...
31020 (ipa_call_summary): ... this one.
31021 (ipa_call_summary_t): New class.
31022 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
31023 (ipa_call_summaries): New.
31024 (inline_edge_summary): Remove.
31025 (estimate_edge_growth): Update.
31026 * ipa-profile.c (ipa_propagate_frequency_1): Update.
31027 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
31028 * ipa-split.c (execute_split_functions): Update.
31029 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
31031 2017-05-19 Richard Biener <rguenther@suse.de>
31033 PR middle-end/80764
31034 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
31036 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
31038 * config/rs6000/rs6000.c (struct machine_function): Add field
31039 fpr_is_wrapped_separately.
31040 (rs6000_get_separate_components): Use 64 components. Handle the
31041 new FPR components.
31042 (rs6000_components_for_bb): Handle the FPR components.
31043 (rs6000_emit_prologue_components): Handle the FPR components.
31044 (rs6000_emit_epilogue_components): Handle the FPR components.
31045 (rs6000_set_handled_components): Handle the FPR components.
31046 (rs6000_emit_prologue): Don't output prologue code for those FPRs
31047 that are already separately shrink-wrapped.
31048 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
31049 that are already separately shrink-wrapped.
31051 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31054 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
31057 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
31058 (define_peephole2 for Altivec d-form load): Add peepholes to catch
31059 cases where the register allocator uses a move and an offsettable
31060 memory operation to/from a FPR register on ISA 2.06/2.07.
31061 (define_peephole2 for Altivec d-form store): Likewise.
31063 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
31066 * config/i386/mmx.md (*mov<mode>_internal): Enable
31067 alternatives 11, 12, 13 and 14 also for 32bit targets.
31068 Remove alternatives 15, 16, 17 and 18.
31069 * config/i386/sse.md (vec_concatv2di): Change
31070 alternative (!x, *y) to (x, ?!*Yn).
31072 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
31074 * dumpfile.h (enum dump_kind): Remove stray comma.
31076 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
31078 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
31079 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
31080 predicate::num_conditions
31081 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
31082 (CHANGED): turn into predicate::changed.
31083 (agg_position_info): Move to ipa-predicate.h
31084 (add_condition, predicate::add_clause, predicate::operator &=,
31085 predicate::or_with, predicate::evaluate, predicate::probability,
31086 dump_condition, dump_clause, predicate::dump,
31087 predicate::remap_after_duplication, predicate::remap_after_inlining,
31088 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
31089 (evaluate_conditions_for_known_args): Update.
31090 (set_cond_stmt_execution_predicate): Update.
31091 * ipa-inline.h: Include ipa-predicate.h
31092 (condition, inline_param_summary, conditions, agg_position_info,
31093 predicate): Move to ipa-predicate.h
31094 * ipa-predicate.c: New file.
31095 * ipa-predicate.h: New file.
31097 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
31099 * final.c (leaf_function_p): Check we are not in a sequence.
31101 2017-05-18 Martin Liska <mliska@suse.cz>
31103 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
31104 * dumpfile.c (dump_register): Use new enum dump_kind.
31105 (get_dump_file_name): Likewise.
31106 (dump_enable_all): Likewise.
31107 (dump_switch_p_1): Likewise.
31108 (enable_rtl_dump_file): Remove usage of TDF_RTL.
31109 * dumpfile.h (enum dump_kind): New enum type.
31110 (struct dump_file_info): Create constructor and
31111 format fields and comments.
31112 * passes.c (pass_manager::register_one_dump_file):
31114 * statistics.c (statistics_early_init): Likewise.
31115 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
31116 TDF_TREE with TDF_SLIM.
31117 (gather_memory_references_ref): Likewise.
31119 2017-05-18 Martin Liska <mliska@suse.cz>
31121 * vec.h (struct vnull): Use it.
31123 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
31125 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
31126 (true_predicate, false_predicate, true_predicate_p,
31127 false_predicate_p): Remove.
31128 (single_cond_predicate, not_inlined_predicate): Turn to member function
31130 (add_condition): Update.
31131 (add_clause): Turn to...
31132 (predicate::add_clause): ... this one; update; allow passing NULL
31134 (and_predicates): Turn to ...
31135 (predicate::operator &=): ... this one.
31136 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
31137 (or_predicates): Turn to ...
31138 (predicate::or_with): ... this one.
31139 (evaluate_predicate): Turn to ...
31140 (predicate::evaluate): ... this one.
31141 (predicate_probability): Turn to ...
31142 (predicate::probability): ... this one.
31143 (dump_condition): Update.
31144 (dump_predicate): Turn to ...
31145 (predicate::dump): ... this one.
31146 (account_size_time): Update.
31147 (edge_set_predicate): Update.
31148 (set_hint_predicate): UPdate.
31149 (evaluate_conditions_for_known_args): Update.
31150 (evaluate_properties_for_edge): Update.
31151 (remap_predicate_after_duplication): Turn to...
31152 (predicate::remap_after_duplication): ... this one.
31153 (remap_hint_predicate_after_duplication): Update.
31154 (inline_summary_t::duplicate): UPdate.
31155 (dump_inline_edge_summary): Update.
31156 (dump_inline_summary): Update.
31157 (set_cond_stmt_execution_predicate): Update.
31158 (set_switch_stmt_execution_predicate): Update.
31159 (compute_bb_predicates): Update.
31160 (will_be_nonconstant_expr_predicate): Update.
31161 (will_be_nonconstant_predicate): Update.
31162 (phi_result_unknown_predicate): Update.
31163 (predicate_for_phi_result): Update.
31164 (array_index_predicate): Update.
31165 (estimate_function_body_sizes): Update.
31166 (estimate_node_size_and_time): Update.
31167 (estimate_ipcp_clone_size_and_time): Update.
31168 (remap_predicate): Rename to ...
31169 (predicate::remap_after_inlining): ... this one.
31170 (remap_hint_predicate): Update.
31171 (inline_merge_summary): Update.
31172 (inline_update_overall_summary): Update.
31173 (estimate_size_after_inlining): Update.
31174 (read_predicate): Rename to ...
31175 (predicate::stream_in): ... this one.
31176 (read_inline_edge_summary): Update.
31177 (write_predicate): Rename to ...
31178 (predicate::stream_out): ... this one.
31179 (write_inline_edge_summary): Update.
31180 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
31181 (clause_t): Turn to uint32_t
31182 (predicate): Turn to class; implement constructor and operators
31184 (size_time_entry): Update.
31185 (inline_summary): Update.
31186 (inline_edge_summary): Update.
31188 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
31190 * fold-const.c (fold_binary_loc): Move transformation...
31191 * match.pd (C - X CMP X): ... here.
31193 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
31195 * config/sparc/sparc.c (sparc_option_override): Set function
31196 alignment for -mcpu=niagara7 to 64 to match the I$ line.
31197 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
31199 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
31201 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
31203 2017-05-18 Marek Polacek <polacek@redhat.com>
31206 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
31207 (pass_ubsan::execute): Call gimple_assign_single_p instead of
31208 gimple_assign_load_p.
31210 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
31212 PR middle-end/80692
31213 * real.c (do_compare): Give decimal_do_compare preference over
31214 comparing just the signs.
31216 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
31218 * doc/md.texi (Canonicalization of Instructions): Describe the
31219 canonical form of instructions that inherently set a condition
31222 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
31224 PR middle-end/80775
31225 * tree-cfg.c: Move deletion of unreachable case statements to after
31226 the merging of consecutive case labels.
31228 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
31230 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
31231 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
31232 restoring of callee-saved registers.
31234 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
31236 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
31237 * config/visium/visium.c (single_set_and_flags): Likewise.
31238 * config/visium/visium.md (Substitutions): Likewise.
31240 2017-05-17 Martin Liska <mliska@suse.cz>
31242 * cfg.c: Introduce dump_flags_t type and
31243 use it instead of int type.
31245 * cfghooks.c: Likewise.
31246 * cfghooks.h (struct cfg_hooks): Likewise.
31247 * cfgrtl.c: Likewise.
31248 * cfgrtl.h: Likewise.
31249 * cgraph.c (cgraph_node::get_body): Likewise.
31250 * coretypes.h: Likewise.
31251 * domwalk.c: Likewise.
31252 * domwalk.h: Likewise.
31253 * dumpfile.c (struct dump_option_value_info): Likewise.
31254 (dump_enable_all): Likewise.
31255 (dump_switch_p_1): Likewise.
31256 (opt_info_switch_p): Likewise.
31257 * dumpfile.h (enum tree_dump_index): Likewise.
31258 (struct dump_file_info): Likewise.
31259 * genemit.c: Likewise.
31260 * generic-match-head.c: Likewise.
31261 * gengtype.c (open_base_files): Likewise.
31262 * gimple-pretty-print.c: Likewise.
31263 * gimple-pretty-print.h: Likewise.
31264 * graph.c (print_graph_cfg): Likewise.
31265 * graphite-scop-detection.c (dot_all_sese): Likewise.
31266 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
31267 * loop-unroll.c (report_unroll): Likewise.
31268 * passes.c (pass_manager::register_one_dump_file): Likewise.
31269 * print-tree.c: Likewise.
31270 * statistics.c: Likewise.
31271 * tree-cfg.c: Likewise.
31272 * tree-cfg.h: Likewise.
31273 * tree-dfa.c: Likewise.
31274 * tree-dfa.h: Likewise.
31275 * tree-dump.c (dump_function): Likewise.
31276 * tree-dump.h (struct dump_info): Likewise.
31277 * tree-pretty-print.c: Likewise.
31278 * tree-pretty-print.h: Likewise.
31279 * tree-ssa-live.c: Likewise.
31280 * tree-ssa-live.h: Likewise.
31281 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
31282 * tree-vect-loop.c: Likewise.
31283 * tree-vect-slp.c: Likewise.
31285 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
31286 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31288 PR tree-optimization/80457
31289 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
31290 of all arguments to a statement as scalar_to_vec operations.
31291 (vectorizable_call): Adjust call to vect_model_simple_cost for
31293 (vectorizable_conversion): Likewise.
31294 (vectorizable_assignment): Likewise.
31295 (vectorizable_shift): Likewise.
31296 (vectorizable_operation): Likewise.
31297 (vectorizable_comparison): Likewise.
31298 (vect_is_simple_cond): Record the def types for operands.
31299 (vectorizable_condition): Likewise, call vect_model_simple_cost.
31300 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
31301 for statement argument count.
31303 2017-05-16 Carl Love <cel@us.ibm.com>
31305 * config/rs6000/rs6000-c: Add support for built-in functions
31306 vector unsigned long long vec_bperm (vector unsigned long long,
31307 vector unsigned char)
31308 vector signed long long vec_mule (vector signed int,
31310 vector unsigned long long vec_mule (vector unsigned int,
31311 vector unsigned int)
31312 vector signed long long vec_mulo (vector signed int,
31314 vector unsigned long long vec_mulo (vector unsigned int,
31315 vector unsigned int)
31316 vector signed char vec_sldw (vector signed char,
31317 vector signed char,
31319 vector unsigned char vec_sldw (vector unsigned char,
31320 vector unsigned char,
31322 vector signed short vec_sldw (vector signed short,
31323 vector signed short,
31325 vector unsigned short vec_sldw (vector unsigned short,
31326 vector unsigned short,
31328 vector signed int vec_sldw (vector signed int,
31331 vector unsigned int vec_sldw (vector unsigned int,
31332 vector unsigned int,
31334 vector signed long long vec_sldw (vector signed long long,
31335 vector signed long long,
31337 vector unsigned long long vec_sldw (vector unsigned long long,
31338 vector unsigned long long,
31340 * config/rs6000/rs6000-c: Add support for built-in functions
31341 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
31342 * config/rs6000/altivec.h: Add defintion for vec_sldw.
31343 * doc/extend.texi: Update the built-in documentation for the
31344 new built-in functions.
31346 2017-05-16 Marek Polacek <polacek@redhat.com>
31350 * tree.c (save_expr): Don't fold the expression.
31352 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
31354 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
31355 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
31356 and (?*y,m). Update insn attributes.
31358 2017-05-16 Martin Liska <mliska@suse.cz>
31360 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
31361 flags argument of print_gimple_stmt, print_gimple_expr,
31362 print_generic_stmt and print_generic_expr.
31363 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
31364 * coretypes.h: Likewise.
31365 * except.c (dump_eh_tree): Likewise.
31366 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
31367 * gimple-pretty-print.h: Likewise.
31368 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
31369 (backprop::push_to_worklist): Likewise.
31370 (backprop::pop_from_worklist): Likewise.
31371 (backprop::process_use): Likewise.
31372 (backprop::intersect_uses): Likewise.
31373 (note_replacement): Likewise.
31374 * gimple-ssa-store-merging.c
31375 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
31376 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
31377 (pass_store_merging::execute): Likewise.
31378 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
31379 (ssa_base_cand_dump_callback): Likewise.
31380 (dump_incr_vec): Likewise.
31381 (replace_refs): Likewise.
31382 (replace_mult_candidate): Likewise.
31383 (create_add_on_incoming_edge): Likewise.
31384 (create_phi_basis): Likewise.
31385 (insert_initializers): Likewise.
31386 (all_phi_incrs_profitable): Likewise.
31387 (introduce_cast_before_cand): Likewise.
31388 (replace_one_candidate): Likewise.
31389 * gimplify.c (gimplify_expr): Likewise.
31390 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
31391 (set_rename): Likewise.
31392 (rename_uses): Likewise.
31393 (copy_loop_phi_nodes): Likewise.
31394 (add_close_phis_to_merge_points): Likewise.
31395 (copy_loop_close_phi_args): Likewise.
31396 (copy_cond_phi_args): Likewise.
31397 (graphite_copy_stmts_from_block): Likewise.
31398 (translate_pending_phi_nodes): Likewise.
31399 * graphite-poly.c (print_pdr): Likewise.
31400 (dump_gbb_cases): Likewise.
31401 (dump_gbb_conditions): Likewise.
31402 (print_scop_params): Likewise.
31403 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
31404 (build_cross_bb_scalars_use): Likewise.
31405 (gather_bbs::before_dom_children): Likewise.
31406 * hsa-dump.c (dump_hsa_immed): Likewise.
31407 * ipa-cp.c (print_ipcp_constant_value): Likewise.
31408 (get_replacement_map): Likewise.
31409 * ipa-inline-analysis.c (dump_condition): Likewise.
31410 (estimate_function_body_sizes): Likewise.
31411 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
31412 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
31413 * ipa-prop.c (ipa_dump_param): Likewise.
31414 (ipa_print_node_jump_functions_for_edge): Likewise.
31415 (ipa_modify_call_arguments): Likewise.
31416 (ipa_modify_expr): Likewise.
31417 (ipa_dump_param_adjustments): Likewise.
31418 (ipa_dump_agg_replacement_values): Likewise.
31419 (ipcp_modif_dom_walker::before_dom_children): Likewise.
31420 * ipa-pure-const.c (check_stmt): Likewise.
31421 (pass_nothrow::execute): Likewise.
31422 * ipa-split.c (execute_split_functions): Likewise.
31423 * omp-offload.c (dump_oacc_loop_part): Likewise.
31424 (dump_oacc_loop): Likewise.
31425 * trans-mem.c (tm_log_emit): Likewise.
31426 (tm_memopt_accumulate_memops): Likewise.
31427 (dump_tm_memopt_set): Likewise.
31428 (dump_tm_memopt_transform): Likewise.
31429 * tree-cfg.c (gimple_verify_flow_info): Likewise.
31430 (print_loop): Likewise.
31431 * tree-chkp-opt.c (chkp_print_addr): Likewise.
31432 (chkp_gather_checks_info): Likewise.
31433 (chkp_get_check_result): Likewise.
31434 (chkp_remove_check_if_pass): Likewise.
31435 (chkp_use_outer_bounds_if_possible): Likewise.
31436 (chkp_reduce_bounds_lifetime): Likewise.
31437 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
31438 (chkp_mark_completed_bounds): Likewise.
31439 (chkp_register_incomplete_bounds): Likewise.
31440 (chkp_mark_invalid_bounds): Likewise.
31441 (chkp_maybe_copy_and_register_bounds): Likewise.
31442 (chkp_build_returned_bound): Likewise.
31443 (chkp_get_bound_for_parm): Likewise.
31444 (chkp_build_bndldx): Likewise.
31445 (chkp_get_bounds_by_definition): Likewise.
31446 (chkp_generate_extern_var_bounds): Likewise.
31447 (chkp_get_bounds_for_decl_addr): Likewise.
31448 * tree-chrec.c (chrec_apply): Likewise.
31449 * tree-data-ref.c (dump_data_reference): Likewise.
31450 (dump_subscript): Likewise.
31451 (dump_data_dependence_relation): Likewise.
31452 (analyze_overlapping_iterations): Likewise.
31453 * tree-inline.c (expand_call_inline): Likewise.
31454 (tree_function_versioning): Likewise.
31455 * tree-into-ssa.c (dump_defs_stack): Likewise.
31456 (dump_currdefs): Likewise.
31457 (dump_names_replaced_by): Likewise.
31458 (dump_update_ssa): Likewise.
31459 (update_ssa): Likewise.
31460 * tree-object-size.c (pass_object_sizes::execute): Likewise.
31461 * tree-parloops.c (build_new_reduction): Likewise.
31462 (try_create_reduction_list): Likewise.
31463 (ref_conflicts_with_region): Likewise.
31464 (oacc_entry_exit_ok_1): Likewise.
31465 (oacc_entry_exit_single_gang): Likewise.
31466 * tree-pretty-print.h: Likewise.
31467 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
31468 (get_scalar_evolution): Likewise.
31469 (add_to_evolution): Likewise.
31470 (get_loop_exit_condition): Likewise.
31471 (analyze_evolution_in_loop): Likewise.
31472 (analyze_initial_condition): Likewise.
31473 (analyze_scalar_evolution): Likewise.
31474 (instantiate_scev): Likewise.
31475 (number_of_latch_executions): Likewise.
31476 (gather_chrec_stats): Likewise.
31477 (final_value_replacement_loop): Likewise.
31478 (scev_const_prop): Likewise.
31479 * tree-sra.c (dump_access): Likewise.
31480 (disqualify_candidate): Likewise.
31481 (create_access): Likewise.
31482 (reject): Likewise.
31483 (maybe_add_sra_candidate): Likewise.
31484 (create_access_replacement): Likewise.
31485 (analyze_access_subtree): Likewise.
31486 (analyze_all_variable_accesses): Likewise.
31487 (sra_modify_assign): Likewise.
31488 (initialize_constant_pool_replacements): Likewise.
31489 (find_param_candidates): Likewise.
31490 (decide_one_param_reduction): Likewise.
31491 (replace_removed_params_ssa_names): Likewise.
31492 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
31493 * tree-ssa-copy.c (dump_copy_of): Likewise.
31494 (copy_prop_visit_cond_stmt): Likewise.
31495 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
31496 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
31497 (record_equivalences_from_stmt): Likewise.
31498 * tree-ssa-dse.c (compute_trims): Likewise.
31499 (delete_dead_call): Likewise.
31500 (delete_dead_assignment): Likewise.
31501 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
31502 (forward_propagate_into_cond): Likewise.
31503 (pass_forwprop::execute): Likewise.
31504 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
31505 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
31507 (move_computations_worker): Likewise.
31508 (execute_sm): Likewise.
31509 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
31510 (remove_exits_and_undefined_stmts): Likewise.
31511 (remove_redundant_iv_tests): Likewise.
31512 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
31513 (adjust_iv_update_pos): Likewise.
31514 * tree-ssa-math-opts.c (bswap_replace): Likewise.
31515 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
31516 (value_replacement): Likewise.
31517 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
31518 * tree-ssa-pre.c (print_pre_expr): Likewise.
31519 (get_representative_for): Likewise.
31520 (create_expression_by_pieces): Likewise.
31521 (insert_into_preds_of_block): Likewise.
31522 (eliminate_insert): Likewise.
31523 (eliminate_dom_walker::before_dom_children): Likewise.
31524 (eliminate): Likewise.
31525 (remove_dead_inserted_code): Likewise.
31526 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
31527 * tree-ssa-reassoc.c (get_rank): Likewise.
31528 (eliminate_duplicate_pair): Likewise.
31529 (eliminate_plus_minus_pair): Likewise.
31530 (eliminate_not_pairs): Likewise.
31531 (undistribute_ops_list): Likewise.
31532 (eliminate_redundant_comparison): Likewise.
31533 (update_range_test): Likewise.
31534 (optimize_range_tests_var_bound): Likewise.
31535 (optimize_vec_cond_expr): Likewise.
31536 (rewrite_expr_tree): Likewise.
31537 (rewrite_expr_tree_parallel): Likewise.
31538 (linearize_expr): Likewise.
31539 (break_up_subtract): Likewise.
31540 (linearize_expr_tree): Likewise.
31541 (attempt_builtin_powi): Likewise.
31542 (attempt_builtin_copysign): Likewise.
31543 (transform_stmt_to_copy): Likewise.
31544 (transform_stmt_to_multiply): Likewise.
31545 (dump_ops_vector): Likewise.
31546 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
31547 (print_scc): Likewise.
31548 (set_ssa_val_to): Likewise.
31549 (visit_reference_op_store): Likewise.
31550 (visit_use): Likewise.
31551 (sccvn_dom_walker::before_dom_children): Likewise.
31552 (run_scc_vn): Likewise.
31553 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
31555 (expr_hash_elt::print): Likewise.
31556 (const_and_copies::pop_to_marker): Likewise.
31557 (const_and_copies::record_const_or_copy_raw): Likewise.
31558 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
31559 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
31560 (dump_predicates): Likewise.
31561 (find_uninit_use): Likewise.
31562 (warn_uninitialized_phi): Likewise.
31563 (pass_late_warn_uninitialized::execute): Likewise.
31564 * tree-ssa.c (verify_vssa): Likewise.
31565 (verify_ssa): Likewise.
31566 (maybe_optimize_var): Likewise.
31567 * tree-vrp.c (dump_value_range): Likewise.
31568 (dump_all_value_ranges): Likewise.
31569 (dump_asserts_for): Likewise.
31570 (register_edge_assert_for_2): Likewise.
31571 (vrp_visit_cond_stmt): Likewise.
31572 (vrp_visit_switch_stmt): Likewise.
31573 (vrp_visit_stmt): Likewise.
31574 (vrp_visit_phi_node): Likewise.
31575 (simplify_cond_using_ranges_1): Likewise.
31576 (fold_predicate_in): Likewise.
31577 (evrp_dom_walker::before_dom_children): Likewise.
31578 (evrp_dom_walker::push_value_range): Likewise.
31579 (evrp_dom_walker::pop_value_range): Likewise.
31580 (execute_early_vrp): Likewise.
31582 2017-05-16 Richard Biener <rguenther@suse.de>
31584 * dwarf2out.c (loc_list_from_tree_1): Do not create
31585 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
31587 2017-05-16 Richard Biener <rguenther@suse.de>
31589 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
31591 (note_variable_value_in_expr): If we resolved the decl ref
31592 do not push to the stack.
31594 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
31596 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
31597 operations in fast-math mode.
31598 (vaddq_f16): Likewise.
31599 (vmul_f16): Likewise.
31600 (vmulq_f16): Likewise.
31601 (vsub_f16): Likewise.
31602 (vsubq_f16): Likewise.
31603 * config/arm/neon.md (add<mode>3): New.
31605 (fma:<VH:mode>3): New. Also remove outdated comment.
31608 2017-05-16 Martin Liska <mliska@suse.cz>
31612 * ipa-devirt.c (warn_types_mismatch): Fix typo.
31613 (odr_types_equivalent_p): Likewise.
31615 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
31617 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
31619 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
31622 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
31623 non-interunit SSE move alternatives with '?'.
31624 (zero-extendsidi peephole2): New peephole to skip intermediate
31625 general register in SSE zero-extend sequence.
31627 2017-05-15 Jeff Law <law@redhat.com>
31629 * reorg.c (relax_delay_slots): Create a new variable to hold
31630 the temporary target rather than clobbering TARGET_LABEL.
31632 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
31633 missing argument to extract_bit_field call.
31634 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
31636 2017-05-15 Martin Liska <mliska@suse.cz>
31639 * gcc.c (process_command): Do not allow empty argument of -o option.
31641 2017-05-15 Renlin Li <renlin.li@arm.com>
31643 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
31644 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
31645 * config/aarch64/constraints.md (Usf): Add long call check.
31646 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
31647 (call_value): Likewise.
31648 (sibcall): Likewise.
31649 (sibcall_value): Likewise.
31651 (call_value_insn): New.
31652 (sibcall_insn): Update rtx pattern.
31653 (sibcall_value_insn): Likewise.
31654 (call_internal): Remove.
31655 (call_value_internal): Likewise.
31656 (sibcall_internal): Likewise.
31657 (sibcall_value_internal): Likewise.
31658 (call_reg): Likewise.
31659 (call_symbol): Likewise.
31660 (call_value_reg): Likewise.
31661 (call_value_symbol): Likewise.
31663 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
31666 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
31668 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
31670 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
31671 compatible with CCGOCmode and with CCZmode.
31673 2017-05-14 Martin Sebor <msebor@redhat.com>
31675 PR middle-end/77671
31676 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
31677 (gimple_fold_builtin_snprintf): Same.
31678 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
31679 (gimple_fold_builtin_snprintf): Same.
31680 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
31681 of character types.
31682 (is_call_safe): New function.
31683 (try_substitute_return_value): Call it.
31684 (try_simplify_call): New function.
31685 (pass_sprintf_length::handle_gimple_call): Call it.
31687 2017-05-14 Martin Sebor <msebor@redhat.com>
31689 PR middle-end/80669
31690 * builtins.c (expand_builtin_stpncpy): Simplify.
31692 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
31694 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
31695 * config/i386/i386.h
31696 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
31697 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
31698 (struct machine_function): Add new members call_ms2sysv,
31699 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
31700 (struct machine_frame_state): New fields sp_realigned and
31701 sp_realigned_offset.
31702 * config/i386/i386.c
31703 (enum xlogue_stub): New enum.
31704 (enum xlogue_stub_sets): New enum.
31705 (class xlogue_layout): New class.
31706 (struct ix86_frame): New fields stack_realign_allocate_offset,
31707 stack_realign_offset and outlined_save_offset. Modify comments to
31708 detail stack layout when using out-of-line stubs.
31709 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
31710 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
31711 -mcall-ms2sysv-xlogues.
31712 (stub_managed_regs): New static variable.
31713 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
31714 registers managed by out-of-line stub.
31715 (disable_call_ms2sysv_xlogues): New function.
31716 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
31717 m->call_ms2sysv when appropriate and compute frame layout for
31719 (sp_valid_at, fp_valid_at): New inline functions.
31720 (choose_basereg): New function.
31721 (choose_baseaddr): Add align parameter, use choose_basereg and modify
31723 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
31724 Use align parameter of choose_baseaddr to generated aligned SSE movs
31726 (pro_epilogue_adjust_stack): Modify to track
31727 machine_frame_state::sp_realigned.
31728 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
31729 (ix86_nsaved_sseregs): Likewise.
31730 (ix86_emit_save_regs): Likewise.
31731 (ix86_emit_save_regs_using_mov): Likewise.
31732 (ix86_emit_save_sse_regs_using_mov): Likewise.
31733 (get_scratch_register_on_entry): Likewise.
31734 (gen_frame_set): New function.
31735 (gen_frame_load): Likewise.
31736 (gen_frame_store): Likewise.
31737 (emit_outlined_ms2sysv_save): Likewise.
31738 (emit_outlined_ms2sysv_restore): Likewise.
31739 (ix86_expand_prologue): Modify stack re-alignment code and call
31740 emit_outlined_ms2sysv_save when appropriate.
31741 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
31742 parameter rtx_insn *insn, which allows the function to be used to only
31743 generate the notes.
31744 (ix86_expand_epilogue): Modify validity checks of frame and stack
31745 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
31746 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
31747 * config/i386/predicates.md
31748 (save_multiple): New predicate.
31749 (restore_multiple): Likewise.
31750 * config/i386/sse.md
31751 (save_multiple<mode>): New pattern.
31752 (save_multiple_realign<mode>): Likewise.
31753 (restore_multiple<mode>): Likewise.
31754 (restore_multiple_and_return<mode>): Likewise.
31755 (restore_multiple_leave_return<mode>): Likewise.
31756 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
31758 2017-05-14 Julia Koval <julia.koval@intel.com>
31760 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
31761 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
31762 (__builtin_ia32_xsetbv): New builtins.
31763 * config/i386/i386.c (ix86_expand_special_args_builtin):
31765 (ix86_expand_builtin): Special expand for new intrinsics.
31766 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
31767 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
31768 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
31770 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31772 * cfganal.c (inverted_post_order_compute): Change argument type
31774 * cfganal.h (inverted_post_order_compute): Adjust prototype.
31775 * df-core.c (rest_of_handle_df_initialize): Adjust.
31776 (rest_of_handle_df_finish): Likewise.
31777 (df_analyze_1): Likewise.
31778 (df_analyze): Likewise.
31779 (loop_inverted_post_order_compute): Change argument to be a vec *.
31780 (df_analyze_loop): Adjust.
31781 (df_get_n_blocks): Likewise.
31782 (df_get_postorder): Likewise.
31783 * df.h (struct df_d): Change field to be a vec.
31784 * lcm.c (compute_laterin): Adjust.
31785 (compute_available): Likewise.
31786 * lra-lives.c (lra_create_live_ranges_1): Likewise.
31787 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
31788 * tree-ssa-pre.c (compute_antic): Likewise.
31790 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31792 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31793 (depth_first_search::depth_first_search): Change structure init
31794 function to this constructor.
31795 (depth_first_search::add_bb): Rename function to this member.
31796 (depth_first_search::execute): Likewise.
31797 (flow_dfs_compute_reverse_finish): Adjust.
31799 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31801 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31802 (longest_simple_path): Likewise.
31803 * shrink-wrap.c (spread_components): Likewise.
31804 (disqualify_problematic_components): Likewise.
31805 (emit_common_heads_for_components): Likewise.
31806 (emit_common_tails_for_components): Likewise.
31807 (insert_prologue_epilogue_for_components): Likewise.
31809 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31811 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31814 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31816 * df-core.c (df_set_blocks): Start using auto_bitmap.
31817 (df_compact_blocks): Likewise.
31818 * df-problems.c (df_rd_confluence_n): Likewise.
31819 * df-scan.c (df_insn_rescan_all): Likewise.
31820 (df_process_deferred_rescans): Likewise.
31821 (df_update_entry_block_defs): Likewise.
31822 (df_update_exit_block_uses): Likewise.
31823 (df_entry_block_bitmap_verify): Likewise.
31824 (df_exit_block_bitmap_verify): Likewise.
31825 (df_scan_verify): Likewise.
31826 * lra-constraints.c (lra_constraints): Likewise.
31827 (undo_optional_reloads): Likewise.
31828 (lra_undo_inheritance): Likewise.
31829 * lra-remat.c (calculate_gen_cands): Likewise.
31830 (do_remat): Likewise.
31831 * lra-spills.c (assign_spill_hard_regs): Likewise.
31832 (spill_pseudos): Likewise.
31833 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31834 (bitmap_set_subtract_values): Likewise.
31836 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31838 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31839 management with auto_bitmap.
31840 (fix_inter_tick): Likewise.
31841 (fix_recovery_deps): Likewise.
31842 * ira.c (add_store_equivs): Likewise.
31843 (find_moveable_pseudos): Likewise.
31844 (split_live_ranges_for_shrink_wrap): Likewise.
31845 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31846 (rtx_reuse_manager::seen_def_p): Likewise.
31847 (rtx_reuse_manager::set_seen_def): Likewise.
31848 * print-rtl.h (class rtx_reuse_manager): Likewise.
31850 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31852 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31854 (migrate_btr_def): Likewise.
31855 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31856 * df-core.c (loop_post_order_compute): Likewise.
31857 (loop_inverted_post_order_compute): Likewise.
31858 * hsa-common.h: Likewise.
31859 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31860 * init-regs.c (initialize_uninitialized_regs): Likewise.
31861 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31862 (inline_small_functions): Likewise.
31863 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31864 * ira.c (combine_and_move_insns): Likewise.
31865 (build_insn_chain): Likewise.
31866 * loop-invariant.c (find_invariants): Likewise.
31867 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31868 * predict.c (tree_predict_by_opcode): Likewise.
31869 (predict_paths_leading_to): Likewise.
31870 (predict_paths_leading_to_edge): Likewise.
31871 (estimate_loops_at_level): Likewise.
31872 (estimate_loops): Likewise.
31873 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31874 (spread_components): Likewise.
31875 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31876 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31877 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31878 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31879 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31880 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31881 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31882 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31883 (mark_threaded_blocks): Likewise.
31884 (thread_through_all_blocks): Likewise.
31885 * tree-ssa.c (verify_ssa): Likewise.
31886 (execute_update_addresses_taken): Likewise.
31887 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31889 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31891 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31893 (post_order_compute): Likewise.
31894 (inverted_post_order_compute): Likewise.
31895 (pre_and_rev_post_order_compute_fn): Likewise.
31897 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31899 * genrecog.c (int_set::int_set): Explicitly construct our
31900 auto_vec base class.
31901 * vec.h (auto_vec::auto_vec): New constructor.
31903 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31905 * bitmap.h (class auto_bitmap): New constructor taking
31906 bitmap_obstack * argument.
31908 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31910 * bitmap.h (class auto_bitmap): Change type of m_bits to
31911 bitmap_head, and adjust ctor / dtor and member operators.
31913 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31915 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31916 when returned register mode doesn't match original mode.
31918 2017-05-12 Jeff Law <law@redhat.com>
31919 Jakub Jelinek <jakub@redhat.com>
31921 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31922 we look for cc setter after the compare-elim changes.
31923 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31924 within the vector to match what compare-elim now expects.
31925 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31926 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31928 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31929 after the compare-elim changes.
31930 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31931 the vector to match what compare-elim now expects.
31932 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31933 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31934 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31935 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31936 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31938 * config/visium/visium.c (single_set_and_flags): Fix where
31939 we look for cc setter after the compare-elim changes.
31940 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31941 with the vector to match what compare-elim now expects.
31942 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31943 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31944 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31945 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31946 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31947 (neg<mode>2_insn_set_overflow): Likewise.
31949 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31951 PR middle-end/79794
31952 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31953 maybe_expand_insn call, set ops[0].target. If still set after call,
31954 set alt_rtl. Add extra arg to recursive calls.
31955 (extract_bit_field): Add alt_rtl argument. Pass to
31957 * expmed.h (extract_bit_field): Fix prototype.
31958 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31959 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31960 to extract_bit_field_calls.
31961 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31962 Pass alt_rtl to extract_bit_field calls.
31963 * calls.c (store_unaligned_arguments_into_psuedos)
31964 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31965 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31967 * optabs.h (struct expand_operand): Add target bitfield.
31969 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31971 * compare-elim.c (try_eliminate_compare): Canonicalize
31972 operation with embedded compare to
31973 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31974 (set (reg) (operation)].
31976 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31978 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31981 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31982 cost of adding a carry flag for ADC instruction.
31983 [case MINUS]: Ignore the cost of subtracting a carry flag
31984 for SBB instruction.
31986 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31988 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31990 * config/rs6000/bmiintrin.h: New file.
31991 * config/rs6000/bmi2intrin.h: New file.
31992 * config/rs6000/x86intrin.h: New file.
31994 2017-05-12 Jeff Law <law@redhat.com>
31996 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31999 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
32001 PR middle-end/80707
32002 * tree-cfg.c: Remove cfg edges of unreachable case statements.
32004 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
32006 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
32007 early expansion of vector divide builtins.
32008 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
32009 builtins identified as having unsigned arguments.
32011 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
32013 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
32014 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
32015 expansion of vector logical operations (and, andc, or, xor,
32018 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
32020 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
32021 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
32023 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
32025 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
32026 early GIMPLE expansion of vector multiplies.
32028 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32030 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
32031 TARGET_HAVE_MOVT conditional.
32032 (movt splitter): Likewise.
32034 2017-05-12 Richard Biener <rguenther@suse.de>
32036 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
32037 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
32038 Fold all stmts not inplace.
32040 2017-05-12 Richard Biener <rguenther@suse.de>
32042 PR tree-optimization/80713
32043 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
32044 inserted_exprs bit for not removed stmts.
32046 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
32048 PR middle-end/69921
32049 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
32050 parallelized" attribute for parallelized OpenACC kernels.
32051 * omp-offload.c (execute_oacc_device_lower): Use it.
32053 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
32054 Set "oacc kernels" attribute.
32055 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
32056 parameter. Adjust all users.
32057 (oacc_fn_attrib_kernels_p): Remove function.
32058 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
32059 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
32060 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
32061 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
32062 assert "oacc kernels" attribute is set.
32064 2017-05-11 Carl Love <cel@us.ibm.com>
32066 * config/rs6000/rs6000-c: Add support for built-in functions
32067 vector unsigned char vec_popcnt (vector signed char)
32068 vector unsigned char vec_popcnt (vector unsigned char)
32069 vector unsigned short vec_popcnt (vector signed short)
32070 vector unsigned short vec_popcnt (vector unsigned short)
32071 vector unsigned int vec_popcnt (vector signed int)
32072 vector unsigned int vec_popcnt (vector unsigned int)
32073 vector unsigned long long vec_popcnt (vector signed long long)
32074 vector unsigned long long vec_popcnt (vector unsigned long long)
32075 vector signed long long vec_slo (vector signed long long,
32076 vector signed char)
32077 vector signed long long vec_slo (vector signed long long,
32078 vector unsigned char)
32079 vector unsigned long long vec_slo (vector unsigned long long,
32080 vector signed char)
32081 vector unsigned long long vec_slo (vector unsigned long long,
32082 vector unsigned char)
32083 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
32084 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
32085 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
32086 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
32087 * doc/extend.texi: Update the built-in documentation file for the
32088 new built-in functions.
32090 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
32092 * attribs.h (sorted_attr_string): Move machine independent
32093 functions for target clone support from the i386 port to common
32094 code. Rename ix86_function_versions to common_function_versions.
32095 Rename make_name to make_unique_name.
32096 (common_function_versions): Likewise.
32097 (make_unique_name): Likewise.
32098 (make_dispatcher_decl): Likewise.
32099 (is_function_default_version): Likewise.
32100 * attribs.c (attr_strcmp): Likewise.
32101 (sorted_attr_string): Likewise.
32102 (common_function_versions): Likewise.
32103 (make_unique_name): Likewise.
32104 (make_dispatcher_decl): Likewise.
32105 (is_function_default_version): Likewise.
32106 * config/i386/i386.c (attr_strcmp): Likewise.
32107 (sorted_attr_string): Likewise.
32108 (ix86_function_versions): Likewise.
32109 (make_name): Likewise.
32110 (make_dispatcher_decl): Likewise.
32111 (is_function_default_version): Likewise.
32112 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
32114 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32117 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
32118 Account for direct move costs for vec_construct of integer
32121 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
32124 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
32125 (UNSPEC_STX_ATOMIC): Ditto.
32126 (loaddi_via_sse): New insn.
32127 (storedi_via_sse): Ditto.
32128 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
32129 Update corresponding peephole2 patterns.
32130 (atomic_storedi_fpu): Ditto.
32132 2017-05-11 Julia Koval <julia.koval@intel.com>
32134 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
32135 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
32137 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
32138 (__builtin_ia32_rsqrt14ss_mask): New builtins.
32139 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
32141 2017-05-11 Nathan Sidwell <nathan@acm.org>
32143 * graphite-poly.c: Include dumpfile.h.
32145 * dumpfle.h (dump_function): Declare here ...
32146 * tree-dump.h (dump_function): ... not here.
32147 * dumpfile.c: #include tree-cfg.h.
32148 (dump_function): Move here from ...
32149 * tree-dump.c (dump_function): ... here.
32150 * gimplify.c: #include splay-tree.h, not tree-dump.h.
32151 * graphite-poly.c: Don't include tree-dump.h.
32152 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
32153 * print-tree.c: Likewise.
32154 * stor-layout.c: Likewise.
32155 * tree-nested.c: Likewise.
32157 * dumpfile.c (dump_start): Use TDF_FLAGS.
32158 (dump_enable_all): Fix TDF_KIND check thinko.
32160 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
32162 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32163 array entries to represent two legal parameterizations of the
32164 overloaded __builtin_cmpb function, as represented by the
32165 P6_OV_BUILTIN_CMPB constant.
32166 (altivec_resolve_overloaded_builtin): Add special case handling
32167 for the __builtin_cmpb function, as represented by the
32168 P6_OV_BUILTIN_CMPB constant.
32169 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
32170 (BU_P6_64BIT_2): New macro.
32171 (BU_P6_OVERLOAD_2): New macro
32172 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
32173 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
32174 (CMPB): Add overload support to represent both 32-bit and 64-bit
32175 compare-bytes function.
32176 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
32177 support for TARGET_CMPB.
32178 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
32179 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32180 documentation of the __builtin_cmpb overloaded built-in function.
32182 2017-05-11 Richard Biener <rguenther@suse.de>
32184 PR tree-optimization/80705
32185 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
32186 bases are not vectorizable.
32188 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32190 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
32191 when counting register pressure.
32193 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32195 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
32196 (iv_ca_more_deps): Renamed to ...
32197 (iv_ca_compare_deps): ... this.
32198 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
32200 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32202 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
32204 (determine_group_iv_costs): ... here.
32205 (find_inv_vars_cb): Record inv var if it's not recorded before.
32207 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32209 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
32210 (get_shiftadd_cost): Ditto.
32212 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32214 * tree-ssa-address.c: Include header file.
32215 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
32217 (add_to_parts): Refactor.
32218 (addr_to_parts): New parameter. Update use of move_hint_to_base.
32219 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
32222 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32224 PR tree-optimization/53090
32225 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
32227 (extract_cond_operands): Detect condition with IV on both sides
32228 and return COMP_IV_EXPR_2.
32229 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
32230 (rewrite_use_compare): Simplify by removing call to function
32231 extract_cond_operands.
32233 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32235 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
32236 (extract_cond_operands): Detect condition comparing against non-
32237 invariant bound and return appropriate enum value.
32238 (find_interesting_uses_cond): Update use of extract_cond_operands.
32239 Handle its return value accordingly.
32240 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
32242 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32244 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
32245 nonlinear iv_use computation in loop invariant sensitive way.
32247 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32249 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
32250 (find_iv_candidates): Call relate_compare_use_with_all_cands.
32252 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32254 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
32255 (dump_cand): Support iv_cand.inv_exprs.
32256 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
32258 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
32261 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32263 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
32265 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
32266 as local function. Include necessary header files.
32267 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
32269 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32271 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
32273 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32275 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
32276 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
32277 RSHIFT_EXPR and BIT_NOT_EXPR.
32279 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32281 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
32282 (adjust_setup_cost): New parameter supporting round up adjustment.
32283 (struct address_cost_data): Delete.
32284 (force_expr_to_var_cost): Don't bound cost with spill_cost.
32285 (split_address_cost, ptr_difference_cost): Delete.
32286 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
32287 (struct ainc_cost_data): New struct.
32288 (get_address_cost_ainc): New function.
32289 (get_address_cost, get_computation_cost): Reimplement.
32290 (determine_group_iv_cost_address): Record inv_expr for all uses of
32292 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
32293 (iv_ca_has_deps): Reimplemented to ...
32294 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
32296 (iv_ca_extend): Call iv_ca_more_deps.
32298 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32300 * tree-ssa-address.c (struct mem_address): Move to header file.
32301 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
32302 * tree-ssa-address.h (struct mem_address): Move from C file.
32303 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
32305 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32307 * tree-affine.h (aff_combination_type): New interface.
32308 (aff_combination_zero_p): Remove static.
32309 (aff_combination_const_p): New interface.
32310 (aff_combination_singleton_var_p): New interfaces.
32312 2017-05-11 Richard Biener <rguenther@suse.de>
32314 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
32315 Skip unreachable blocks and destinations.
32316 (eliminate): Move stmt removal and fixup ...
32317 (fini_eliminate): ... here. Skip inserted exprs.
32318 (pass_pre::execute): Move fini_pre after fini_eliminate.
32319 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
32320 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
32321 PRE to get rid of dead code that has invalid SSA form and
32322 split critical edges again.
32324 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32326 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32328 2017-05-11 Richard Biener <rguenther@suse.de>
32330 * passes.c (execute_function_todo): Verify loops if they are
32331 said to be up-to-date.
32332 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
32333 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
32335 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
32338 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
32339 handle calling assemble_external ourself.
32342 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
32343 modes with zero size. Enhance comment.
32345 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32347 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
32348 built-ins for vec_xl and vec_xst with short and char pointer
32351 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32353 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
32354 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
32355 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
32356 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
32357 (_mm_maskz_min_round_ss): New intrinsics.
32358 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
32359 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
32360 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
32361 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
32362 (__builtin_ia32_minss_mask_round): New builtins.
32363 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32364 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32365 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
32367 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
32368 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
32370 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
32373 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32375 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
32376 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
32377 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
32378 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
32379 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
32380 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
32381 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
32382 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32383 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32384 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
32385 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
32386 (__builtin_ia32_mulss_mask_round): New builtins.
32387 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32388 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32389 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
32391 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
32392 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32394 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32397 2017-05-10 Julia Koval <julia.koval@intel.com>
32399 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
32400 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
32401 (_mm256_setr_m128i): New intrinsics.
32403 2017-05-10 Julia Koval <julia.koval@intel.com>
32405 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
32406 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
32407 (_mm_maskz_rcp14_ss): New intrinsics.
32408 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
32409 (__builtin_ia32_rcp14ss_mask): New builtins.
32410 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
32412 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
32414 PR tree-optimization/51513
32415 * tree-cfg.c (gimple_seq_unreachable_p): New function.
32416 (assert_unreachable_fallthru_edge_p): Use it.
32417 (group_case_labels_stmt): Likewise.
32418 * tree-cfg.h: Prototype it.
32419 * stmt.c: Include cfghooks.h and tree-cfg.h.
32420 (emit_case_dispatch_table) <gap_label>: New local variable.
32421 Use it to fill dispatch table gaps.
32422 Test for default_label before updating probabilities.
32423 (expand_case) <default_label>: Remove unneeded initialization.
32424 Test for unreachable default case statement and remove its edge.
32425 Set default_label accordingly.
32426 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
32428 2017-05-10 Carl Love <cel@us.ibm.com>
32430 * config/rs6000/rs6000-c: Add support for built-in functions
32431 vector signed char vec_neg (vector signed char)
32432 vector signed short int vec_neg (vector short int)
32433 vector signed int vec_neg (vector signed int)
32434 vector signed long long vec_neg (vector signed long long)
32435 vector float vec_neg (vector float)
32436 vector double vec_neg (vector double)
32437 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
32439 * config/rs6000/altivec.h: Add define for vec_neg
32440 * doc/extend.texi: Update the built-in documentation for the
32441 new built-in functions.
32443 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32445 PR tree-optimization/77644
32446 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
32448 2017-05-10 Nathan Sidwell <nathan@acm.org>
32450 * dumpfile.h (TDI_lang_all): New.
32451 (TDF_KIND): New. Renumber others
32452 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
32454 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
32456 (get_dump_file_name): Adjust suffix generation.
32457 (dump_enable_all): Use TDF_KIND.
32458 * doc/invoke.texi (-fdump-lang-all): Document.
32460 * dumpfile.h: Tabify.
32462 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
32465 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
32466 Move member access before delete.
32468 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
32470 * tree-inline.c (expand_call_inline): Split block at stmt
32473 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
32476 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
32477 are now unused after splitting mov{sf,sd}_hardfloat.
32478 (f32_lr2): Likewise.
32479 (f32_lm): Likewise.
32480 (f32_lm2): Likewise.
32481 (f32_li): Likewise.
32482 (f32_li2): Likewise.
32483 (f32_lv): Likewise.
32484 (f32_sr): Likewise.
32485 (f32_sr2): Likewise.
32486 (f32_sm): Likewise.
32487 (f32_sm2): Likewise.
32488 (f32_si): Likewise.
32489 (f32_si2): Likewise.
32490 (f32_sv): Likewise.
32491 (f32_dm): Likewise.
32492 (f32_vsx): Likewise.
32493 (f32_av): Likewise.
32494 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
32495 For movsf, order stores so the VSX stores occur before the GPR
32496 store which encourages the register allocator to use a traditional
32497 FPR instead of a GPR. For movsd, order the stores so that the GPR
32498 store comes before the VSX stores to allow the power6 to work.
32499 This is due to the power6 not having a 32-bit integer store
32500 instruction from a FPR.
32501 (movsf_hardfloat): Likewise.
32502 (movsd_hardfloat): Likewise.
32504 2017-05-09 Martin Sebor <msebor@redhat.com>
32506 PR translation/80280
32507 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
32510 PR translation/80280
32511 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
32512 data member added in r247778.
32513 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
32515 2017-05-09 Nathan Sidwell <nathan@acm.org>
32517 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
32519 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
32522 2017-05-09 Marek Polacek <polacek@redhat.com>
32524 * doc/invoke.texi: Fix typo.
32526 2017-05-09 Richard Biener <rguenther@suse.de>
32528 * tree-vrp.c (vrp_val_is_max): Adjust comment.
32529 (vrp_val_is_min): Likewise.
32530 (set_value_range_to_value): Likewise.
32531 (set_value_range_to_nonnegative): Likewise.
32532 (gimple_assign_nonzero_p): Likewise.
32533 (gimple_stmt_nonzero_p): Likewise.
32534 (vrp_int_const_binop): Likewise. Remove unreachable case.
32535 (adjust_range_with_scev): Adjust comments.
32536 (compare_range_with_value): Likewise.
32537 (extract_range_from_phi_node): Likewise.
32538 (test_for_singularity): Likewise.
32540 2017-05-09 Richard Biener <rguenther@suse.de>
32542 * tree-vrp.c (get_single_symbol): Add assert that we don't
32543 get overflowed constants as invariant part.
32544 (compare_values_warnv): Add comment before the TREE_NO_WARNING
32545 checks. Use wi::cmp instead of recursing for integer constants.
32546 (compare_values): Just ignore whether we assumed undefined
32547 overflow instead of failing the compare.
32548 (extract_range_for_var_from_comparison_expr): Add comment before the
32549 TREE_NO_WARNING sets.
32550 (test_for_singularity): Likewise.
32551 (extract_range_from_comparison): Do not disable optimization
32552 when we assumed undefined overflow.
32553 (extract_range_basic): Remove init of unused var.
32555 2017-05-09 Richard Biener <rguenther@suse.de>
32557 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
32558 (extract_range_from_multiplicative_op_1): Adjust.
32559 (extract_range_from_binary_expr_1): Use int_const_binop.
32561 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
32564 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
32565 rs6000_store_data_bypass_p in seven define_bypass directives and
32566 in several comments.
32567 * config/rs6000/rs6000-protos.h: Add prototype for
32568 rs6000_store_data_bypass_p function.
32569 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
32570 function implements slightly different (rs6000-specific) semantics
32571 than store_data_bypass_p, returning false rather than aborting
32572 with assertion error when arguments do not satisfy the
32573 requirements of store data bypass.
32574 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
32575 rs6000_store_data_bypass_p.
32577 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
32579 * config/xtensa/xtensa-protos.h
32580 (xtensa_initial_elimination_offset): New declaration.
32581 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
32582 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
32583 macro definition, add case for FRAME_POINTER_REGNUM when
32584 FRAME_GROWS_DOWNWARD.
32585 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
32586 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
32587 xtensa_initial_elimination_offset.
32589 2017-05-08 Nathan Sidwell <nathan@acm.org>
32591 * doc/invoke.texi: Alphabetize -fdump options.
32593 2017-05-08 Martin Sebor <msebor@redhat.com>
32595 PR translation/80280
32596 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
32598 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
32600 * target.def (compute_frame_layout): New optional target hook.
32601 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
32602 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
32603 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
32605 * reload1.c (verify_initial_elim_offsets): Likewise.
32606 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
32607 (use_simple_return_p): Call arm_compute_frame_layout if needed.
32608 (arm_get_frame_offsets): Split up into this ...
32609 (arm_compute_frame_layout): ... and this function.
32611 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
32613 * config/aarch64/constraints.md (Usa): New constraint.
32614 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
32616 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
32618 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
32619 with_multilib_list after it has been checked.
32621 2017-05-08 Richard Biener <rguenther@suse.de>
32623 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
32624 (bitmap_set_subtract_values): Likewise.
32626 2017-05-08 Richard Biener <rguenther@suse.de>
32628 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
32629 (gimple_assign_nonzero): ... this and remove strict_overflow_p
32631 (gimple_stmt_nonzero_warnv_p): Rename to ...
32632 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
32634 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
32635 (extract_range_basic): Adjust, do not disable propagation on
32636 strict overflow sensitive simplification.
32637 (vrp_visit_cond_stmt): Likewise.
32639 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
32641 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
32642 body size unconditionally.
32644 2017-05-07 Jeff Law <law@redhat.com>
32647 2017-05-06 Jeff Law <law@redhat.com>
32648 PR tree-optimization/78496
32649 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32652 PR tree-optimization/78496
32653 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32654 (simplify_stmt_using_ranges): Call it.
32655 (vrp_dom_walker::before_dom_children): Extract equivalences
32656 from an ASSERT_EXPR with an equality comparison against a
32659 2017-05-06 Jeff Law <law@redhat.com>
32661 PR tree-optimization/78496
32662 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32665 PR tree-optimization/78496
32666 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32667 (simplify_stmt_using_ranges): Call it.
32668 (vrp_dom_walker::before_dom_children): Extract equivalences
32669 from an ASSERT_EXPR with an equality comparison against a
32672 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32674 * lra-constraints.c (lra_copy_reg_equiv): New function.
32675 (split_reg): Use it to copy equivalence information from the
32676 original register to the spill register.
32678 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32680 PR rtl-optimization/75964
32681 * simplify-rtx.c (simplify_const_relational_operation): Remove
32682 invalid handling of comparisons of integer ABS.
32684 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
32686 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
32687 initialize to zero.
32688 (init_regs): Remove declaration.
32689 (function_arg_advance_32): Initialize error_p as boolean variable.
32691 2017-05-05 Nathan Sidwell <nathan@acm.org>
32693 * store-motion.c (remove_reachable_equiv_notes): Reformat long
32694 lines. Use for (;;).
32696 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32698 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
32699 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
32700 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
32701 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
32702 VF=2 that require versioning.
32704 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32706 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
32709 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32711 * diagnostic.h (diagnostic_override_option_index): Convert from
32712 macro to inline function.
32714 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32716 * diagnostic.c (last_module_changed_p): New function.
32717 (set_last_module): New function.
32718 (diagnostic_report_current_module): Convert macro usage to
32719 the above functions.
32720 * diagnostic.h (diagnostic_context::last_module): Strengthen
32721 from const line_map * to const line_map_ordinary *.
32722 (diagnostic_last_module_changed): Delete macro.
32723 (diagnostic_set_last_module): Delete macro.
32725 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32727 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
32728 with diagnostic_report_diagnostic.
32729 (diagnostic_n_impl_richloc): Likewise.
32730 * diagnostic.h (report_diagnostic): Delete macro.
32731 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
32732 with diagnostic_report_diagnostic.
32733 * substring-locations.c (format_warning_va): Likewise.
32735 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32737 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
32738 save/restor of format_spec. Move option-printing code to...
32739 (print_option_information): ...this new function, and
32740 reimplement by simply printing to the pretty_printer,
32741 rather than appending to the format string.
32743 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32745 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
32746 handling logic into...
32747 (update_effective_level_from_pragmas): ...this new function.
32749 2017-05-04 Andrew Waterman <andrew@sifive.com>
32751 * config/riscv/riscv.opt (mstrict-align): New option.
32752 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
32753 (SLOW_UNALIGNED_ACCESS): Define.
32754 (riscv_slow_unaligned_access): Declare.
32755 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
32757 (riscv_slow_unaligned_access): New variable.
32758 (rocket_tune_info): Set slow_unaligned_access to true.
32759 (optimize_size_tune_info): Set slow_unaligned_access to false.
32760 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
32761 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
32762 (riscv_option_override): Set riscv_slow_unaligned_access.
32763 * doc/invoke.texi: Add -mstrict-align to RISC-V.
32765 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
32767 * config/riscv/riscv.md: Unify indentation.
32769 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
32774 * config/rs6000/rs6000.md (u code attribute): Add FIX and
32776 (extendsi<mode>2): Add support for doing sign extension via
32777 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
32778 don't have ISA 3.0 instructions.
32779 (extendsi<mode>2 splitter): Likewise.
32780 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
32781 generate the normal insns since SImode can now go in vector
32782 registers. Disallow the special UNSPECs needed for previous
32783 machines to hide SImode being used. Add new insns
32784 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
32785 (fix_trunc<mode>si2_stfiwx): Likewise.
32786 (fix_trunc<mode>si2_internal): Likewise.
32787 (fixuns_trunc<mode>si2): Likewise.
32788 (fixuns_trunc<mode>si2_stfiwx): Likewise.
32789 (fctiw<u>z_<mode>_smallint): Likewise.
32790 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32791 of floating point to 32-bit integer from doing a direct move to
32792 the GPR registers to do a store.
32793 (fctiwz_<mode>): Break long line.
32795 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32797 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32798 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32799 (addr_list, addr_offset_valid_p): New.
32800 (split_address_groups): Check offset validity with above function.
32801 (gt-tree-ssa-loop-ivopts.h): Include header file.
32803 2017-05-05 Nathan Sidwell <nathan@acm.org>
32805 * config.gcc (arm*-*-*): Add missing 'fi'.
32807 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32809 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32810 included in -fopt-info does not matter.
32811 * doc/optinfo.texi (-fopt-info): Fix description of default
32812 behavour. Explicitly say order of options included in -fopt-info
32815 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32817 * config.gcc: Allow combinations of aprofile and rmprofile values for
32818 --with-multilib-list.
32819 * config/arm/t-multilib: New file.
32820 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32821 variables. Remove setting of ISA and floating-point ABI in
32822 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32823 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32824 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32825 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32827 * config/arm/t-rmprofile: Likewise except for the matches changes.
32828 * doc/install.texi (--with-multilib-list): Document the combination of
32829 aprofile and rmprofile values and warn about pitfalls in doing that.
32831 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32833 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32834 (movdi_aarch64): Likewise.
32836 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32838 PR tree-optimization/80632
32839 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32841 (build_arrays): Initialize it for virtual phis.
32842 (fix_phi_nodes): Use it for virtual phis.
32844 PR tree-optimization/80558
32845 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32846 [x, y] op z into [x op, y op z] for op & or | if conditions
32849 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32850 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32853 * config/arm/arm.md (use_literal_pool): Remove.
32854 (64-bit immediate split): No longer takes cost into consideration
32855 if arm_disable_literal_pool is enabled.
32856 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32857 used when arm_disable_literal_pool is enabled.
32858 (arm_max_const_double_inline_cost): Remove use of
32859 arm_disable_literal_pool.
32860 (push_minipool_fix): Add assert.
32861 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32862 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32863 (no_literal_pool_sf_immediate): New.
32865 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32867 PR tree-optimization/80613
32868 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32869 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32871 2017-05-05 Richard Biener <rguenther@suse.de>
32873 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32875 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32877 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32878 of this flag from insn conditions due to removal from r247495.
32880 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32882 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32884 (arm_early_store_addr_dep_ptr): Likewise.
32885 * config/arm/aarch-common-protos.h
32886 (arm_early_load_addr_dep_ptr): Add prototype.
32887 (arm_early_store_addr_dep_ptr): Likewise.
32888 * config/arm/cortex-a53.md: Add new bypasses.
32890 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32892 * tree.c (next_type_uid): Change type to unsigned.
32893 (type_hash_canon): Decrement back next_type_uid if
32894 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32895 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32898 2017-05-04 Martin Sebor <msebor@redhat.com>
32900 * builtins.c: Fix a trivial typo in a comment.
32902 PR middle-end/79234
32903 * builtins.c (check_sizes): Adjust to handle reading past the end.
32904 Avoid printing excessive upper bound of ranges. Use %E to print
32905 tree nodes instead of converting them to %wu.
32906 (expand_builtin_memchr): New function.
32907 (compute_dest_size): Rename...
32908 (compute_objsize): ...to this.
32909 (expand_builtin_memcpy): Adjust.
32910 (expand_builtin_mempcpy): Adjust.
32911 (expand_builtin_strcat): Adjust.
32912 (expand_builtin_strcpy): Adjust.
32913 (check_strncat_sizes): Adjust.
32914 (expand_builtin_strncat): Adjust.
32915 (expand_builtin_strncpy): Adjust and simplify.
32916 (expand_builtin_memset): Adjust.
32917 (expand_builtin_bzero): Adjust.
32918 (expand_builtin_memcmp): Adjust.
32919 (expand_builtin): Handle memcmp.
32920 (maybe_emit_chk_warning): Check strncat just once.
32922 2017-05-04 Martin Sebor <msebor@redhat.com>
32924 PR preprocessor/79214
32925 PR middle-end/79222
32926 PR middle-end/79223
32927 * builtins.c (check_sizes): Add inlining context and issue
32928 warnings even when -Wno-system-headers is set.
32929 (check_strncat_sizes): Same.
32930 (expand_builtin_strncat): Same.
32931 (expand_builtin_memmove): New function.
32932 (expand_builtin_stpncpy): Same.
32933 (expand_builtin): Handle memmove and stpncpy.
32935 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32937 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32938 which is not used any more.
32940 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32942 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32944 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32946 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32947 (cortexa53_tunings): Likewise.
32948 (cortexa57_tunings): Likewise.
32949 (cortexa72_tunings): Likewise.
32950 (cortexa73_tunings): Likewise.
32952 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32954 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32955 Set loop alignment to 8.
32957 2017-05-04 Martin Sebor <msebor@redhat.com>
32959 PR translation/80280
32960 * builtins.c (expand_builtin_object_size): Add missing quoting to
32961 %D and like directives.
32962 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32963 (hsa_type_for_tree_type): Same.
32964 (verify_function_arguments): Same.
32965 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32966 * varasm.c (get_section): Same.
32969 2017-05-04 Martin Sebor <msebor@redhat.com>
32971 PR translation/80280
32972 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32974 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32976 * config/aarch64/aarch64.c (generic_addrcost_table):
32977 Change HI/TI mode setting.
32979 2017-05-04 Martin Jambor <mjambor@suse.cz>
32981 PR tree-optimization/80622
32982 * tree-sra.c (comes_initialized_p): New function.
32983 (build_accesses_from_assign): Only set write lazily when
32984 comes_initialized_p is false.
32985 (analyze_access_subtree): Use comes_initialized_p.
32986 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32987 instead of testing for PARM_DECL.
32989 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32991 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32992 constraint on operand 0 to allow more general addressing modes.
32993 Adjust output template.
32994 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32996 * config/aarch64/aarch64-protos.h
32997 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32998 * config/aarch64/constraints.md (Dp): New address constraint.
32999 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
33002 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
33004 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
33005 update use of estimate_ipcp_clone_size_and_time.
33006 (estimate_local_effects): Update use of
33007 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
33008 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
33009 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
33010 Return nonspecialized time.
33012 2017-05-04 Richard Biener <rguenther@suse.de>
33014 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
33015 for the last VUSE which def dominates the PHI. Directly call
33017 (get_continuation_for_phi_1): Remove.
33019 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
33021 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
33022 to explain the use of truncating division. Cap the number of
33023 iterations to the maximum given by nb_iterations_upper_bound,
33026 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33028 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
33029 * configure: Regenerate.
33030 * config.in: Regenerate.
33031 * config/i386/driver-mingw32.c: new file.
33032 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
33033 * config.host: Link driver-mingw32.o on MinGW host.
33034 * doc/install.texi: Document new --enable-mingw-wildcard configure
33037 2017-05-04 Marek Polacek <polacek@redhat.com>
33039 PR tree-optimization/80612
33040 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
33042 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
33043 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
33045 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
33046 (movt splitter): Likewise.
33047 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
33048 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
33049 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
33050 block for Thumb-1 with MOVT.
33051 (thumb2_legitimate_address_p): Move code block ...
33052 (can_avoid_literal_pool_for_label_p): ... into this new function.
33053 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
33055 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
33056 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
33057 "M-profile targets with the MOVT instruction".
33059 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
33061 * config/arm/arm-builtins.c (arm_init_builtins): Rename
33062 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
33063 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
33065 2017-05-04 Martin Liska <mliska@suse.cz>
33067 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
33068 variable cond_code.
33070 2017-05-04 Richard Biener <rguenther@suse.de>
33072 * tree.c (array_at_struct_end_p): Handle arrays at struct
33073 end with flexarrays more conservatively. Refactor and treat
33074 arrays of arrays or aggregates more strict. Fix
33075 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
33076 * tree.c (array_at_struct_end_p): Adjust prototype.
33077 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
33078 * gimple-fold.c (get_range_strlen): Likewise.
33079 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
33081 2017-05-04 Richard Biener <rguenther@suse.de>
33083 PR tree-optimization/31130
33084 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
33086 (supports_overflow_infinity): Likewise.
33087 (is_negative_overflow_infinity): Likewise.
33088 (is_positive_overflow_infinity): Likewise.
33089 (is_overflow_infinity): Likewise.
33090 (stmt_overflow_infinity): Likewise.
33091 (overflow_infinity_range_p): Likewise.
33092 (usable_range_p): Remove as always returning true.
33093 (make_overflow_infinity): Remove.
33094 (negative_overflow_infinity): Likewise.
33095 (positive_overflow_infinity): Likewise.
33096 (avoid_overflow_infinity): Likewise.
33097 (set_value_range): Adjust accordingly.
33098 (set_value_range_to_nonnegative): Likewise, remove now unused
33099 overflow_infinity arg.
33100 (vrp_operand_equal_p): Adjust.
33101 (update_value_range): Likewise.
33102 (range_int_cst_singleton_p): Likewise.
33103 (operand_less_p): Likewise.
33104 (compare_values_warnv): Likewise.
33105 (extract_range_for_var_from_comparison_expr): Likewise.
33106 (vrp_int_const_binop): Likewise.
33107 (zero_nonzero_bits_from_vr): Likewise.
33108 (extract_range_from_multiplicative_op_1): Likewise.
33109 (extract_range_from_binary_expr_1): Likewise.
33110 (extract_range_from_unary_expr): Likewise.
33111 (extract_range_from_comparison): Likewise.
33112 (extract_range_basic): Likewise.
33113 (adjust_range_with_scev): Likewise.
33114 (compare_ranges): Likewise.
33115 (compare_range_with_value): Likewise.
33116 (dump_value_range): Likewise.
33117 (test_for_singularity): Likewise, remove strict_overflow_p parameter
33119 (simplify_cond_using_ranges): Adjust.
33121 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
33123 * brig-builtins.def: Added a builtin for class_f64.
33124 * builtin-types.def: Added a builtin type needed by class_f64.
33126 2017-05-03 Jason Merrill <jason@redhat.com>
33128 * timevar.def: Add TV_CONSTEXPR.
33130 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33132 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
33134 2017-05-03 Martin Jambor <mjambor@suse.cz>
33136 * ipa-prop.c (ipa_update_after_lto_read): Removed.
33137 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
33138 * ipa-cp.c (ipcp_propagate_stage): Do not call
33139 ipa_update_after_lto_read.
33140 * ipa-inline.c (ipa_inline): Likewise.
33142 2017-05-03 Martin Jambor <mjambor@suse.cz>
33144 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
33145 tag. Added a default constructor and a destructor.
33146 (ipa_edge_args_sum_t): New class;
33147 (ipa_edge_args_sum): Declare.
33148 (ipa_edge_args_vector): Remove declaration.
33149 (IPA_EDGE_REF): Use ipa_edge_args_sum.
33150 (ipa_free_edge_args_substructures): Remove declaration.
33151 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
33152 (ipa_edge_args_info_available_for_edge_p): Likewise.
33153 * ipa-prop.c (ipa_edge_args_vector): Removed.
33154 (edge_removal_hook_holder): Likewise.
33155 (edge_duplication_hook_holder): Likewise.
33156 (ipa_edge_args_sum): New variable.
33157 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
33158 ipa_edge_args_vector.
33159 (ipa_free_edge_args_substructures): Likewise.
33160 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
33161 ipa_edge_args_vector.
33162 (ipa_edge_removal_hook): Turned into method
33163 ipa_edge_args_sum_t::remove.
33164 (ipa_edge_duplication_hook): Turned into method
33165 ipa_edge_args_sum_t::duplicate.
33166 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
33167 registering edge hooks.
33168 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
33169 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
33170 ipa_edge_args_sum instead of ipa_edge_args_vector.
33171 * ipa-profile.c (ipa_profile): Likewise.
33173 2017-05-03 Martin Jambor <mjambor@suse.cz>
33175 * symbol-summary.h (function_summary): New method exists.
33176 (function_summary::symtab_removal): Deallocate through release.
33177 (call_summary): New class.
33178 (gt_ggc_mx): New overload.
33179 (gt_pch_nx): Likewise.
33180 (gt_pch_nx): Likewise.
33182 2017-05-03 Jeff Law <law@redhat.com>
33184 PR tree-optimization/78496
33185 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
33186 from simplify_cond_using_ranges. Split off code to walk
33187 backwards through casts into ...
33188 (simplify_cond_using_ranges_2): New function.
33189 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
33190 (execute_vrp): After identifying jump threads, call
33191 simplify_cond_using_ranges_2.
33193 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
33196 * ipa-inline.h (inline_summary): Add ctor.
33197 (create_ggc): Do not use ggc_cleared_alloc.
33199 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
33200 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33202 * gcc.c (handle_braces): Support escaping in switch matching
33204 * doc/invoke.texi (Spec Files): Document it.
33205 Remove superfluous @code markup in items.
33207 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33209 * diagnostic-show-locus.c (struct column_range): New struct.
33210 (get_affected_columns): New function.
33211 (get_printed_columns): New function.
33212 (struct correction): New struct.
33213 (correction::ensure_capacity): New function.
33214 (correction::ensure_terminated): New function.
33215 (struct line_corrections): New struct.
33216 (line_corrections::~line_corrections): New dtor.
33217 (line_corrections::add_hint): New function.
33218 (layout::print_trailing_fixits): Reimplement in terms of the new
33220 (selftest::test_overlapped_fixit_printing): New function.
33221 (selftest::diagnostic_show_locus_c_tests): Call it.
33223 2017-05-03 Nathan Sidwell <nathan@acm.org>
33225 Canonicalize canonical type hashing
33226 * tree.h (type_hash_canon_hash): Declare.
33227 * tree.c (type_hash_list, attribute_hash_list): Move into
33228 type_hash_canon_hash.
33229 (build_type_attribute_qual_variant): Break out hash code calc into
33230 type_hash_canon_hash.
33231 (type_hash_canon_hash): New. Generic type hash computation.
33232 (build_range_type_1, build_array_type_1, build_function_type,
33233 build_method_type_directly, build_offset_type, build_complex_type,
33234 make_vector_type): Call it.
33236 2017-05-03 Richard Biener <rguenther@suse.de>
33238 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
33239 When all DRs have unknown misaligned do not always peel
33240 when there is a store but apply the same costing model as if
33241 there were only loads.
33243 2017-05-03 Richard Biener <rguenther@suse.de>
33246 PR tree-optimization/80492
33247 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33248 compare_base_decls returning dont-know properly.
33250 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
33252 * config/arm/iterators.md (CCSI): New mode iterator.
33253 (arch): New mode attribute.
33254 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
33255 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
33256 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
33257 code iterator for success result mode.
33258 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
33259 the corresponding new insn generators.
33261 2017-05-03 Bin Cheng <bin.cheng@arm.com>
33264 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33265 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33267 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
33269 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
33270 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
33271 (DDR_A): Wrap DDR argument in brackets.
33272 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
33273 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
33274 (DDR_REVERSED_P): Likewise.
33276 2017-05-03 Jakub Jelinek <jakub@redhat.com>
33278 PR tree-optimization/79472
33279 * tree-switch-conversion.c (struct switch_conv_info): Add
33280 contiguous_range and default_case_nonstandard fields.
33281 (collect_switch_conv_info): Compute contiguous_range and
33282 default_case_nonstandard fields, don't clear final_bb if
33283 contiguous_range and only the default case doesn't have the required
33285 (check_all_empty_except_final): Set default_case_nonstandard instead
33286 of failing if contiguous_range and the default case doesn't have empty
33288 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
33289 and only the default case doesn't have the required constants. Skip
33291 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
33292 if default_case_nonstandard.
33293 (build_constructors): Build constant 1 just once. Assert that default
33294 values aren't inserted in between cases if contiguous_range. Skip
33296 (build_arrays): Skip virtual phis.
33297 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
33298 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
33299 Handle virtual phis.
33300 (gen_inbound_check): Handle default_case_nonstandard case.
33301 (process_switch): Adjust check_final_bb caller. Call
33302 gather_default_values with the first non-default case instead of
33303 default case if default_case_nonstandard.
33305 2017-05-02 Nathan Sidwell <nathan@acm.org>
33307 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
33308 check. Fix formatting.
33310 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
33312 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
33313 errors when comparing specialized and unspecialized times.
33315 2017-05-02 David Malcolm <dmalcolm@redhat.com>
33317 * diagnostic-show-locus.c
33318 (layout::should_print_annotation_line_p): Make private.
33319 (layout::print_annotation_line): Make private.
33320 (layout::annotation_line_showed_range_p): Make private.
33321 (layout::show_ruler): Make private.
33322 (layout::print_source_line): Make private. Pass in line and
33323 line_width, rather than calling location_get_source_line. Drop
33325 (layout::print_leading_fixits): New method.
33326 (layout::print_any_fixits): Rename to...
33327 (layout::print_trailing_fixits): ...this, and make private.
33328 Don't print newline fixits.
33329 (diagnostic_show_locus): Move logic for printing one row into...
33330 (layout::print_line): ...this new function. Move the
33331 location_get_source_line call and error-handling from
33332 print_source_line to here. Call print_leading_fixits, and rename
33333 print_any_fixits to print_trailing_fixits.
33334 (selftest::test_fixit_insert_containing_newline): Update now that
33335 newlines are partially supported.
33336 (selftest::test_fixit_insert_containing_newline_2): New test.
33337 (selftest::test_fixit_replace_containing_newline): Update comments.
33338 (selftest::diagnostic_show_locus_c_tests): Call the new test.
33339 * edit-context.c (class added_line): New class.
33340 (class edited_line): Describe newline handling in comment.
33341 (edited_line::actually_edited_p): New method.
33342 (edited_line::print_content): Delete redundant decl.
33343 (edited_line::m_predecessors): New field.
33344 (edited_file::print_content): Call edited_line::print_content.
33345 (edited_file::print_diff): Update to support newlines.
33346 (edited_file::print_diff_hunk): Likewise.
33347 (edited_file::print_run_of_changed_lines): New function.
33348 (edited_file::print_diff_line): Convert to...
33349 (print_diff_line): ...this.
33350 (edited_file::get_effective_line_count): New function.
33351 (edited_line::edited_line): Initialize new field m_predecessors.
33352 (edited_line::~edited_line): Clean up m_predecessors.
33353 (edited_line::apply_fixit): Handle newlines.
33354 (edited_line::get_effective_line_count): New function.
33355 (edited_line::print_content): New function.
33356 (edited_line::print_diff_lines): New function.
33357 (selftest::test_applying_fixits_insert_containing_newline): New
33359 (selftest::test_applying_fixits_replace_containing_newline): New
33361 (selftest::insert_line): New function.
33362 (selftest::test_applying_fixits_multiple_lines): Add example of
33364 (selftest::edit_context_c_tests): Call the new tests.
33366 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33368 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
33369 parameter cand. Update dump information.
33370 (get_computation_cost): Update uses.
33372 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33374 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
33375 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
33376 (get_computation_at, rewrite_use_address): Update use of
33377 get_computation_aff.
33379 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33381 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
33382 (get_computation): Delete.
33383 (get_computation_cost): Implement like get_computation_cost_at.
33384 Use get_computation_at.
33385 (get_computation_cost_at): Delete.
33386 (rewrite_use_nonlinear_expr): Use get_computation_at.
33387 (rewrite_use_compare, remove_unused_ivs): Ditto.
33389 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33391 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
33393 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33395 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
33396 (ivopts_global_cost_for_size): Rename parameter and update uses.
33397 (iv_ca_recount_cost): Update uses.
33398 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
33399 candidates seperately in n_invs and n_cands.
33400 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
33402 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33404 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
33405 (find_inv_vars_cb): New.
33406 (find_depends): Renamed to ...
33407 (find_inv_vars): ... this.
33408 (add_candidate_1, force_var_cost): Call find_inv_vars.
33409 (split_address_cost, determine_group_iv_cost_cond): Ditto.
33411 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33413 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
33414 inv_vars. Add inv_exprs.
33415 (struct iv_cand): Rename depends_on to inv_vars.
33416 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
33417 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
33418 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
33419 (dump_cand): Dump inv_vars.
33420 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
33421 (record_invariant, find_depends, add_candidate_1): Ditto.
33422 (set_group_iv_cost, force_var_cost): Ditto.
33423 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
33424 (get_computation_cost_at, get_computation_cost): Ditto.
33425 (determine_group_iv_cost_generic): Ditto.
33426 (determine_group_iv_cost_address): Ditto.
33427 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
33428 (determine_group_iv_costs): Ditto.
33429 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
33430 (iv_ca_set_remove_invariants): Renamed to ...
33431 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
33432 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
33433 (iv_ca_set_add_invariants): Renamed to ...
33434 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
33435 (iv_ca_set_cp): Use iv_ca_set_add_invs.
33436 (iv_ca_has_deps): Support inv_vars and inv_exprs.
33437 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
33438 (create_new_ivs): Remove useless dump.
33440 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33442 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
33444 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
33445 (iv_ca_set_no_cp, create_new_iv): Ditto.
33447 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33449 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33451 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33453 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
33454 function tree_check2.
33456 2017-05-02 Martin Liska <mliska@suse.cz>
33458 * doc/gcov.texi: Add missing preposition.
33459 * gcov.c (function_info::function_info): Properly fill up
33460 all member variables.
33462 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33464 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
33466 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33468 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
33470 2017-05-02 Martin Liska <mliska@suse.cz>
33473 * lto-streamer-in.c (lto_read_tree_1): Remove
33474 LTO_STREAMER_DEBUG.
33475 * lto-streamer.c (struct tree_hash_entry): Likewise.
33476 (struct tree_entry_hasher): Likewise.
33477 (tree_entry_hasher::hash): Likewise.
33478 (tree_entry_hasher::equal): Likewise.
33479 (lto_streamer_init): Likewise.
33480 (lto_orig_address_map): Likewise.
33481 (lto_orig_address_get): Likewise.
33482 (lto_orig_address_remove): Likewise.
33483 * lto-streamer.h: Likewise.
33484 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
33485 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
33487 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
33489 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
33490 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
33491 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
33492 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
33493 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
33494 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
33495 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
33496 (mm_maskz_sub_ss): New intrinsics.
33497 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33498 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
33499 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
33500 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
33501 (__builtin_ia32_subss_mask_round): New builtins.
33502 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33503 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
33504 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
33506 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
33507 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
33509 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
33512 2017-05-02 Martin Jambor <mjambor@suse.cz>
33514 PR tree-optimization/78687
33515 * tree-sra.c (access): New field parent.
33516 (process_subtree_disqualification): New function.
33517 (disqualify_candidate): Call it.
33518 (build_accesses_from_assign): Reset write flag if creating an
33520 (build_access_subtree): Fill in parent field and also prpagate
33521 down grp_write flag.
33522 (create_artificial_child_access): New parameter set_grp_write, set
33523 grp_write to its value.
33524 (propagate_subaccesses_across_link): Also propagate grp_write flag
33526 (propagate_all_subaccesses): Push the closest parent back to work
33527 queue if add_access_to_work_queue returned true.
33529 2017-05-02 Richard Biener <rguenther@suse.de>
33531 * common.opt (fstrict-overflow): Alias negative to fwrapv.
33532 * doc/invoke.texi (fstrict-overflow): Remove all traces of
33533 -fstrict-overflow documentation.
33534 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
33535 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
33536 flag_strict_overflow.
33537 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
33538 * lto-opts.c (lto_write_options): Do not stream it.
33539 * lto-wrapper.c (merge_and_complain): Do not handle it.
33540 * opts.c (default_options_table): Do not set -fstrict-overflow.
33541 (finish_options): Likewise do not clear it when sanitizing.
33542 * simplify-rtx.c (simplify_const_relational_operation): Do not
33543 test flag_strict_overflow.
33545 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33547 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
33548 using enabled attribute.
33549 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
33550 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
33551 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
33552 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
33553 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
33554 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
33555 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
33556 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
33557 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
33558 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
33560 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33562 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
33564 2017-05-02 Richard Biener <rguenther@suse.de>
33566 PR tree-optimization/80591
33568 2017-04-10 Richard Biener <rguenther@suse.de>
33570 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33573 2017-05-02 Richard Biener <rguenther@suse.de>
33575 PR tree-optimization/80549
33576 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
33577 (cleanup_tree_cfg_noloop): Create forwarders to known loop
33578 headers if they do not have a preheader.
33580 2017-05-02 Martin Liska <mliska@suse.cz>
33583 * common.opt: Fix typo.
33584 * doc/invoke.texi: Likewise.
33586 2017-05-01 Jan Beulich <jbeulich@suse.com>
33588 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
33589 swapping, add (x,x,m,x,n) alternative.
33591 2017-05-01 Nathan Sidwell <nathan@acm.org>
33593 * calls.c (combine_pending_stack_adjustment_and_call): Remove
33594 unnecessary unadjusted_alignment check.
33596 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33599 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
33601 * gimplify.c (gimplify_cilk_detach): New function.
33602 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
33603 * tree-core.h: Document EXPR_CILK_SPAWN.
33604 * tree.h (EXPR_CILK_SPAWN): Define.
33606 2017-05-01 David Malcolm <dmalcolm@redhat.com>
33608 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
33609 to use new fixit_hint representation, using the "replace" logic.
33610 (get_line_span_for_fixit_hint): Likewise.
33611 (layout::print_any_fixits): Likewise.
33612 (selftest::test_one_liner_many_fixits): Rename to...
33613 (selftest::test_one_liner_many_fixits_1): ...this, and update
33614 comment and expected output to reflect that the multiple fix-it
33615 hints are now consolidated into one insertion.
33616 (selftest::test_one_liner_many_fixits_2): New test.
33617 (selftest::test_diagnostic_show_locus_one_liner): Update for
33619 (selftest::test_fixit_consolidation): Update for fix-it API
33621 * diagnostic.c (print_parseable_fixits): Likewise.
33622 * edit-context.c (edited_line::m_line_events): Convert from
33623 auto_vec <line_event *> to auto_vec <line_event>.
33624 (class line_event): Convert from abstract base class to a concrete
33625 class, taking over the role of replace_event.
33626 (class insert_event): Delete.
33627 (class replace_event): Rename to class line_event. Convert to
33629 (edit_context::add_fixits): Reimplement.
33630 (edit_context::apply_insert): Delete.
33631 (edit_context::apply_replace): Rename to...
33632 (edit_context::apply_fixit): ...this. Convert to half-open range.
33633 (edited_file::apply_insert): Delete.
33634 (edited_file::apply_replace): Rename to...
33635 (edited_file::apply_fixit): ...this.
33636 (edited_line::~edited_line): Drop deletion of events.
33637 (edited_line::apply_insert): Delete.
33638 (edited_line::apply_replace): Rename to...
33639 (edited_line::apply_fixit): ...this. Convert to half-open range.
33640 Update for change to type of m_line_events.
33641 * edit-context.h (edit_context::apply_insert): Delete.
33642 (edit_context::apply_replace): Rename to...
33643 (edit_context::apply_fixit): ...this.
33645 2017-05-01 Martin Sebor <msebor@redhat.com>
33647 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
33650 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
33653 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
33654 __get_cpuid_max returns 0.
33655 (__get_cpuid_count): Ditto.
33657 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
33659 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
33660 replacement expression is another instance of one of its arguments.
33662 2017-05-01 Jakub Jelinek <jakub@redhat.com>
33665 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
33666 check for stack push/pop autoinc.
33667 * config/i386/i386.c (ix86_agi_dependent): Return false
33668 if the only reason why modified_in_p returned true is that
33669 addr is SP based and set_insn is a push or pop.
33671 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33673 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
33676 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33679 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
33680 (account_size_time): Use two predicates - exec_pred and
33682 (evaluate_conditions_for_known_args): Compute both clause and
33684 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
33685 (inline_summary_t::duplicate): Update.
33686 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
33688 (compute_inline_parameters): Likewise.
33689 (estimate_edge_size_and_time): Update caluclation of time.
33690 (estimate_node_size_and_time): Compute both time and nonspecialized
33692 (estimate_ipcp_clone_size_and_time): Update.
33693 (inline_merge_summary): Update.
33694 (do_estimate_edge_time): Update.
33695 (do_estimate_edge_size): Update.
33696 (do_estimate_edge_hints): Update.
33697 (inline_read_section, inline_write_summary): Stream both new predicates.
33698 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
33700 (compute_inlined_call_time): Cleanup.
33701 (big_speedup_p): Update.
33702 (edge_badness): Update.
33703 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
33704 (size_time_entry): Replace predicate by exec_predicate and
33705 nonconst_predicate.
33706 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
33707 (estimate_edge_time): Return also nonspec_time.
33708 (reset_edge_growth_cache): Update.
33710 2017-04-29 Jakub Jelinek <jakub@redhat.com>
33712 PR rtl-optimization/80491
33713 * ifcvt.c (noce_process_if_block): When looking for x setter
33714 with missing else_bb, don't check only the insn right before
33715 cond_earliest, but look for the last insn that x is modified in
33716 within the same bb.
33718 PR rtl-optimization/80491
33719 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
33721 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
33723 PR tree-optimization/80487
33724 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
33726 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33728 PR tree-optimization/79697
33729 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
33730 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
33731 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
33733 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
33734 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
33736 2017-04-28 Martin Sebor <msebor@redhat.com>
33738 PR tree-optimization/80523
33739 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
33740 (init_target_to_host_charmap, target_to_host, target_strtol10): New
33742 (maybe_warn, format_directive, parse_directive): Use new functions.
33743 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
33745 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
33747 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
33749 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
33751 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
33752 target_header_dir): Set correctly.
33753 * configure: Regenerated.
33754 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33755 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33756 instead of SYSTEM_HEADER_DIR.
33758 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
33760 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
33761 (estimate_local_effects): Likewise.
33762 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
33763 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
33764 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
33765 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
33766 do_estimate_edge_time, estimate_edge_time): Likewise.
33767 * ipa-inline-analysis.c (estimate_node_size_and_time,
33768 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
33769 (estimate_time_after_inlining): Remove.
33771 2017-04-28 Martin Liska <mliska@suse.cz>
33773 * doc/gcov.texi: Enhance documentation of gcov.
33775 2017-04-28 Martin Liska <mliska@suse.cz>
33777 * doc/gcov.texi: Sort options in alphabetic order.
33778 * doc/gcov-dump.texi: Likewise.
33779 * doc/gcov-tool.texi: Likewise.
33780 * gcov.c (print_usage): Likewise.
33781 * gcov-dump.c (print_usage): Likewise.
33782 * gcov-tool.c (print_merge_usage_message): Likewise.
33783 (print_rewrite_usage_message): Likewise.
33784 (print_overlap_usage_message): Likewise.
33786 2017-04-28 Martin Liska <mliska@suse.cz>
33788 PR gcov-profile/53915
33789 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33791 2017-04-28 Martin Liska <mliska@suse.cz>
33793 PR gcov-profile/79891
33794 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33795 is marked by compiler as living on a line.
33796 (get_cycles_count): Remove usage of the union.
33797 (output_intermediate_file): Likewise.
33798 (find_source): Fix GNU coding style.
33799 (accumulate_line_counts): Remove old non-all block mode.
33800 (output_lines): Remove usage of the union.
33801 * profile.c (output_location): Include all BBs, even if
33802 belonging to a same line (and file) as a previous BB.
33804 2017-04-28 Martin Liska <mliska@suse.cz>
33806 * gcov.c (process_args): Handle new argument 'w'.
33807 (read_graph_file): Assign ID to BBs.
33808 (output_branch_count): Display BB # if verbose flag is set.
33809 (output_lines): Likewise for arcs.
33810 (print_usage): Add '--verbose' option help.
33811 * doc/gcov.texi: Document --verbose (-w) option.
33813 2017-04-28 Martin Liska <mliska@suse.cz>
33815 * gcov.c (struct block_location_info): New struct.
33816 (process_file): Fill up the new structure.
33817 (read_graph_file): Replace usage of encoding by the newly added
33819 (add_line_counts): Likewise.
33820 (accumulate_line_counts): Remove usage of the union.
33821 (function_info::function_info): New function.
33822 (function_info::~function_info): Likewise.
33823 (process_file): Call delete instead of release_function.
33824 (release_function): Release the function.
33825 (release_structures): Call delete instead of release_function.
33826 (solve_flow_graph): Replace usage of num_blocks.
33827 (find_exception_blocks): Likewise.
33828 (output_lines): Fix GNU coding style.
33830 2017-04-28 Martin Liska <mliska@suse.cz>
33833 * coverage.c (coverage_remove_note_file): New function.
33834 * coverage.h: Declare the function.
33835 * toplev.c (finalize): Clean if an error has been seen.
33837 2017-04-28 Martin Liska <mliska@suse.cz>
33839 PR gcov-profile/80031
33840 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33841 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33842 * gcov.c (read_graph_file): Read just number of blocks.
33843 * profile.c (branch_prob): Do not stream 0 flags per a basic
33846 2017-04-28 Martin Liska <mliska@suse.cz>
33848 * gcov-dump.c (tag_*): Add new argument to declarations.
33849 (dump_gcov_file): Likewise.
33850 (tag_blocks): Add and use new argument depth.
33851 (tag_arcs): Likewise.
33852 (tag_lines): Likewise.
33853 (tag_counters): Likewise.
33854 (tag_summary): Likewise.
33855 (dump_working_sets): Use depth to do a proper indentation.
33857 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33860 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33861 * symtab.c (symtab_node::debug_symtab): Move definition here.
33863 2017-04-28 Richard Biener <rguenther@suse.de>
33865 * lto-streamer.h (LTO_major_version): Bump to 7.
33867 2017-04-28 Richard Biener <rguenther@suse.de>
33869 * tree-vrp.c (assert_info): New struct.
33870 (add_assert_info): New helper.
33871 (register_edge_assert_for_2): Refactor to add asserts to a vector
33873 (register_edge_assert_for_1): Likewise.
33874 (register_edge_assert_for): Likewise.
33875 (finish_register_edge_assert_for): New helper actually registering
33876 asserts where live on edge.
33877 (find_conditional_asserts): Adjust.
33878 (find_switch_asserts): Likewise.
33879 (evrp_dom_walker::try_find_new_range): Generalize.
33880 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33882 2017-04-27 Marek Polacek <polacek@redhat.com>
33885 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33886 arg10 and arg11 to itype.
33888 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33890 * doc/extend.texi (Object Size Checking): Improve grammar.
33892 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33895 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33896 that the logic for permitting reciprocal estimates matches that
33899 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33902 * tree.c (type_cache_hasher::equal): Only compare
33903 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33904 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33905 non-aggregate element types.
33906 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33907 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33909 2017-04-27 Richard Biener <rguenther@suse.de>
33911 PR middle-end/80533
33912 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33913 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33914 keeping a reference to a trailing array.
33916 2017-04-27 Richard Biener <rguenther@suse.de>
33918 PR middle-end/80539
33919 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33920 being in loop-closed SSA form conservatively.
33921 (chrec_fold_multiply_poly_poly): Likewise.
33923 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33925 PR middle-end/79665
33926 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33927 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33929 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33932 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33933 (aarch64_function_arg_alignment): Return unsigned int again, but still
33934 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33935 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33936 Don't emit -Wpsabi note.
33937 (aarch64_function_arg_boundary): Likewise.
33938 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33941 2017-04-26 Nathan Sidwell <nathan@acm.org>
33943 * tree.h (crc32_unsigned_n): Declare.
33944 (crc32_unsigned, crc32_unsigned): Make inline.
33945 * tree.c (crc32_unsigned_bits): Replace with ...
33946 (crc32_unsigned_n): ... this.
33947 (crc32_unsigned, crc32_byte): Remove.
33948 (crc32_string): Remove unnecessary braces.
33950 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33952 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33953 * ipa-inline-analysis.c (MAX_TIME): Remove.
33954 (account_size_time): Use sreal for time.
33955 (dump_inline_summary): Update.
33956 (estimate_function_body_sizes): Update.
33957 (estimate_edge_size_and_time): Update.
33958 (estimate_calls_size_and_time): Update.
33959 (estimate_node_size_and_time): Update.
33960 (inline_merge_summary): Update.
33961 (inline_update_overall_summary): Update.
33962 (estimate_time_after_inlining): Update.
33963 (inline_read_section): Update.
33964 (inline_write_summary): Update.
33965 * ipa-inline.c (compute_uninlined_call_time): Update.
33966 (compute_inlined_call_time): Update.
33967 (recursive_inlining): Update.
33968 (inline_small_functions): Update.
33969 (dump_overall_stats): Update.
33970 * ipa-inline.h: Include sreal.h.
33971 (size_time_entry): Turn time to sreal.
33972 (inline_summary): Turn self_time nad time to sreal.
33974 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33976 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33978 (sreal::stream_out, sreal::stream_in): New.
33979 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33981 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33983 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33986 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33989 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33991 (dimode_scalar_chain::compute_convert_gain): Ditto.
33992 (dimode_scalar_chain::make_vector_copies): Ditto.
33993 (dimode_scalar_chain::convert_reg): Ditto.
33994 (dimode_scalar_chain::convert_insn): Ditto.
33995 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33996 (VI248_AVX512BW_1): New mode iterator.
33997 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33998 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
34001 2017-04-25 Martin Sebor <msebor@redhat.com>
34003 PR tree-optimization/80497
34004 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
34005 constants are representable in HOST_WIDE_INT.
34006 (parse_directive): Ditto.
34008 2017-04-25 Martin Sebor <msebor@redhat.com>
34011 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
34012 (new_zero_array): Adjust signature.
34013 (dom_info::dom_init): Used unsigned rather that size_t.
34014 (dom_info::dom_info): Same.
34016 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
34017 Jakub Jelinek <jakub@redhat.com>
34020 * config/arm/arm.c: Include gimple.h.
34021 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
34022 returns negative, increment ncrn only if it returned positive.
34023 (arm_needs_doubleword_align): Return int instead of bool,
34024 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
34025 members, but if there is any such non-FIELD_DECL
34026 > PARM_BOUNDARY aligned decl, return -1 instead of false.
34027 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
34028 returns negative, increment nregs only if it returned positive.
34029 (arm_setup_incoming_varargs): Likewise.
34030 (arm_function_arg_boundary): Emit -Wpsabi note if
34031 arm_needs_doubleword_align returns negative, return
34032 DOUBLEWORD_ALIGNMENT only if it returned positive.
34034 2017-04-25 Marek Polacek <polacek@redhat.com>
34037 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34038 first argument to type.
34040 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
34043 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
34044 type checks to test for compatibility instead of equality.
34046 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
34047 Jakub Jelinek <jakub@redhat.com>
34050 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
34052 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
34053 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
34054 the alignment computation, but return their maximum in warn_alignment.
34055 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
34056 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
34058 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
34059 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
34062 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34064 * config/arc/simdext.md (dmpyh): Fix typo.
34066 2017-04-25 Richard Biener <rguenther@suse.de>
34068 PR tree-optimization/80492
34069 * alias.c (compare_base_decls): Handle registers with asm
34070 specification conservatively.
34071 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
34072 compare_base_decls returning dont-know properly.
34074 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34076 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
34077 (legitimate_offset_address_p): New function.
34078 (arc_legitimate_address_p): Use above function.
34080 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34082 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
34084 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34086 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
34087 ACCH registers whenever they are available.
34089 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34091 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
34092 double regs fix when not used.
34094 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34096 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
34098 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
34099 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
34101 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34103 * config/arc/arc.c (arc_output_addsi): Check for h-register class
34104 when emitting short ADD instructions.
34106 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34108 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
34110 (cmpsi_cc_c_insn): Likewise.
34111 (cbranchsi4_scratch): Compute proper instruction length using
34112 compact_hreg_operand.
34113 * config/arc/predicates.md (compact_hreg_operand): New predicate.
34115 2017-04-25 Richard Biener <rguenther@suse.de>
34117 PR middle-end/80509
34118 * passes.c (pass_manager::pass_manager): Initialize
34119 m_name_to_pass_map.
34121 2017-04-25 Richard Biener <rguenther@suse.de>
34123 PR tree-optimization/79201
34124 * tree-ssa-sink.c (statement_sink_location): Handle calls.
34126 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34129 * config/s390/vector.md: Split MEM->GPR vector moves for
34130 non-s_operand addresses.
34132 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34135 * config/s390/predicates.md (reload_const_wide_int_operand): New
34137 * config/s390/s390.md ("movti"): Remove d/P alternative.
34138 ("movti_bigconst"): New pattern definition.
34140 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34143 * s390-protos.h (s390_expand_cs_hqi): Removed.
34144 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
34145 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
34146 modes as well as CCZ1mode and CCZmode.
34147 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
34148 signature of s390_emit_compare_and_swap.
34149 (s390_expand_cs_hqi): Likewise, make static.
34150 (s390_expand_cs_tdsi): Generate an explicit compare before trying
34151 compare-and-swap, in some cases.
34152 (s390_expand_cs): Wrapper function.
34153 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
34155 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
34156 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
34157 patterns for small and large integers. Forbid symref memory operands.
34158 Move expander to s390.c. Require cc register.
34159 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
34160 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
34161 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
34162 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
34163 symref memory operands. Remove CC mode and call s390_match_ccmode
34165 ("atomic_exchange<mode>"): Allow and implement all integer modes.
34167 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34169 * config/s390/s390.md (define_peephole2): New peephole to help
34170 combining the load-and-test pattern with volatile memory.
34172 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34174 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
34175 with CCZmode for TARGET_Z196.
34177 2017-04-25 Jakub Jelinek <jakub@redhat.com>
34179 PR rtl-optimization/80501
34180 * combine.c (make_compound_operation_int): Set subreg_code to SET
34181 even for AND with mask of the sign bit of mode.
34183 PR rtl-optimization/80500
34184 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
34185 sum's initial value.
34187 2017-04-25 Julian Brown <julian@codesourcery.com>
34188 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34190 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
34192 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
34194 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
34196 2017-04-25 Julian Brown <julian@codesourcery.com>
34197 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34199 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
34200 (thunderx2t99_sha): New Reservation.
34202 2017-04-25 Julian Brown <julian@codesourcery.com>
34203 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34205 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
34206 type for 1-element load.
34208 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
34210 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
34212 2017-04-24 Martin Jambor <mjambor@suse.cz>
34214 PR tree-optimization/80293
34215 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
34216 char arrays not totally scalarizable if it is false.
34217 (analyze_all_variable_accesses): Pass correct value in the new
34218 parameter. Add a statistics counter.
34220 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
34222 PR middle-end/79931
34223 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
34225 2017-04-24 Richard Biener <rguenther@suse.de>
34227 PR tree-optimization/80494
34228 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
34229 out for complex types.
34231 2017-04-24 Richard Biener <rguenther@suse.de>
34233 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
34234 * tree-ssa-sccvn.c (print_scc): Print SCC size.
34235 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
34236 (DFS): Adjust and never fail.
34237 (sccvn_dom_walker::fail): Remove.
34238 (sccvn_dom_walker::before_dom_children): Adjust.
34239 (run_scc_vn): Likewise and never fail.
34240 * tree-ssa-pre.c (pass_pre::execute): Adjust.
34241 (pass_fre::execute): Likewise.
34243 2017-04-24 Richard Biener <rguenther@suse.de>
34245 PR tree-optimization/79725
34246 * tree-ssa-sink.c (statement_sink_location): Return whether
34247 failure reason was zero uses. Move that check later.
34248 (sink_code_in_bb): Deal with zero uses by removing the stmt
34251 2017-04-24 Richard Biener <rguenther@suse.de>
34254 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
34255 pointer-based references.
34257 2017-04-24 Richard Biener <rguenther@suse.de>
34260 * pass_manager.h (pass_manager::operator new): Remove.
34261 (pass_manager::operator delete): Likewise.
34262 * passes.c (pass_manager::operator new): Remove.
34263 (pass_manager::operator delete): Likewise.
34264 (pass_manager::pass_manager): Zero individual pass members.
34266 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
34269 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
34270 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
34271 Check "XEXP (src, 1)" operand here.
34272 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
34273 Check "XEXP (src, 1)" operand here.
34274 (dimode_scalar_chain::make_vector_copies): Detect count register
34275 of a shift instruction. Zero extend count register from QImode
34276 to DImode to satisfy vector shift pattern count operand predicate.
34277 Substitute vector shift count operand with a DImode copy.
34278 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
34281 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
34283 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
34284 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34285 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
34286 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34287 (UNSPEC_NOREX_MEM): Remove definition.
34289 2017-04-21 Richard Biener <rguenther@suse.de>
34291 PR tree-optimization/79547
34292 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
34293 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
34294 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
34295 without any constraints.
34297 2017-04-21 Richard Biener <rguenther@suse.de>
34299 PR tree-optimization/78847
34300 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
34302 2017-04-21 Richard Biener <rguenther@suse.de>
34304 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
34305 (build_distinct_type_copy): Likewise.
34306 (build_variant_type_copy): Likewise.
34307 * tree.c (build_qualified_type): Pass down mem-stat info.
34308 (build_distinct_type_copy): Likewise.
34309 (build_variant_type_copy): Likewise.
34311 2017-04-21 Richard Biener <rguenther@suse.de>
34313 PR tree-optimization/80237
34314 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
34316 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
34317 for a simplified result.
34319 2016-04-21 Richard Biener <rguenther@suse.de>
34321 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
34322 sth as strict as a simple_iv but a chrec without symbols and an
34323 operand defined in the loop we are peeling (and not some subloop).
34324 (propagate_constants_for_unrolling): Propagate all constants.
34326 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34329 * config/i386/i386.c (print_reg): Remove assert for disalowed
34330 regno values, call output_operand_lossage instead.
34332 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34335 * config/i386/constraints.md (Yc): New register constraint.
34336 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
34337 Use Yc constraint for alternative 2 of operand 0. Remove
34338 preferred_for_speed attribute.
34340 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34342 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
34343 lastprivate clauses in SIMT case.
34345 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
34347 * doc/invoke.texi (-Wextra-semi): Document new warning option.
34349 2017-04-20 Richard Biener <rguenther@suse.de>
34351 PR tree-optimization/57796
34352 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
34353 as N scalar stores.
34354 (vect_model_load_cost): Cost gathers as N scalar loads.
34356 2017-04-20 Richard Biener <rguenther@suse.de>
34358 * ggc-page.c (ggc_allocated_p): Rename to ...
34359 (safe_lookup_page_table_entry): ... this and return the lookup
34361 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
34363 2017-04-20 Richard Biener <rguenther@suse.de>
34365 PR tree-optimization/80453
34366 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
34367 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
34368 from the conditions.
34369 (vn_phi_eq): Pass them down.
34370 (vn_phi_lookup): Record them.
34371 (vn_phi_insert): Likewise.
34373 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
34375 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
34376 uninitialized variable warning to avoid buffer overrun.
34378 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34381 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
34382 is suppressed for '{ 0 }' in C.
34384 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34386 * BASE-VER: Set to 8.0.0.
34388 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
34390 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
34391 priority .init_array and .fini_array section with SECTION_NOTYPE
34394 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34396 PR middle-end/80423
34397 * tree.h (build_array_type): Add typeless_storage default argument.
34398 * tree.c (type_cache_hasher::equal): Also compare
34399 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
34400 (build_array_type): Add typeless_storage argument, set
34401 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
34403 (build_nonshared_array_type): Adjust build_array_type_1 caller.
34404 (build_array_type): Likewise. Add typeless_storage argument.
34406 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34407 Jakub Jelinek <jakub@redhat.com>
34409 PR tree-optimization/80426
34410 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
34411 operation on symbolic operands, also compute the overflow for the
34412 invariant part when the operation degenerates into a negation.
34414 2017-04-19 Jakub Jelinek <jakub@redhat.com>
34417 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
34418 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
34421 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
34423 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
34426 * config/avr/avr.c (tree.h): Include it.
34427 (cgraph.h): Include it.
34428 (avr_encode_section_info): Don't warn for uninitialized progmem
34429 variable if it's just an alias.
34431 2017-04-19 Richard Biener <rguenther@suse.de>
34434 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
34435 when needed by AutoPGO.
34437 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
34440 * doc/lto.texi: Remove an extra 'that'.
34442 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
34444 PR rtl-optimization/80429
34445 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
34446 are only used in debug insns.
34448 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34449 Vladimir Makarov <vmakarov@redhat.com>
34451 * config/sparc/predicates.md (input_operand): Add comment. Return
34452 true for any memory operand when LRA is in progress.
34453 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
34455 2017-04-18 Jeff Law <law@redhat.com>
34458 * mips.md ({return,simple_return}_internal): Do not overwrite
34461 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34463 PR tree-optimization/80443
34464 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
34465 instead of adding 1, subtract -1 and similarly instead of subtracting
34468 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
34470 PR rtl-optimization/80357
34471 * haifa-sched.c (tmp_bitmap): New variable.
34472 (model_recompute): Handle duplicate use records.
34473 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
34474 (free_global_sched_pressure_data): Free it.
34476 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
34479 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34480 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34481 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34482 instead of SYSTEM_HEADER_DIR.
34484 2017-04-18 Jeff Law <law@redhat.com>
34486 PR middle-end/80422
34487 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
34488 predecessors after walking up the insn chain.
34490 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34493 * dwarf2out.c (modified_type_die): Try harder not to emit internal
34494 sizetype type into debug info.
34496 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34499 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
34500 unneeded test for TARGET_UPPER_REGS_SF.
34501 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
34503 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34506 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
34507 instead of gsi_after_labels.
34509 2017-04-18 Jeff Law <law@redhat.com>
34511 * regcprop.c (maybe_mode_change): Avoid creating copies of the
34515 2017-04-13 Jeff Law <law@redhat.com>
34516 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34517 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34519 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
34522 * config/avr/avr.c (intl.h): Include it.
34523 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
34525 2017-04-18 Martin Liska <mliska@suse.cz>
34527 PR gcov-profile/78783
34528 * gcov-tool.c (gcov_output_files): Validate that destination
34529 file is either removed by the tool or by a user.
34531 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34532 Guy Benyei <guybe@mellanox.com>
34534 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
34535 block, and do not negate it, the stored id is already negative.
34537 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34539 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
34541 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34544 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
34545 masks of options that should be turned off if the VSX vector
34546 options are turned off.
34547 (OTHER_P8_VECTOR_MASKS): Likewise.
34548 (OTHER_VSX_VECTOR_MASKS): Likewise.
34549 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
34550 rs6000_disable_incompatible_switches to validate no type switches
34552 (rs6000_incompatible_switch): New function to disallow turning on
34553 other vector options if -mno-vsx, -mno-power8-vector, or
34554 -mno-power9-vector are specified.
34556 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34558 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
34560 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34562 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
34563 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
34564 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
34565 (ARG_POINTER_CFA_OFFSET): Likewise.
34567 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34569 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
34570 conditions to take advantage of various optimizations.
34572 2017-04-13 Jeff Law <law@redhat.com>
34574 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34575 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34576 (zero_extendsidi2_dext): Likewise.
34578 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34581 * fold-const.c (fold_ternary_loc): Revert
34582 use op0 instead of fold_convert_loc (loc, type, arg0) part of
34585 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
34587 PR rtl-optimization/80343
34588 * lra-remat.c (update_scratch_ops): Assign original hard reg to
34589 new scratch pseudo.
34591 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
34594 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
34595 to ubsan_encode_value.
34597 2017-04-13 Jeff Law <law@redhat.com>
34599 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
34600 appearing in DEBUG_INSNs.
34602 2017-04-13 Martin Liska <mliska@suse.cz>
34604 PR gcov-profile/80413
34605 * gcov-io.c (gcov_write_string): Copy to buffer just when
34606 allocated size is greater than zero.
34608 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34611 * dwarf2out.c (decls_for_scope): Ignore declarations of
34612 current_function_decl in BLOCK_NONLOCALIZED_VARS.
34614 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
34617 * ipa-visibility.c (non_local_p): Fix typos.
34618 (localize_node): When localizing symbol in same comdat group,
34619 dissolve the group only when we know external symbols are going
34621 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
34623 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34625 PR tree-optimization/79390
34626 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
34627 order does not result in usable sequence, retry with reversed operand
34633 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
34634 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
34635 op0 instead of fold_convert_loc (loc, type, arg0).
34637 2017-04-12 Jeff Law <law@redhat.com>
34639 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
34640 has a delay slot in the generated code.
34642 * config/cris/cris.md (cris_preferred_reload_class): Return
34643 GENNONACR_REGS rather than GENERAL_REGS.
34645 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34648 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
34649 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
34650 signedness of the result type.
34652 2017-04-12 Richard Biener <rguenther@suse.de>
34653 Jeff Law <law@redhat.com>
34655 PR tree-optimization/80359
34656 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
34657 trim stores to TARGET_MEM_REFs.
34659 2017-04-12 Richard Biener <rguenther@suse.de>
34661 PR tree-optimization/79390
34662 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
34663 threading case even more.
34665 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
34668 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
34669 for quad_address_p for TImode, instead of just not indexed_address.
34671 2017-04-12 Richard Biener <rguenther@suse.de>
34672 Bernd Edlinger <bernd.edlinger@hotmail.de>
34674 PR middle-end/79671
34675 * alias.c (component_uses_parent_alias_set_from): Handle
34676 TYPE_TYPELESS_STORAGE.
34677 (get_alias_set): Likewise.
34678 * tree-core.h (tree_type_common): Add typeless_storage flag.
34679 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
34680 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
34681 for types containing members with TYPE_TYPELESS_STORAGE.
34682 (place_field): Likewise.
34683 (layout_type): Likewise for ARRAY_TYPE.
34684 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
34685 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
34686 TYPE_TYPELESS_STORAGE.
34687 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
34689 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34692 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34693 first argument to type.
34695 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34699 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
34700 CONST0_RTX (mode) rather than const0_rtx where appropriate.
34701 (rs6000_expand_binop_builtin): Likewise.
34702 (rs6000_expand_ternop_builtin): Likewise; also add missing
34703 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
34704 vshasigma built-ins.
34705 * doc/extend.texi: Document that vec_xxpermdi's third argument
34706 must be a constant.
34708 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
34710 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
34711 Use shift_const cost parameter when calculating gain of STV shifts.
34713 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
34715 PR rtl-optimization/70478
34716 * lra-constraints.c (process_alt_operands): Check memory for
34717 disfavoring memory insn operand.
34719 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34721 PR middle-end/80100
34722 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
34723 left shift in unsigned HOST_WIDE_INT type.
34725 PR rtl-optimization/80385
34726 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
34727 (not (neg X)) into (plus X -1) for complex or non-integral modes.
34730 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
34731 if they have any depend clauses.
34733 2017-04-11 Martin Liska <mliska@suse.cz>
34736 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
34737 * ipa-split.c (split_function): Create a local comdat symbol
34738 if caller is in a comdat group.
34740 2017-04-11 Martin Liska <mliska@suse.cz>
34743 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
34746 2017-04-11 Martin Sebor <msebor@redhat.com>
34748 PR middle-end/80364
34749 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
34750 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
34752 (directive::set_width, directive::set_precision, format_character):
34754 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
34757 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
34760 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
34761 conflict, set target->arch_name instead of target->cpu_name.
34763 2017-04-11 Richard Biener <rguenther@suse.de>
34765 PR tree-optimization/80374
34766 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
34767 build_zero_cst, remove fold_convertible_p check again.
34769 2017-04-11 Martin Liska <mliska@suse.cz>
34772 * ubsan.c (instrument_object_size): Do not instrument register
34775 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34778 * config/i386/i386-builtin-types.def
34779 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
34780 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
34781 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
34782 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
34783 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
34784 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
34785 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
34786 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
34787 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
34788 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
34789 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34790 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34791 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34792 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34793 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34794 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34795 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34796 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34797 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34798 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34799 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34800 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34801 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34802 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34803 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34804 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34805 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34806 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34807 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34808 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34809 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34810 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34811 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34812 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34813 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34814 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34815 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34816 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34817 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34818 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34819 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34820 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34821 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34822 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34823 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34824 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34826 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34827 flag to second_arg_count, handle 4 argument function type _COUNT
34828 aliases, handle second_arg_count on second argument rather than last.
34830 2017-04-10 Jeff Law <law@redhat.com>
34832 PR tree-optimization/80374
34833 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34834 record anything if we can not convert integer_zero_node to the
34837 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34840 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34841 Enhance special handling given to the TARGET_P9_MINMAX option in
34842 relation to certain other options.
34844 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34846 PR tree-optimization/80153
34847 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34848 remove POINTER_PLUS_EXPR's base part directly, rather than through
34851 2017-04-10 Richard Biener <rguenther@suse.de>
34852 Bin Cheng <bin.cheng@arm.com>
34854 PR tree-optimization/80153
34855 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34856 the first element of pointer type aff_tree. Build result expr in
34858 (add_elt_to_tree): Convert to type unconditionally. Remove other
34859 fold_convert calls.
34860 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34861 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34863 2017-04-10 Richard Biener <rguenther@suse.de>
34865 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34868 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34870 PR rtl-optimization/70478
34871 * lra-constraints.c (curr_small_class_check): New.
34872 (update_and_check_small_class_inputs): New.
34873 (process_alt_operands): Update curr_small_class_check. Disfavor
34874 alternative insn memory operands. Check available regs for small
34877 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34880 * config/mips/mips.opt (-mvirt): Update description.
34881 * doc/invoke.texi (-mvirt): Likewise.
34883 2017-04-10 Richard Biener <rguenther@suse.de>
34885 PR middle-end/80362
34886 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34887 looking for NEGATE_EXPR in -A / -B to A / B folding.
34889 2017-04-10 Martin Liska <mliska@suse.cz>
34891 PR gcov-profile/80224
34892 * gcov.c (print_usage): Fix usage string.
34893 (get_gcov_intermediate_filename): Remove.
34894 (output_gcov_file): Use both for normal and intermediate format.
34895 (generate_results): Do not initialize special file for
34896 intermediate format.
34898 2017-04-10 Richard Biener <rguenther@suse.de>
34900 PR tree-optimization/80304
34901 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34904 2017-04-10 Nathan Sidwell <nathan@acm.org>
34907 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34908 (rs6000_init_builtins): Use it.
34910 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34912 * config/arm/arm.md (<mrc>): Add mode to SET source.
34913 (<mrrc>): Likewise.
34915 2017-04-10 Richard Biener <rguenther@suse.de>
34917 PR middle-end/80344
34918 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34920 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34923 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34924 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34925 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34926 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34927 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34928 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34929 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34930 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34931 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34932 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34933 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34934 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34935 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34936 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34937 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34938 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34939 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34940 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34941 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34942 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34943 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34944 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34945 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34947 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34949 PR rtl-optimization/70478
34950 * lra-constraints.c: Reverse the last patch.
34952 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34954 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34955 Add comment for WCHAR_T.
34957 2017-04-08 Martin Liska <mliska@suse.cz>
34960 2017-04-07 Martin Liska <mliska@suse.cz>
34963 * ipa-split.c (split_function): Add function part to a same comdat
34966 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34969 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34971 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34973 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34974 (vec_pairing): ...to this.
34975 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34976 (rs6000_sched_init): Adjust for name change.
34977 (struct rs6000_sched_context): Likewise.
34978 (rs6000_init_sched_context): Likewise.
34979 (rs6000_set_sched_context): Likewise.
34981 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34987 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34989 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34990 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34991 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34993 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34995 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34997 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34999 PR rtl-optimization/70703
35000 * ira-color.c (update_conflict_hard_regno_costs): Use
35001 int64_t instead of HOST_WIDE_INT.
35003 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
35005 PR rtl-optimization/70478
35006 * lra-constraints.c (process_alt_operands): Disfavor alternative
35007 insn memory operands.
35009 2017-04-07 Jeff Law <law@redhat.com>
35011 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
35012 CALL and NOTE_INSN_CALL_ARG_LOCATION.
35014 2017-04-07 Martin Liska <mliska@suse.cz>
35017 * config/aarch64/aarch64.c (aarch64_process_target_attr):
35018 Show error message instead of an ICE.
35020 2017-04-07 Martin Liska <mliska@suse.cz>
35023 * ipa-split.c (split_function): Add function part to a same comdat
35026 2017-04-07 Richard Biener <rguenther@suse.de>
35028 PR middle-end/80341
35029 * tree.c (get_unwidened): Also handle ! for_type case for
35031 * convert.c (do_narrow): Split out from ...
35032 (convert_to_integer_1): ... here. Do not pass final truncation
35033 type to get_unwidened for TRUNC_DIV_EXPR.
35035 2017-04-07 Richard Biener <rguenther@suse.de>
35037 * tree-affine.c (wide_int_ext_for_comb): Take type rather
35039 (aff_combination_const): Adjust.
35040 (aff_combination_scale): Likewise.
35041 (aff_combination_add_elt): Likewise.
35042 (aff_combination_add_cst): Likewise.
35043 (aff_combination_convert): Likewise.
35044 (add_elt_to_tree): Likewise. Remove unused argument.
35045 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
35047 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
35049 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
35051 * config/arm/arm.c (arm_default_short_enums): Use
35052 ARM_DEFAULT_SHORT_ENUMS.
35053 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
35055 2017-04-06 Jakub Jelinek <jakub@redhat.com>
35058 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
35059 members with redundant out-of-class redeclaration.
35061 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
35064 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
35065 * config/i386/i386.md (*zero_extendsidi2):
35066 Add (?*x,*x) and (?*v,*v) alternatives.
35068 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
35071 * config/i386/i386.c (ix86_expand_builtin)
35072 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
35073 mode from insn data. Convert operands to insn operand mode.
35074 Copy operands that don't satisfy insn predicate to a register.
35076 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
35078 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
35081 2017-04-06 Richard Biener <rguenther@suse.de>
35083 PR tree-optimization/80334
35084 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
35085 preserve alignment of accesses.
35087 2017-04-06 Richard Biener <rguenther@suse.de>
35089 PR tree-optimization/80262
35090 * tree-sra.c (build_ref_for_offset): Preserve address-space
35092 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
35093 Drop useless address-space information on MEM_REF offsets.
35095 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
35097 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
35099 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
35101 PR rtl-optimization/70703
35102 * ira-color.c (update_conflict_hard_regno_costs): Use
35103 HOST_WIDE_INT instead of long.
35105 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
35108 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
35109 not defined for x86_64 target. Add -mmmx target option when __SSE2__
35111 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
35112 for x86_64 target. Handle -m3dnowa option.
35114 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
35116 PR rtl-optimization/70703
35117 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
35118 (update_conflict_hard_regno_costs): Use long instead of unsigned
35119 arithmetic for cost calculation.
35121 2017-04-05 Jakub Jelinek <jakub@redhat.com>
35122 Bernd Edlinger <bernd.edlinger@hotmail.de>
35125 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
35128 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
35131 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
35132 ptr_mode with Pmode throughout.
35133 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
35134 into probe_stack_range and use DImode.
35136 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
35139 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
35140 call_eh_return is true.
35142 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35144 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
35145 Initialize last_match_fntype_index.
35147 2017-04-05 Jakub Jelinek <jakub@redhat.com>
35150 * tree-nvr.c: Include internal-fn.h.
35151 (pass_return_slot::execute): Ignore internal calls without
35154 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35155 Richard Biener <rguenther@suse.de>
35158 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
35159 captures used multiple times, except for the last use.
35160 * generic-match-head.c: Include gimplify.h.
35162 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35164 PR tree-optimization/79390
35165 * target.h (struct noce_if_info): Declare.
35166 * targhooks.h (default_noce_conversion_profitable_p): Declare.
35167 * target.def (noce_conversion_profitable_p): New target hook.
35168 * ifcvt.h (struct noce_if_info): New type, moved from ...
35169 * ifcvt.c (struct noce_if_info): ... here.
35170 (noce_conversion_profitable_p): Renamed to ...
35171 (default_noce_conversion_profitable_p): ... this. No longer
35173 (noce_try_store_flag_constants, noce_try_addcc,
35174 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
35175 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
35176 instead of noce_conversion_profitable_p.
35177 * config/i386/i386.c: Include ifcvt.h.
35178 (ix86_option_override_internal): Don't override
35179 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
35180 (ix86_noce_conversion_profitable_p): New function.
35181 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
35182 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
35183 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
35184 * doc/tm.texi: Regenerated.
35186 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35188 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
35191 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
35194 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
35195 instructions for small multiply cores.
35197 2017-04-04 Jeff Law <law@redhat.com>
35199 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
35201 (mips_expand_vec_perm_const): Initialize elements in orig_perm
35202 that are not set by the loop over the elements.
35204 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35207 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
35208 int mode, convert_modes it to mode as unsigned, otherwise use
35209 lowpart_subreg to mode rather than SImode.
35210 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
35211 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
35212 Use DImode instead of SImode for the shift count operand.
35213 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
35216 2017-04-04 Richard Biener <rguenther@suse.de>
35218 PR middle-end/80281
35219 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
35220 arithmetic done for the negate or the plus. Simplify.
35221 (A - (-B) -> A + B): Likewise.
35222 * fold-const.c (split_tree): Make sure to not negate pointers.
35224 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
35226 PR rtl-optimization/60818
35227 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
35228 a compare of comparisons with the thing compared if this results
35229 in a different machine mode.
35231 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
35233 * alias.c (base_alias_check): Fix typo in comment.
35234 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
35235 * cgraphunit.c (symbol_table::compile): Likewise.
35236 * collect2.c (maybe_run_lto_and_relink): Likewise.
35237 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
35238 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
35239 * config/avr/avr.c (avr_map_op_t): Likewise.
35240 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
35241 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
35242 * config/epiphany/epiphany.md (movcc): Likewise.
35243 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
35244 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
35246 * config/mips/mips.c (mips_save_restore_reg): Likewise.
35247 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
35248 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
35249 * config/sh/sh.c (sh_rtx_costs): Likewise.
35250 * fold-const.c (fold_truth_andor): Likewise.
35251 * genautomata.c (collapse_flag): Likewise.
35252 * gengtype.h (struct type::u::s): Likewise.
35253 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
35254 * input.c (FORMAT_AMOUNT): Likewise.
35255 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
35256 (known_aggs_to_agg_replacement_list): Likewise.
35257 * ipa-inline-analysis.c: Likewise.
35258 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
35259 * ipa-polymorphic-call.c
35260 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
35261 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
35262 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
35264 * modulo-sched.c (apply_reg_moves): Likewise.
35265 * omp-expand.c (build_omp_regions_1): Likewise.
35266 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
35267 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
35268 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
35269 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
35270 * value-prof.c: Likewise.
35271 * var-tracking.c (val_reset): Likewise.
35273 2017-04-03 Richard Biener <rguenther@suse.de>
35275 PR tree-optimization/80275
35276 * fold-const.c (split_address_to_core_and_offset): Handle
35279 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
35281 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
35282 descriptors is at least equal to that of functions.
35284 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35286 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
35288 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35291 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
35292 (mov<IMOD4:mode>): New expander.
35293 (*mov<IMOD4:mode>_internal): New insn and split pattern.
35295 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
35297 PR rtl-optimization/79405
35298 * fwprop.c (propagations_left): New variable.
35299 (forward_propagate_into): Decrement it.
35300 (fwprop_init): Initialize it.
35301 (fw_prop): If the variable has reached zero, stop propagating.
35302 (fwprop_addr): Ditto.
35304 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35307 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
35308 a FUNCTION_DECL, pass it as decl instead of origin to
35311 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
35313 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
35316 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
35319 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
35320 TARGET_VSX_SMALL_INTEGER.
35322 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35324 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
35325 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
35327 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
35329 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
35330 extraction from odd-numbered MSA register.
35332 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35334 PR middle-end/80173
35335 * expmed.c (store_bit_field_1): Don't attempt to create
35336 a word subreg out of hard registers wider than word if they
35337 have HARD_REGNO_NREGS of 1 for their mode.
35339 PR middle-end/80163
35340 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
35341 conversions to integer types wider than word and pointer.
35344 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
35345 (rtx_equal_for_cselib_p): Pass 0 to it.
35346 * cselib.c (cselib_hasher::equal): Likewise.
35347 (rtx_equal_for_cselib_1): Add depth argument. If depth
35348 is 128, don't look up VALUE locs and punt. Increment
35349 depth in recursive calls when walking VALUE locs.
35351 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
35353 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
35354 (make_gcov_file_name): Use the canonical path name for generating
35356 (read_line): Fix handling of files with ascii null bytes.
35358 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
35360 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
35361 to initialise a vector register instead
35362 of using a const_int.
35364 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35366 PR translation/80189
35367 * gimplify.c (omp_default_clause): Use %qs instead of %s in
35368 diagnostic messages.
35370 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
35373 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
35374 (dfp_diex_<mode>): Update mode of operand 1.
35375 * doc/extend.texi (dxex, dxexq): Document change to return type.
35376 (diex, diexq): Document change to argument type.
35378 2017-03-30 Martin Jambor <mjambor@suse.cz>
35381 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
35382 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
35383 it reflects the signature changes performed at the callee side.
35384 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
35385 to cgraph_build_function_type_skip_args.
35386 (build_function_decl_skip_args): Adjust call to the above function.
35388 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35391 * config/i386/sse.md
35392 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
35393 register as dest whenever it is a MEM not rtx_equal_p to the
35394 corresponding dup operand, and when forcing into reg move the
35395 reg into the memory afterwards.
35396 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
35397 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
35398 for the force_reg mode.
35399 (avx512vl_vextractf128<mode>): Use register as dest either
35400 always when a MEM, or when it is a MEM not rtx_equal_p to the
35401 corresponding dup operand, or even not when it is a CONST_VECTOR
35402 depending on the mode and lo vs. hi.
35403 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
35405 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
35406 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
35407 Likewise. Require that operands[2] is even.
35408 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
35409 Remove extraneous parens. Require that operands[2] is a multiple
35411 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
35412 operands[0] is a MEM if <mask_applied>, the predicates/constraints
35413 disallow memory then.
35415 2017-03-30 Richard Biener <rguenther@suse.de>
35417 PR tree-optimization/77498
35418 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
35419 to non-constants over backedges.
35421 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
35423 PR rtl-optimization/80233
35424 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
35425 as last_combined_insn. Do not test for BARRIER_P separately.
35427 2017-03-29 Andreas Schwab <schwab@suse.de>
35430 * calls.c (prepare_call_address): Convert funexp to Pmode before
35431 copying to temp reg.
35433 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35435 PR tree-optimization/80158
35436 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
35437 Handle possible future case of more than one alternate
35439 (replace_rhs_if_not_dup): Likewise.
35440 (replace_one_candidate): Likewise.
35442 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
35444 PR rtl-optimization/80193
35445 * ira.c (ira): Do not check allocation for LRA.
35447 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
35449 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
35450 (nvptx_output_simt_exit): Declare.
35451 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
35452 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
35453 (init_softstack_frame): Move initialization of crtl->is_leaf to...
35454 (nvptx_declare_function_name): ...here. Emit declaration of local
35455 memory space buffer for omp_simt_enter insn.
35456 (nvptx_output_unisimt_switch): New.
35457 (nvptx_output_softstack_switch): New.
35458 (nvptx_output_simt_enter): New.
35459 (nvptx_output_simt_exit): New.
35460 * config/nvptx/nvptx.h (struct machine_function): New fields
35461 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
35462 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
35463 (UNSPECV_SIMT_EXIT): Ditto.
35464 (omp_simt_enter_insn): New insn.
35465 (omp_simt_enter): New expansion.
35466 (omp_simt_exit): New insn.
35467 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
35469 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
35470 (expand_GOMP_SIMT_ENTER_ALLOC): New.
35471 (expand_GOMP_SIMT_EXIT): New.
35472 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
35473 (GOMP_SIMT_ENTER_ALLOC): Ditto.
35474 (GOMP_SIMT_EXIT): Ditto.
35475 * target-insns.def (omp_simt_enter): New insn.
35476 (omp_simt_exit): Ditto.
35477 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
35479 (lower_rec_simd_input_clauses): Implement SIMT privatization.
35480 (lower_rec_input_clauses): Likewise.
35481 (lower_lastprivate_clauses): Handle SIMT privatization.
35483 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
35484 (ompdevlow_adjust_simt_enter): New.
35485 (find_simtpriv_var_op): New.
35486 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
35487 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
35489 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
35490 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
35491 (copy_decl_for_dup_finish): Ditto.
35493 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
35495 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
35498 * config/i386/i386.c (ix86_option_override_internal): Always
35499 allow -mpreferred-stack-boundary=3 for 64-bit targets.
35501 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35503 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
35505 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35507 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
35508 mark new edge's irreducible flag accordign to it.
35509 (vect_do_peeling): Check loop preheader edge's irreducible flag
35510 and pass it to function slpeel_add_loop_guard.
35512 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
35514 PR tree-optimization/80218
35515 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35516 Update block frequencies and counts.
35518 2017-03-28 Richard Biener <rguenther@suse.de>
35520 PR tree-optimization/78644
35521 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
35522 of a simplification result we may not use it at all.
35524 2017-03-28 Richard Biener <rguenther@suse.de>
35527 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
35528 without arguments, generate default definition of a SSA name.
35530 2017-03-28 Richard Biener <rguenther@suse.de>
35532 PR middle-end/80222
35533 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
35534 TYPE_REF_CAN_ALIAS_ALL references.
35535 * fold-const.c (fold_indirect_ref_1): Likewise.
35537 2017-03-28 Martin Liska <mliska@suse.cz>
35540 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
35541 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
35543 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35544 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
35546 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
35547 (EXTRA_SPECS): Define.
35548 (SUBTARGET_EXTRA_SPECS): Likewise.
35549 (SUBTARGET_CPP_SPEC): Likewise.
35550 * config/arc/elf.h (EXTRA_SPECS): Renamed to
35551 SUBTARGET_EXTRA_SPECS.
35552 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
35554 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35556 * config/arc/simdext.md (vst64_insn): Update pattern.
35557 (vld32wh_insn): Likewise.
35558 (vld32wl_insn): Likewise.
35559 (vld64_insn): Likewise.
35560 (vld32_insn): Likewise.
35562 2017-03-28 Marek Polacek <polacek@redhat.com>
35565 * fold-const.c (fold_comparison): Use protected_set_expr_location
35566 instead of SET_EXPR_LOCATION.
35568 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
35570 * tree.c (add_expr): Avoid name lookup warning.
35572 2017-03-27 Jeff Law <law@redhat.com>
35574 PR tree-optimization/80216
35575 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
35576 function name. Limit recursion depth.
35577 (record_temporary_equivalences): Corresponding changes.
35579 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
35581 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
35584 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35587 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
35589 * cfgcleanup.c (reg_note_cfa_p): New array.
35590 (insns_have_identical_cfa_notes): New function.
35591 (old_insns_match_p): Don't cross-jump in between /f
35592 and non-/f instructions. If both i1 and i2 are frame related,
35593 verify all CFA notes, their order and content.
35595 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
35598 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
35599 HImode and SImode with zero extend to DImode to one insn.
35600 (bswap<mode>2_extenddi): Likewise.
35601 (bswapsi2_extenddi): Likewise.
35602 (bswaphi2_extendsi): Likewise.
35603 (bswaphi2): Combine bswap HImode and SImode into one insn.
35604 Separate memory insns from swapping register.
35605 (bswapsi2): Likewise.
35606 (bswap<mode>2): Likewise.
35607 (bswaphi2_internal): Delete, no longer used.
35608 (bswapsi2_internal): Likewise.
35609 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
35610 store, and gpr<-gpr swap insns.
35611 (bswap<mode>2_store): Likewise.
35612 (bswaphi2_reg): Register only splitter, combine with the splitter.
35613 (bswaphi2 splitter): Likewise.
35614 (bswapsi2_reg): Likewise.
35615 (bswapsi2 splitter): Likewise.
35616 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
35617 the insns into load, store, and register/register insns.
35618 (bswapdi2_ldbrx): Likewise.
35619 (bswapdi2_load): Likewise.
35620 (bswapdi2_store): Likewise.
35621 (bswapdi2_reg): Likewise.
35623 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
35625 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
35626 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
35628 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
35631 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
35633 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35634 special handling for target option conflicts between dform
35635 options (-mpower9-dform, -mpower9-dform-vector,
35636 -mpower9-dform-scalar) and -mno-direct-move.
35638 2017-03-27 Richard Biener <rguenther@suse.de>
35640 PR tree-optimization/80181
35641 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
35643 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35645 * config/arc/predicates.md (move_double_src_operand): Replace the
35646 call to move_double_src_operand with a call to address_operand.
35648 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35650 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
35651 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
35652 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
35654 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35656 * config/arc/predicates.md (long_immediate_loadstore_operand):
35657 Consider scaled addresses cases.
35659 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35661 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
35662 restored when in interrupt.
35663 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
35664 doesn't have delay slot.
35666 2017-03-27 Richard Biener <rguenther@suse.de>
35669 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
35670 inlined thunk clones.
35672 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35675 * asan.c (instrument_derefs): Copy over last operand from
35676 original COMPONENT_REF to the new COMPONENT_REF with
35677 DECL_BIT_FIELD_REPRESENTATIVE.
35678 * ubsan.c (instrument_object_size): Likewise.
35680 2017-03-27 Richard Biener <rguenther@suse.de>
35682 PR tree-optimization/80170
35683 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
35684 sure DR/SCEV didnt fold in constants we do not see when looking
35685 at the reference base alignment.
35687 2017-03-27 Richard Biener <rguenther@suse.de>
35689 PR middle-end/80171
35690 * gimple-fold.c (fold_ctor_reference): Properly guard against
35691 NULL return value from canonicalize_constructor_val.
35693 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
35696 * config/i386/i386.c (ix86_expand_builtin)
35697 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
35698 flags reg setting and flags reg using instructions.
35699 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
35700 clobbering instructions to zero extend op2.
35702 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
35704 * doc/install.texi (Configuration) <--with-aix-soname>:
35705 Update link to AIX ld.
35707 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
35709 PR rtl-optimization/80160
35710 PR rtl-optimization/80159
35711 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
35712 reg_alternate_class into account.
35714 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
35717 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
35718 to consider in curr_insn_transform.
35720 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35722 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
35723 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
35724 and emit_mode_inner.
35726 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35728 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
35729 argument to the overloaded builtin variants. Use the new flag to
35730 deprecate certain builtin variants.
35731 * config/s390/s390-builtin-types.def: Add new builtin types.
35732 * config/s390/s390-builtins.h: Support new flags field for
35733 overloaded builtins.
35734 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
35735 (s390_macro_to_expand): Enable vector float data type.
35736 (s390_cpu_cpp_builtins_internal): Indicate support of the new
35737 builtins by incrementing the __VEC__ version number.
35738 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
35740 (s390_resolve_overloaded_builtin): Emit error messages depending
35741 on the builtin flags.
35742 * config/s390/s390.c (s390_expand_builtin): Support additional
35743 flags argument. Change error message to match the messages
35744 emitted in s390-c.c.
35745 * config/s390/s390.md: New UNSPEC_* constants.
35746 (op_type): Add new instruction types.
35747 * config/s390/vecintrin.h: Add new builtins and test data class
35749 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
35750 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
35751 (VEC_INEXACT, VEC_NOINEXACT): New constants.
35752 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
35753 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
35754 ("vec_mergel<mode>"): V_HW -> VEC_HW.
35756 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
35757 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
35758 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
35759 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
35761 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
35762 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
35763 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
35764 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
35766 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
35767 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
35768 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
35769 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
35770 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
35771 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
35772 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
35774 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
35775 ("vec_scatter_element<V_HW_4:mode>_DI")
35776 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
35777 ("vec_fpint<mode>", "vflls")
35778 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
35779 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
35780 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
35781 ("*vec_cmphe<mode>_cc"): ... these.
35783 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
35784 mode constant instead of magic value.
35786 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35788 * config/s390/s390.c (s390_expand_vec_compare): Support other
35789 vector floating point modes than just V2DF.
35790 (s390_expand_vcond): Likewise.
35791 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35792 (s390_cannot_change_mode_class): Prevent mode changes between TF
35793 and V1TF in vector registers.
35794 * config/s390/s390.md (DF, SF): New mode attributes.
35795 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35796 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35797 SFmode support for VRs.
35798 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35800 (VFT, VF_HW): New mode iterators.
35801 (vw, sdx): New mode attributes.
35802 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35803 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35804 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35805 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35806 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35807 also the new vector floating point modes. Renaming to ...
35809 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35810 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35811 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35812 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35813 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35814 ("vec_unordered<mode>"): ... these.
35816 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35817 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35818 ("*vec_extendv2df"): New insn definitions.
35820 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35822 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35823 ("mulditi3_2", "*muldi3_sign"): New patterns.
35824 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35825 rename the pattern definition.
35827 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35829 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35831 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35833 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35835 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35836 instruction if possible.
35837 * config/s390/vector.md (vec_halfnumelts): New mode
35839 ("*vec_vllezlf<mode>"): New pattern.
35841 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35843 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35844 ("popcountv4si2", "popcountv2di2"): Rename to ...
35845 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35846 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35848 ("popcount<mode>2_vxe"): New pattern.
35850 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35852 * common/config/s390/s390-common.c (processor_flags_table): Add
35854 * config.gcc: Add arch12.
35855 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35856 Default to arch12 for unknown CPU model numbers.
35857 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35858 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35859 PROCESSOR_max sanity check.
35860 * config/s390/s390-opts.h (enum processor_type): Add
35862 * config/s390/s390.c (processor_table): Add arch12.
35863 (s390_expand_builtin): Add check for B_VXE flag.
35864 (s390_issue_rate): Add PROCESSOR_ARCH12.
35865 (s390_get_sched_attrmask): Likewise.
35866 (s390_get_unit_mask): Likewise.
35867 (s390_sched_score): Enable z13 scheduling for arch12.
35868 (s390_sched_reorder): Likewise.
35869 (s390_sched_variable_issue): Likewise.
35870 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35872 (s390_tune_attr): Use z13 scheduling also for arch12.
35873 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35874 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35875 (TARGET_VXE_P): New macros.
35876 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35877 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35878 * config/s390/s390.opt: Add arch12 as processor_type.
35880 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35882 * config/s390/s390.md
35883 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35884 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35885 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35887 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35888 Rename expanders to ...
35890 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35891 ("fixuns_truncdddi2_emu"): ... these.
35893 ("fixuns_trunc<mode>si2_emu"): New expander.
35895 ("*fixuns_truncdfdi2_z13"): Rename to ...
35896 ("*fixuns_truncdfdi2_vx"): ... this.
35898 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35900 * config/s390/2964.md: Remove the single element vector compare
35901 instructions which are no longer used.
35902 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35904 (s390_canonicalize_comparison): Remove handling of DFmode
35906 (s390_expand_vec_compare_scalar): Remove function.
35907 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35908 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35910 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35912 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35914 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35916 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35917 will anyway by matched by mov<mode>_64dfp.
35919 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35921 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35922 vlef/vstef. Add missing operand to vleif.
35924 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35926 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35927 pair for all vector types with 64 bit elements.
35928 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35929 * config/s390/vector.md (V_HW_64): ... here.
35930 (V_128_NOSINGLE): New mode iterator.
35931 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35932 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35933 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35934 ("*vec_load_pairv2di"): Change to ...
35935 ("*vec_load_pair<mode>"): ... this one.
35937 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35939 * config/s390/constraints.md: Add comments.
35940 (jKK): Reject element sizes > 8 bytes.
35941 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35943 * config/s390/s390.md: Add the s_operand checks formerly in
35944 s390_split_ok_p to various splitters where they are still
35946 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35947 for 128 bit vectors. Plus two splitters.
35949 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35951 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35954 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35957 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35958 error if the boundary argument is not constant.
35960 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35962 PR rtl-optimization/80112
35963 * loop-doloop.c (doloop_condition_get): Don't check condition
35964 if cmp isn't SET with IF_THEN_ELSE src.
35966 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35968 PR tree-optimization/80158
35969 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35970 replacing a candidate statement, also replace it for the
35971 candidate's alternate interpretation.
35972 (replace_rhs_if_not_dup): Likewise.
35973 (replace_one_candidate): Likewise.
35975 2017-03-24 Richard Biener <rguenther@suse.de>
35977 PR tree-optimization/80167
35978 * graphite-isl-ast-to-gimple.c
35979 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35981 (translate_isl_ast_to_gimple::get_rename): Likewise.
35983 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35985 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35986 handling of certain combinations of target options, including the
35987 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35988 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35990 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35993 * config/arm/arm.md (*load_multiple): Add reload_completed to
35994 matching condition.
35996 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35997 Richard Biener <rguenth@suse.de>
35999 PR tree-optimization/79908
36000 PR tree-optimization/80136
36001 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
36002 been cast away, gimplify_and_add suffices.
36004 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
36006 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
36008 2017-03-23 Richard Biener <rguenther@suse.de>
36010 PR tree-optimization/80032
36011 * gimplify.c (gimple_push_cleanup): Forced unconditional
36012 cleanups still have to go to the conditional_cleanups
36015 2017-03-22 Jakub Jelinek <jakub@redhat.com>
36017 PR tree-optimization/80072
36018 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
36020 (next_operand_entry_id): Change type to unsigned int.
36021 (sort_by_operand_rank): Make sure to return the right return value
36022 even if unsigned fields are bigger than INT_MAX.
36023 (struct oecount): Change cnt and id type to unsigned int.
36024 (oecount_hasher::equal): Formatting fix.
36025 (oecount_cmp): Make sure to return the right return value
36026 even if unsigned fields are bigger than INT_MAX.
36027 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
36030 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
36031 TREE_READONLY on result if writing it more than once.
36034 * doc/invoke.texi (-fsanitize=thread): Document that with
36035 -fnon-call-exceptions atomics are not able to throw
36039 * tsan.c: Include tree-eh.h.
36040 (instrument_builtin_call): Call maybe_clean_eh_stmt or
36041 maybe_clean_or_replace_eh_stmt where needed.
36042 (instrument_memory_accesses): Add cfg_changed argument.
36043 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
36044 if it returned true.
36045 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
36047 PR rtl-optimization/63191
36048 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
36049 wrapper function, moved the whole old content into ...
36050 (ix86_delegitimize_address_1): ... this. New inline function.
36051 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
36052 true as last argument instead of ix86_delegitimize_address.
36054 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
36056 * config/aarch64/aarch64.c (generic_branch_cost): Copy
36057 cortexa57_branch_cost.
36059 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
36061 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
36063 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36066 * doc/md.texi (Constraints): Document wA constraint.
36067 * config/rs6000/constraints.md (wA): New.
36068 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
36069 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
36070 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
36071 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
36073 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
36076 * gimplify.c (is_oacc_declared): New function.
36077 (oacc_default_clause): Use it to set default flags for acc declared
36078 variables inside parallel regions.
36079 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
36080 declared variables.
36081 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
36082 declare attribute to any decl as necessary.
36084 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
36087 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
36088 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
36089 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
36090 (arm_arch_lpae): This.
36091 * config/arm/arm.c (arm_arch7ve): Rename into ...
36092 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
36093 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
36096 2017-03-22 Martin Liska <mliska@suse.cz>
36099 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
36100 error message instead of an ICE.
36102 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36104 * doc/extend.texi (6.11 Additional Floating Types): Revise.
36106 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
36108 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36110 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
36113 2017-03-21 Martin Sebor <msebor@redhat.com>
36115 * doc/extend.texi: Use "cannot" instead of "can't."
36116 * doc/hostconfig.texi: Same.
36117 * doc/install.texi: Same.
36118 * doc/invoke.texi: Same.
36119 * doc/loop.texi: Same.
36120 * doc/md.texi: Same.
36121 * doc/objc.texi: Same.
36122 * doc/rtl.texi: Same.
36123 * doc/tm.texi: Same.
36124 * doc/tm.texi.in: Same.
36125 * doc/trouble.texi: Same.
36127 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
36130 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
36131 (collect_checksum_attributes): Set it.
36132 (die_checksum_ordered): Use it.
36134 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36136 PR tree-optimization/79908
36137 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
36138 change: For a VA_ARG whose LHS has been cast away, use
36139 force_gimple_operand to construct the side effects.
36141 2017-03-21 David Malcolm <dmalcolm@redhat.com>
36143 PR translation/80001
36144 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
36145 more amenable to translation.
36146 (oacc_loop_auto_partitions): Likewise.
36148 2017-03-21 Marek Polacek <polacek@redhat.com>
36149 Martin Sebor <msebor@redhat.com>
36151 PR tree-optimization/80109
36152 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
36153 on INTEGRAL_TYPE_P.
36155 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36156 Segher Boessenkool <segher@kernel.crashing.org>
36159 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
36160 check reg_used_between_p between insn and one of succ or succ2
36161 depending on if succ is artificial insn not inserted into insn
36164 2017-03-21 Martin Liska <mliska@suse.cz>
36166 PR gcov-profile/80081
36167 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
36168 * doc/gcc.texi: Include gcov-dump stuff.
36169 * doc/gcov-dump.texi: New file.
36171 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
36173 PR rtl-optimization/79150
36174 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
36175 conditional jump, if the jump is the last insn of the loop.
36177 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36178 Richard Biener <rguenth@suse.de>
36180 PR tree-optimization/79908
36181 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
36182 been cast away, use force_gimple_operand to construct the side
36185 2017-03-21 Martin Liska <mliska@suse.cz>
36187 PR libfortran/79956
36188 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
36191 2017-03-21 Brad Spengler <spender@grsecurity.net>
36194 * plugin.c (htab_hash_plugin): New function.
36195 (add_new_plugin): Use it and adjust.
36196 (parse_plugin_arg_opt): Adjust.
36197 (init_one_plugin): Likewise.
36199 2017-03-21 Richard Biener <rguenther@suse.de>
36201 PR tree-optimization/80032
36202 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
36203 if set force the cleanup to happen unconditionally.
36204 (gimplify_target_expr): Push inserted clobbers with force_uncond
36205 to avoid them being removed by control-dependent DCE.
36207 2017-03-21 Richard Biener <rguenther@suse.de>
36209 PR tree-optimization/80122
36210 * tree-inline.c (copy_bb): Do not expans va-arg packs or
36211 va_arg_pack_len when the inlined call stmt requires pack
36213 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
36215 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36218 * tsan.c (instrument_builtin_call): If the memory model argument
36219 is not a constant, assume it is valid.
36222 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
36225 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36227 PR rtl-optimization/79910
36228 * combine.c (can_combine_p): Do not allow combining an I0 or I1
36229 if its dest is used by an insn before I2 (other than the combined
36230 insns themselves, which are properly handled already).
36232 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36235 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36237 * combine.c (record_used_regs): New static function.
36238 (try_combine): Handle situations where there is an additional
36239 instruction between I2 and I3 which needs to have a LOG_LINK
36243 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36245 * combine.c (try_combine): Delete redundant i1 test. Call
36246 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36248 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36251 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
36252 alternatives 13/14.
36254 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36256 PR tree-optimization/80054
36257 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
36258 the optimization if a PHI or any of its arguments is not dominated
36259 by the candidate's basis. Use gphi* rather than gimple* as
36261 (replace_profitable_candidates): Clean up a gimple* variable that
36262 should be a gphi* variable.
36264 2017-03-20 Martin Sebor <msebor@redhat.com>
36267 * doc/extend.texi (attribute constructor): Document present limitation.
36269 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
36272 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
36273 __POWER9_VECTOR__ #ifdef control, change template definition to
36274 use Power9-specific built-in function.
36275 (vec_any_eq): Likewise.
36276 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
36277 to control outcomes from this test.
36278 (vector_ae_<mode>p): For VEC_F modes, likewise.
36280 2017-03-20 Ian Lance Taylor <iant@google.com>
36282 * config/i386/i386.c (ix86_function_regparm): Save an extra
36283 register for -fsplit-stack with DECL_STATIC_CHAIN.
36285 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36288 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
36289 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
36291 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36293 * config/riscv/riscv.c (riscv_print_operand): Use "fence
36295 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
36298 2017-03-20 Marek Polacek <polacek@redhat.com>
36301 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
36303 2017-03-20 Richard Biener <rguenther@suse.de>
36305 PR tree-optimization/80113
36306 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
36307 allocate extra SSA name for PHI def.
36308 (add_close_phis_to_outer_loops): Likewise.
36309 (add_close_phis_to_merge_points): Likewise.
36310 (copy_loop_close_phi_args): Likewise.
36311 (copy_cond_phi_nodes): Likewise.
36313 2017-03-20 Martin Liska <mliska@suse.cz>
36315 PR middle-end/79753
36316 * tree-chkp.c (chkp_build_returned_bound): Do not build
36317 returned bounds for a LHS that's not a BOUNDED_P type.
36319 2017-03-20 Martin Liska <mliska@suse.cz>
36323 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
36324 COMPLEX_CST and VECTOR_CST.
36326 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36329 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
36330 target operand. A new splitter adds the clobber statement in case
36331 the target operand is dead anyway.
36333 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
36335 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
36336 to age-old versions of binutils and glibc.
36338 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
36340 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
36342 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36344 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
36346 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36348 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
36349 requirement for binutils 2.13.
36351 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36353 * combine.c (try_combine): Delete redundant i1 test. Call
36354 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36356 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
36358 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
36359 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
36361 <riscv64-*-elf>: Re-arrange section
36362 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
36363 <riscv32-*-linux>: Likewise.
36364 <riscv64-*-elf>: Likewise
36365 <riscv64-*-linux>: Likewise.
36367 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
36370 * aarch64.opt(verbose-cost-dump): Fix typo.
36372 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
36375 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
36376 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
36378 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36380 * reload.c (find_reloads): When reloading a nonoffsettable address,
36381 use RELOAD_OTHER for it and its address reloads.
36383 PR rtl-optimization/79910
36384 * combine.c (record_used_regs): New static function.
36385 (try_combine): Handle situations where there is an additional
36386 instruction between I2 and I3 which needs to have a LOG_LINK
36389 2017-03-17 Jeff Law <law@redhat.com>
36391 PR tree-optimization/71437
36392 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
36393 conditional in the hash table first.
36394 (vrp_dom_walker::before_dom_children): Extract condition from
36395 ASSERT_EXPR. Record condition, its inverion and any implied
36396 conditions as well.
36398 2017-03-17 Marek Polacek <polacek@redhat.com>
36399 Markus Trippelsdorf <markus@trippelsdorf.de>
36401 PR tree-optimization/80079
36402 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
36405 2017-03-17 Richard Biener <rguenther@suse.de>
36407 PR middle-end/80075
36408 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
36409 Properly verify the LHS before the RHS possibly claims to be
36411 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
36414 2017-03-17 Martin Jambor <mjambor@suse.cz>
36416 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
36417 (List of -O2 options): Likewise.
36418 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
36421 2017-03-17 Tom de Vries <tom@codesourcery.com>
36423 * gcov-dump.c (print_usage): Print bug_report_url.
36425 2017-03-17 Richard Biener <rguenther@suse.de>
36427 PR middle-end/80050
36428 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
36429 (parser::peek): Likewise.
36431 2017-03-17 Richard Biener <rguenther@suse.de>
36433 PR tree-optimization/80048
36434 * sese.c (free_sese_info): Properly release rename_map and
36435 copied_bb_map elements.
36437 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
36439 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
36440 Add linked-list forward and backlinks. Insert on
36441 construction, remove on destruction.
36442 (class pass_store_merging): Add m_stores_head field.
36443 (pass_store_merging::terminate_and_process_all_chains):
36444 Iterate over m_stores_head list.
36445 (pass_store_merging::terminate_all_aliasing_chains):
36447 (pass_store_merging::execute): Check for debug stmts first.
36448 Push new chains onto the m_stores_head stack.
36450 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
36453 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
36454 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
36455 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
36457 2017-03-16 Jeff Law <law@redhat.com>
36459 PR tree-optimization/71437
36460 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
36461 member function. Implementation moved into after_dom_children
36462 member function and into the threader's thread_outgoing_edges
36464 (dom_opt_dom_walker::after_dom_children): Simplify by moving
36465 some code into new thread_outgoing_edges.
36466 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
36467 definition. Simplify marker handling (do it here). Assume we always
36468 have the available expression and the const/copies tables.
36469 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
36471 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
36472 * tree-vrp.c (equiv_stack): No longer file scoped.
36473 (vrp_dom_walker): New class.
36474 (vrp_dom_walker::before_dom_children): New member function.
36475 (vrp_dom_walker::after_dom_children): Likewise.
36476 (identify_jump_threads): Setup domwalker. Use it rather than
36477 walking edges in a random order by hand. Simplify setup/finalization.
36478 (finalize_jump_threads): Remove.
36479 (vrp_finalize): Do not call identify_jump_threads here.
36480 (execute_vrp): Do it here instead and call thread_through_all_blocks
36483 PR tree-optimization/71437
36484 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
36486 (simplify_stmt_for_jump_threading): Add basic_block argument. All
36488 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
36489 (dom_opt_dom_walker::thread_across_edge): Remove
36490 handle_dominating_asserts argument. All callers changed.
36491 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
36492 changes. Remove calls to lhs_of_dominating_assert. Other
36493 uses of handle_dominating_asserts turn into unconditional code
36494 (simplify_control_stmt_condition_1): Likewise.
36495 (simplify_control_stmt_condition): Likewise.
36496 (thread_through_normal_block, thread_across_edge): Likewise.
36497 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
36498 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
36499 object if it is not an SSA_NAME.
36500 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
36501 before calling into the VRP specific simplifiers.
36502 (identify_jump_threads): Remove handle_dominating_asserts
36505 2017-03-16 Jakub Jelinek <jakub@redhat.com>
36508 * tree-diagnostic.c (default_tree_printer): No longer static.
36509 * tree-diagnostic.h (default_tree_printer): New prototype.
36511 2017-03-16 Tamar Christina <tamar.christina@arm.com>
36513 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
36514 Change ins into fmov.
36516 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36518 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
36519 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
36520 Use h_con constraint for operand 1.
36521 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
36522 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
36524 2017-03-15 Jeff Law <law@redhat.com>
36526 PR tree-optimization/71437
36527 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
36528 (record_temporary_equivalences): Use it.
36530 PR tree-optimization/71437
36531 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
36532 tree-ssa-scopedtables.
36533 (lookup_avail_expr, build_and_record_new_cond): Likewise.
36534 (record_conditions, record_cond, vuse_eq): Likewise.
36535 (record_edge_info): Adjust to API tweak of record_conditions.
36536 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
36537 (record_temporary_equivalences, optimize_stmt): Likewise.
36538 (eliminate_redundant_computations): Likewise.
36539 (record_equivalences_from_stmt): Likewise.
36540 * tree-ssa-scopedtables.c: Include options.h and params.h.
36541 (vuse_eq): New function, moved from tree-ssa-dom.c
36542 (build_and_record_new_cond): Likewise.
36543 (record_conditions): Likewise. Accept vector of conditions rather
36544 than edge_equivalence structure for first argument.
36545 for the first argument.
36546 (avail_exprs_stack::lookup_avail_expr): New member function, moved
36547 from tree-ssa-dom.c.
36548 (avail_exprs_stack::record_cond): Likewise.
36549 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
36550 from tree-ssa-dom.c.
36551 (avail_exprs_stack): Add new member functions lookup_avail_expr
36553 (record_conditions): Declare.
36555 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
36558 * lra-constraints.c (process_alt_operands): Increase reject for
36559 reloading an input/output operand.
36561 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
36564 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
36565 insns to convert from signed/unsigned char/short to IEEE 128-bit
36567 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
36569 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
36572 * config/i386/i386.c (ix86_vector_duplicate_value): Create
36573 subreg of inner mode for values already in registers.
36575 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
36577 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
36578 iteration reg is used after the loop.
36580 2017-03-14 Martin Sebor <msebor@redhat.com>
36582 PR tree-optimization/79800
36583 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
36584 precision in negative-positive range.
36585 (format_floating): Call non-const overload with adjusted precision.
36587 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
36590 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
36593 2017-03-14 Martin Sebor <msebor@redhat.com>
36595 PR middle-end/80020
36596 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
36597 * builtins.def (aligned_alloc): Use it.
36600 * Makefile.in (GTFILES): Add calls.c.
36601 * calls.c: Include "gt-calls.h".
36603 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
36605 PR rtl-optimization/79728
36606 * regs.h (struct target_regs): New field
36607 x_contains_allocatable_regs_of_mode.
36608 (contains_allocatable_regs_of_mode): New macro.
36609 * reginfo.c (init_reg_sets_1): Initialize it, and change
36610 contains_reg_of_mode so it includes global regs as well.
36611 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
36612 rather than contains_regs_of_mode.
36614 2017-03-14 Martin Liska <mliska@suse.cz>
36616 * doc/invoke.texi: Document options that can't be combined with
36617 -fcheck-pointer-bounds.
36619 2017-03-14 Martin Liska <mliska@suse.cz>
36621 PR middle-end/79831
36622 * doc/invoke.texi (-Wchkp): Document the option.
36624 2017-03-14 Martin Liska <mliska@suse.cz>
36626 * Makefile.in: Install gcov-dump.
36628 2017-03-14 Martin Liska <mliska@suse.cz>
36630 * multiple_target.c (expand_target_clones): Bail out for
36631 an invalid attribute.
36633 2017-03-14 Richard Biener <rguenther@suse.de>
36635 * alias.c (struct alias_set_entry): Pack properly.
36636 * cfgloop.h (struct loop): Likewise.
36637 * cse.c (struct set): Likewise.
36638 * ipa-utils.c (struct searchc_env): Likewise.
36639 * loop-invariant.c (struct invariant): Likewise.
36640 * lra-remat.c (struct cand): Likewise.
36641 * recog.c (struct change_t): Likewise.
36642 * rtl.h (struct address_info): Likewise.
36643 * symbol-summary.h (function_summary): Likewise.
36644 * tree-loop-distribution.c (struct partition): Likewise.
36645 * tree-object-size.c (struct object_size_info): Likewise.
36646 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
36647 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
36648 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
36649 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
36650 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
36651 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
36652 (struct _stmt_vec_info): Likewise.
36654 2017-03-14 Martin Liska <mliska@suse.cz>
36657 * multiple_target.c (create_dispatcher_calls): Check that
36658 a target can create a function dispatcher.
36660 2017-03-14 Martin Liska <mliska@suse.cz>
36663 * multiple_target.c (expand_target_clones): Drop local.local
36664 flag for default implementation.
36666 2017-03-14 Richard Biener <rguenther@suse.de>
36668 PR tree-optimization/80030
36669 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
36671 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
36673 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
36674 gcc_fallthrough() instead of __attribute__((fallthrough));
36676 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36678 * doc/gcc.texi: Remove "up" link to (DIR).
36679 * doc/gccint.texi: Ditto.
36681 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36683 * doc/install.texi (Specific) <avr>: Remove reference to
36686 2017-03-13 Jeff Law <law@redhat.com>
36688 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
36689 attribute rather than comments.
36691 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
36692 match_scratch operand is highest.
36694 2017-03-13 Martin Liska <mliska@suse.cz>
36696 PR middle-end/78339
36697 * ipa-pure-const.c (warn_function_noreturn): If the declarations
36698 is a CHKP clone, use original declaration.
36700 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36702 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
36703 (arc_conditional_register_usage): Use a different allocation order
36704 when optimizing for size.
36705 * common/config/arc/arc-common.c (arc_option_optimization_table):
36706 Section anchors default on when optimizing for size.
36708 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36710 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
36712 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36714 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
36715 * config/arc/arc.md (cpu_facility): Add cd variant.
36716 (*movqi_insn): Add code density variant.
36717 (*movhi_insn): Likewise.
36718 (*movqi_insn): Likewise.
36719 (*addsi3_mixed): Likewise.
36720 (subsi3_insn): Likewise.
36722 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36724 * config/arc/arc.md (movsi_cond_exec): Update constraint.
36726 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36728 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
36729 expressions with MINUS and UNARY ops.
36731 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36734 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
36736 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
36737 between vec_select and vector argument.
36738 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
36739 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
36740 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
36741 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
36742 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
36743 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
36745 2017-03-13 Richard Biener <rguenther@suse.de>
36748 * params.def (vect-max-peeling-for-alignment): Fix typo.
36750 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36752 * doc/install.texi (Specific) <mips-*-*>: Remove description of
36753 issue that only occurred with binutils below 2.18.
36755 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36757 * doc/install.texi (Specific) <cris-axis-elf>: No longer
36758 refer to binutils 2.11/2.12 minimum.
36760 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36762 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
36763 ftp.kernel.org and simplify binutils requirement.
36765 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
36767 * doc/invoke.texi (Warning Options): Fix spelling of link-time
36769 (Optimize Options): Ditto. Also remove redundancy.
36771 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36773 PR translation/79848
36774 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
36776 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
36777 to G_ to avoid double translation.
36779 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36781 PR translation/79923
36782 * auto-profile.c (get_combined_location): Convert leading
36783 character of diagnostics to lower case and remove trailing period.
36784 (read_profile): Likewise for various diagnostics.
36785 * config/arm/arm.c (arm_option_override): Remove trailing period
36786 from various diagnostics.
36787 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
36788 (msp430_expand_delay_cycles): Likewise.
36790 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36793 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36794 full command-line argument, rather than just "str".
36795 (aarch64_validate_march): Likewise.
36796 (aarch64_validate_mtune): Likewise.
36798 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36800 PR rtl-optimization/78911
36801 * lra-assigns.c (must_not_spill_p): New function.
36802 (spill_for): Use it.
36804 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36806 PR tree-optimization/79981
36807 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36808 ATOMIC_COMPARE_EXCHANGE ifn result.
36809 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36810 IFN_ATOMIC_COMPARE_EXCHANGE.
36812 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36815 * opts.c (parse_sanitizer_options): Add missing question mark to
36816 "did you mean" message.
36818 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36820 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36822 (VMULEUH_UNS): Likewise.
36823 (VMULOUB_UNS): Likewise.
36824 (VMULOUH_UNS): Likewise.
36825 * config/rs6000/rs6000.c (builtin_function_type): Remove
36826 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36828 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36831 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36832 x with result of extra_parsing_for_operand_code_0.
36833 (function_reader::extra_parsing_for_operand_code_0): Convert
36834 return type from void to rtx, returning x. When reading
36835 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36836 larger size containing struct block_symbol.
36838 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36840 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36841 -mfloat128-hardware without -m64.
36843 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36846 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36847 entries to the case statement that marks unsigned arguments to
36848 overloaded functions.
36850 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36852 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36853 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36855 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36858 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36859 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36861 2017-03-10 Martin Liska <mliska@suse.cz>
36865 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36866 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36869 2017-03-10 Olivier Hainque <hainque@adacore.com>
36871 * tree-switch-conversion (array_value_type): Start by resetting
36872 candidate type to it's main variant.
36874 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36876 PR rtl-optimization/79909
36877 * combine.c (try_combine): Use simplify_replace_rtx on individual
36878 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36879 of the whole CALL_INSN_FUNCTION_USAGE.
36881 PR tree-optimization/79972
36882 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36883 get_range_info on SSA_NAMEs. Formatting fixes.
36885 2017-03-10 Richard Biener <rguenther@suse.de>
36886 Jakub Jelinek <jakub@redhat.com>
36888 PR tree-optimization/77975
36889 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36890 edge to be constant.
36891 (get_val_for): For constant x return it. Formatting fix.
36892 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36893 would use the same bases as the current one.
36895 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36897 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36898 instead of vec_select for V1TImode.
36899 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36901 (VSX_LE_128): Add V1TI to this mode iterator.
36902 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36903 (*vsx_le_perm_store_<mode>): Likewise.
36904 (pre-reload splitter for VSX stores): Likewise.
36905 (post-reload splitter for VSX stores): Likewise.
36906 (*vsx_xxpermdi2_le_<mode>): Likewise.
36907 (*vsx_lxvd2x2_le_<mode>): Likewise.
36908 (*vsx_stxvd2x2_le_<mode>): Likewise.
36910 2017-03-09 Michael Eager <eager@eagercon.com>
36912 Correct failures with --enable-checking=yes,rtl.
36914 * config/microblaze/microblaze.c (microblaze_expand_shift):
36915 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36916 test for const0_rtx.
36917 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36918 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36920 2017-03-09 Richard Biener <rguenther@suse.de>
36922 PR tree-optimization/79977
36923 * graphite-scop-detection.c (scop_detection::merge_sese):
36924 Handle the case of extra exits to blocks dominating the entry.
36926 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36928 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36931 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36933 PR rtl-optimization/79949
36934 * lra-constraints.c (process_alt_operands): Check memory when
36935 trying to predict a cycle. Print about the overall increase.
36937 2017-03-09 Richard Biener <rguenther@suse.de>
36939 PR middle-end/79971
36940 * gimple-expr.c (useless_type_conversion_p): Preserve
36941 TYPE_SATURATING for fixed-point types.
36943 2017-03-09 Richard Biener <rguenther@suse.de>
36946 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36947 alignment of BLKmode params.
36949 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36952 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36953 (VALL_NO_V2Q): Likewise.
36955 * config/aarch64/aarch64-simd.md
36956 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36958 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36959 VALL_NO_V2Q mode iterator.
36960 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36962 2017-03-09 Martin Liska <mliska@suse.cz>
36964 PR tree-optimization/79631
36965 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36966 tree_int_cst_sign_bit just for INTEGER constants.
36968 2017-03-09 Martin Liska <mliska@suse.cz>
36972 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36975 2017-03-09 Marek Polacek <polacek@redhat.com>
36978 * tree.c (inchash::add_expr): Handle TREE_VEC.
36980 2017-03-09 Martin Liska <mliska@suse.cz>
36983 (chkp_narrow_size_and_offset): New function.
36984 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36985 (void chkp_parse_bit_field_ref): New function.
36986 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36987 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36989 2017-03-09 Martin Liska <mliska@suse.cz>
36992 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36993 (chkp_find_bounds_1): Remove gcc_unreachable.
36995 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36998 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36999 BUILT_IN_SYNC*, determine the access type from the size suffix and
37000 always build a MEM_REF with that type. Handle forgotten
37001 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
37004 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
37005 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
37006 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
37007 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
37008 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
37009 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
37010 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
37011 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
37012 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
37013 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
37014 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
37015 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
37016 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
37017 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
37018 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
37019 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
37020 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
37021 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
37022 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
37023 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
37024 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
37025 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
37026 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
37027 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
37028 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
37029 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
37030 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
37031 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
37032 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
37033 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
37034 definitions outside of __OPTIMIZE__ guarded section.
37037 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
37038 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
37039 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
37040 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
37043 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37045 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
37046 ("vfenez<mode>"): Add missing constraints.
37048 2017-03-08 Martin Sebor <msebor@redhat.com>
37051 * config/nds32/nds32.c (nds32_option_override):
37052 Fix misspelled diagnostic.
37054 2017-03-08 Jakub Jelinek <jakub@redhat.com>
37057 * gimplify.c (gimplify_omp_for): Replace index var in outer
37058 taskloop statement with an artificial variable and add
37059 OMP_CLAUSE_PRIVATE clause for it.
37061 2017-03-08 Richard Biener <rguenther@suse.de>
37063 PR tree-optimization/79955
37064 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
37065 for accesses that are completely outside of the variable.
37067 2017-03-08 Andrew Haley <aph@redhat.com>
37069 PR tree-optimization/79943
37070 * tree-ssa-loop-split.c (compute_new_first_bound): When
37071 calculating the new upper bound, (END-BEG) should be added, not
37074 2017-03-08 Jakub Jelinek <jakub@redhat.com>
37076 * config/avr/avr.md (setmemhi): Make sure match_dup
37077 operand number comes before match_scratch.
37079 2017-03-08 Richard Biener <rguenther@suse.de>
37081 PR tree-optimization/79920
37082 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
37083 with ncopies == 1 to ...
37084 (vect_transform_slp_perm_load): ... here. Properly compute
37085 all element loads by iterating VF times over the group. Do
37086 not handle ncopies (computed in a broken way) in
37087 vect_create_mask_and_perm.
37089 2017-03-08 Jakub Jelinek <jakub@redhat.com>
37092 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
37093 is a uniform vector, use uniform_vector_p return value instead of
37094 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
37096 2017-03-07 Marek Polacek <polacek@redhat.com>
37098 PR middle-end/79809
37099 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
37100 (alloca_call_type): Likewise.
37102 2017-03-07 Martin Liska <mliska@suse.cz>
37104 * gcov.c (process_args): Put comment to correct location.
37106 2017-03-07 Martin Liska <mliska@suse.cz>
37108 PR middle-end/68270
37109 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
37110 Use array_at_struct_end_p instead of DECL_CHAIN (field).
37111 (chkp_narrow_bounds_for_field): Likewise.
37112 (chkp_parse_array_and_component_ref): Pass one more argument to
37115 2017-03-07 Richard Biener <rguenther@suse.de>
37117 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
37120 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
37122 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
37123 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
37125 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37128 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
37129 to end of description.
37130 (PARAM_MAX_STORES_TO_MERGE): Likewise.
37132 2017-03-07 Jakub Jelinek <jakub@redhat.com>
37134 PR rtl-optimization/79901
37135 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
37137 (*avx512f_<code><mode>3<mask_name>): ... this.
37138 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
37139 iterator instead of VI8_AVX2_AVX512BW.
37141 PR rtl-optimization/79901
37142 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
37143 min/max expander, expand it using expand_vec_cond_expr.
37146 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
37149 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37152 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
37153 to void * for PCH reasons.
37154 * dwarf2out.c (output_loc_operands, output_die): Cast
37155 v.val_vec.array to unsigned char *.
37157 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
37160 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
37163 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
37165 PR rtl-optimization/79571
37166 * lra-constraints.c (process_alt_operands): Calculate static
37167 reject and subtract it from overall when only addresses will be
37170 2017-03-06 Julia Koval <julia.koval@intel.com>
37173 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
37174 incoming stack boundary to 128 for 64-bit targets.
37176 2017-03-06 Richard Biener <rguenther@suse.de>
37178 PR tree-optimization/79894
37179 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
37180 to NULL after folding it.
37182 2017-03-06 Richard Biener <rguenther@suse.de>
37184 PR tree-optimization/79824
37185 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
37186 check disabling peeling for gaps.
37188 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
37190 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
37191 attributes): Document gettimeofday.
37193 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37195 * config/s390/s390.c (s390_option_override_internal): Set
37196 PARAM_MIN_VECT_LOOP_BOUND
37198 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37200 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
37201 * config/s390/s390.md: Likewise.
37203 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37206 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
37207 (<avx2_avx512>_perm<mode>): Rename to ...
37208 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
37210 (<avx512>_perm<mode>_mask): Rename to ...
37211 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
37213 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
37214 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
37215 instead of VI8F_256_512.
37216 (avx512f_perm<mode>): New define_expand.
37217 (avx512f_perm<mode>_mask): Likewise.
37218 (avx512f_perm<mode>_1<mask_name>): New define_insn.
37219 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
37221 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37223 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
37224 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
37226 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
37228 2017-03-06 Martin Liska <mliska@suse.cz>
37231 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
37232 when having a SSA NAME w/o VAR_DECL assigned to it.
37234 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37236 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
37237 msa_dpsub_<su>_d): Fix MODE for vec_select.
37239 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37241 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
37243 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
37245 2017-03-06 Richard Biener <rguenther@suse.de>
37247 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
37248 * plugin.c (register_plugin_info): Likewise.
37249 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
37251 2017-03-05 Jakub Jelinek <jakub@redhat.com>
37253 * config/i386/sse.md (sse_storehps, sse_storelps,
37254 avx_<castmode><avxsizesuffix>_<castmode>,
37255 avx512f_<castmode><avxsizesuffix>_<castmode>,
37256 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
37257 in condition that at least one operand is not a MEM.
37259 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37261 PR middle-end/79805
37262 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
37263 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
37265 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
37266 gimple_call_nothrow_p flag based on whether original builtin can throw.
37267 If it can, emit following stmts on the fallthrough edge.
37268 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
37269 don't create new bb if inserting just debug stmts on the edge, try to
37270 insert them on the fallthru bb or just reset debug stmts.
37272 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
37275 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
37276 restore recog_data (including the operand rtxes inside it) around
37277 the call to get_insn_template.
37279 2017-03-03 Martin Sebor <msebor@redhat.com>
37281 PR tree-optimization/79699
37282 * context.c (context::~context): Free MPFR caches to avoid
37283 a memory leak on program exit.
37285 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37287 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
37288 Use wide_int::ulow () instead of .elt (0).
37290 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37292 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
37293 (*pushxf): Limit oF constraint to 32bit targets and add oC
37294 constraint for 64bit targets.
37295 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
37296 (*pushdf): Change rmF constraint to rmC.
37298 2017-03-03 Martin Liska <mliska@suse.cz>
37300 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
37301 Remove unused variable.
37303 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37306 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
37307 is a memory operand, increase num_memory.
37308 (ix86_expand_args_builtin): Likewise.
37310 2017-03-03 Jan Hubicka <jh@suse.cz>
37313 * ipa-devirt.c (maybe_record_node): Properly handle
37314 __cxa_pure_virtual visibility.
37316 2017-03-03 Martin Liska <mliska@suse.cz>
37318 PR tree-optimization/79803
37319 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
37321 (pass_loop_prefetch::execute): Disabled optimization if an
37322 assumption about L1 cache size is not met.
37324 2017-03-03 Martin Liska <mliska@suse.cz>
37326 PR rtl-optimization/79574
37327 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
37328 (hash_scan_set): Likewise.
37329 (dump_hash_table): Likewise.
37330 (hoist_code): Likewise.
37332 2017-03-03 Richard Biener <rguenther@suse.de>
37334 * fixed-value.c (fixed_from_string): Restore use of elt (1)
37335 in place of uhigh ().
37336 (fixed_convert_from_real): Likewise.
37338 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37341 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
37343 2017-03-03 Richard Biener <rguenther@suse.de>
37345 PR middle-end/79818
37346 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
37347 TYPE_OVERFLOW_UNDEFINED check.
37349 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37351 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
37353 (vector_ae_<mode>_p): Likewise.
37354 (vector_nez_<mode>_p): Likewise.
37355 (vector_ne_v2di_p): Likewise.
37356 (vector_ae_v2di_p): Likewise.
37357 (vector_ne_<mode>_p): Likewise.
37358 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
37360 (vsx_tsqrt<mode>2_fe): Likewise.
37362 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
37365 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
37367 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37369 PR rtl-optimization/79780
37370 * cprop.c (one_cprop_pass): When second and further conditional trap
37371 in a single basic block is turned into an unconditional trap, turn it
37372 into a deleted note to avoid RTL verification failures.
37374 2017-03-02 Richard Biener <rguenther@suse.de>
37376 * fold-const.c (const_binop): Use ulow () instead of elt (0).
37378 2017-03-02 Richard Biener <rguenther@suse.de>
37380 PR tree-optimization/79345
37382 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
37383 param and abort the walk, returning -1 if it is hit.
37384 (walk_aliased_vdefs): Take a limit param and pass it on.
37385 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
37386 defaulting to 0 and return a signed int.
37387 * tree-ssa-uninit.c (struct check_defs_data): New struct.
37388 (check_defs): New helper.
37389 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
37390 about uninitialized memory.
37391 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
37392 bogus uninitialized warning.
37393 (fixed_convert_from_real): Likewise.
37395 2017-03-02 Bin Cheng <bin.cheng@arm.com>
37397 PR tree-optimization/66768
37398 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
37399 iv_use if base object can't be determined.
37401 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37403 PR tree-optimization/79345
37404 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
37405 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
37406 (get_pattern_stats): Initialize it.
37407 * genemit.c (gen_expand): Verify match_scratch numbers come after
37408 match_operand/match_dup numbers.
37409 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
37410 match_scratch numbers.
37411 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
37413 * config/s390/s390.md (trunctdsd2): Likewise.
37415 2017-03-02 Richard Biener <rguenther@suse.de>
37417 * wide-int.h (wide_int_storage::operator=): Implement in terms
37420 2017-03-02 Richard Biener <rguenther@suse.de>
37422 PR tree-optimization/79777
37423 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
37424 the to insert expression to sth existing.
37426 2017-03-01 Martin Sebor <msebor@redhat.com>
37428 PR middle-end/79692
37429 * gimple-ssa-sprintf.c
37430 (directive::known_width_and_precision): New function.
37431 (format_integer): Use it.
37432 (get_mpfr_format_length): Consider the full range of precision
37433 when computing %g output with the # flag. Set the likely byte
37434 count to 3 rather than 1 when precision is indeterminate.
37435 (format_floating): Correct the lower bound of precision.
37437 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37439 * doc/invoke.texi: Document default code model for 64-bit Linux.
37441 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37444 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
37445 udiv rather than div since input pattern is unsigned.
37447 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
37449 * config/i386/i386.c (print_reg): Warn for values of
37450 unsupported size in integer register.
37452 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
37455 * config/rs6000/predicates.md (current_file_function_operand): Do
37456 not allow self calls to be local if the function is replaceable.
37458 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
37461 * config/rs6000/altivec.h (vec_ctz and others): Change the
37462 preprocessor macro that controls conditional compilation from
37463 _ARCH_PWR9 to __POWER9_VECTOR__.
37464 (vec_all_ne): Change parameterization of __altivec_scalar_pred
37465 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
37466 control (instead of _ARCH_PWR9 control) so that template
37467 definition uses power9-specific function.
37468 (vec_any_eq): Likewise.
37469 (vec_all_ne): Change macro definition to use a power9-specific
37470 expansion under #ifdef __POWER9_VECTOR__ control (instead of
37471 _ARCH_PWR9 control).
37472 (vec_any_eq) Likewise.
37473 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
37474 expansion for CMPNEF to remove support for xvcmpnesp instruction.
37475 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
37476 support for xvcmpnedp instruction.
37477 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
37478 macro expansion so that Power9 implementation of vec_all_ne does
37479 not use the AltiVec predicate framework.
37480 (VCMPNEH_P): Likewise.
37481 (VCMPNEW_P): Likewise.
37482 (VCMPNED_P): Likewise.
37483 (VCMPNEFP_P): Likewise.
37484 (VCMPNEDP_P): Likewise.
37485 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
37486 implementation of vec_any_eq to not use AltiVec predicate
37488 (VCMPAEH_P): Likewise.
37489 (VCMPAEW_P): Likewise.
37490 (VCMPAED_P): Likewise.
37491 (VCMPAEFP_P): Likewise.
37492 (VCMPAEDP_P): Likewise.
37493 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
37494 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
37495 not use the AltiVec predicate framework.
37496 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
37497 of vec_any_eq to not use AltiVec predicate framework.
37498 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
37499 support for predefined __POWER9_VECTOR__ macro to indicate that
37500 Power9 instruction selection is enabled.
37501 (altivec_overloaded_builtins): Remove extraneous
37502 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
37503 function argument types RS6000_BTI_bool_V16QI and
37504 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
37505 entry for overloaded function argument types RS6000_BTI_bool_V4SI
37506 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
37507 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
37508 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
37509 Power9 for implementations of vec_cmpne. Change the signature for
37510 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
37511 (representing vec_all_ne) to remove the previously described first
37512 argument of type RS6000_BTI_INTSI, as this was an artifact of
37513 reliance on the AltiVec predicate framework, which is no longer
37514 used in the implementation of these functions. Add
37515 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
37516 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
37517 since, unlike the AltiVec predicate framework implementation, we
37518 do not share function descriptors between vec_alle and vec_anyeq.
37519 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
37520 set of modes that receive special treatment even when
37521 TARGET_P9_VECTOR is true. The special treatment emits code that
37522 does not depend on Power9 instructions.
37523 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
37524 define_expand to not rely on AltiVec predicate framework.
37525 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37527 (vector_ne_v2di_p): Change this define_expand to not rely on
37528 AltiVec predicate framework.
37529 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
37531 (vector_ne_<mode>_p): Change this define_expand to not rely on
37532 AltiVec predicate framework.
37533 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37535 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
37536 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
37537 define_insn pattern.
37538 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
37539 define_insn pattern because the xvcmpne<VSs>. instruction is not
37541 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
37542 instruction is not supported.
37544 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37546 * config/nvptx/nvptx.c: Include intl.h.
37548 2017-03-01 Martin Jambor <mjambor@suse.cz>
37551 * ipa-prop.h (ipa_bits): Removed field known.
37552 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
37553 to pointers. Adjusted their comments to warn about their sharing.
37554 (ipcp_transformation_summary): Change bits to a vector of pointers.
37555 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
37556 (ipa_get_ipa_bits_for_value): Declare.
37557 * tree-vrp.h (value_range): Mark as GTY((for_user)).
37558 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
37559 (ipa_bits_hash_table): Likewise.
37560 (ipa_vr_ggc_hash_traits): Likewise.
37561 (ipa_vr_hash_table): Likewise.
37562 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
37563 being pointers and vr_known being removed.
37564 (ipa_set_jf_unknown): Likewise.
37565 (ipa_get_ipa_bits_for_value): New function.
37566 (ipa_set_jfunc_bits): Likewise.
37567 (ipa_get_value_range): New overloaded functions.
37568 (ipa_set_jfunc_vr): Likewise.
37569 (ipa_compute_jump_functions_for_edge): Use the above functions to
37570 construct bits and vr parts of jump functions.
37571 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
37572 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37574 (ipcp_grow_transformations_if_necessary): Also allocate
37575 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37577 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
37578 them. Fix too long lines.
37579 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
37580 vr_known being removed.
37581 (ipa_read_jump_function): Use new setter functions to construct bits
37582 and vr parts of jump functions or set them to NULL.
37583 (write_ipcp_transformation_info): Adjust for bits being pointers.
37584 (read_ipcp_transformation_info): Likewise.
37585 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
37587 Include gt-ipa-prop.h.
37588 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
37590 (ipcp_store_bits_results): Likewise.
37591 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
37592 Do not write to existing jump functions but use a temporary instead.
37594 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37597 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
37598 attempt to use its first operand as BIT_FIELD_REF base.
37600 2017-03-01 Richard Biener <rguenther@suse.de>
37602 PR middle-end/79721
37603 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
37604 interpolating formula in wrapping arithmetic.
37605 (chrec_apply): Convert chrec_evaluate return value to wanted type.
37607 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37609 PR tree-optimization/79734
37610 * tree-vect-generic.c (expand_vector_condition): Optimize
37611 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
37612 Handle VEC_COND_EXPR where comparison has different inner width from
37613 type's inner width.
37615 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
37617 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
37618 markup, and similar issues. Remove @opindex entries for things
37619 that aren't options. Add missing -mmpy-option entries.
37621 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37623 PR tree-optimization/79737
37624 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
37625 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
37626 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
37627 instead of byte_size. Formatting fix.
37628 (shift_bytes_in_array_right): Formatting fix.
37630 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
37633 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
37634 condition on optimize for the leaf function test.
37636 2017-02-28 Martin Liska <mliska@suse.cz>
37639 * read-rtl-function.c (function_reader::handle_unknown_directive):
37640 Bail out when one uses -flto.
37642 2017-02-28 Martin Liska <mliska@suse.cz>
37644 * common.opt: Replace space with tabular for options of <number>
37646 * config/i386/i386.opt: Show <number> value for
37647 -mlarge-data-threshold.
37648 * opts.c (print_filtered_help): Do not display number in hexadecimal
37651 2017-02-28 Martin Liska <mliska@suse.cz>
37653 * common.opt: Fix --help=option -Q for options which are of
37656 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
37658 * config/i386/i386.c (print_reg): Error out for values
37659 of 8-bit size in invalid integer register.
37661 2017-02-28 Martin Sebor <msebor@redhat.com>
37663 PR tree-optimization/79691
37664 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
37666 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37669 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
37670 gcc_unreachable with output_operand_lossage.
37672 2017-02-28 Richard Biener <rguenther@suse.de>
37674 PR tree-optimization/79740
37675 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
37677 (visit_nary_op): Insert the nary into the hashtable if we
37678 pattern-matched sth.
37679 * tree-ssa-pre.c (eliminate_insert): Robustify.
37681 2017-02-28 Richard Biener <rguenther@suse.de>
37683 PR middle-end/79731
37684 * fold-const.c (decode_field_reference): Reject out-of-bound
37687 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37689 * config/i386/i386.c: Include intl.h.
37690 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
37691 instead of just cond ? "..." : "...".
37692 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
37693 * coverage.c (read_counts_file): Likewise.
37694 * omp-offload.c: Include intl.h.
37695 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
37696 of just cond ? "..." : "...".
37697 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
37698 of just cond ? "..." : "...".
37700 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
37703 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
37705 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
37706 'tune for' CPU name.
37707 * config/arm/arm-cpu-data.h: Regenerated.
37709 2017-02-28 Richard Biener <rguenther@suse.de>
37711 PR tree-optimization/79732
37712 * tree-inline.c (expand_call_inline): Do not shadow var.
37714 2017-02-28 Richard Biener <rguenther@suse.de>
37716 PR tree-optimization/79723
37717 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
37718 address-space properly.
37720 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
37722 * doc/optinfo.texi (Optimization groups): Fix option used for
37724 * doc/invoke.texi (-fopt-info): Document "omp".
37725 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
37726 (OPTGROUP_ALL): Add OPTGROUP_OMP.
37727 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
37728 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
37729 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
37731 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
37733 * dumpfile.c (optgroup_options): Instead of "openmp", associate
37734 OPTGROUP_OMP with "omp".
37736 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
37739 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
37740 for arithmetic shift of unsigned V2DI.
37742 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
37744 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
37745 arc/linux.h headers.
37746 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
37747 (LINK_SPEC): Likewise.
37748 (ARC_TLS_EXTRA_START_SPEC): Likewise.
37749 (EXTRA_SPECS): Likewise.
37750 (STARTFILE_SPEC): Likewise.
37751 (ENDFILE_SPEC): Likewise.
37752 (LIB_SPEC): Likewise.
37753 (TARGET_SDATA_DEFAULT): Likewise.
37754 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
37755 (MULTILIB_DEFAULTS): Likewise.
37756 (DWARF2_UNWIND_INFO): Likewise.
37757 * config/arc/big.h: New file.
37758 * config/arc/elf.h: Likewise.
37759 * config/arc/linux.h: Likewise.
37760 * config/arc/t-uClibc: Remove.
37762 2017-02-27 Bin Cheng <bin.cheng@arm.com>
37764 PR tree-optimization/77536
37765 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
37766 (tree_transform_and_unroll_loop): Use above function to compute the
37767 estimated niter of unrolled loop and use it when scaling profile.
37768 Also use count info rather than frequency if it's non-zero.
37769 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
37770 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
37771 (vect_transform_loop): Call above function.
37773 2017-02-27 Richard Biener <rguenther@suse.de>
37775 PR tree-optimization/45397
37776 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
37777 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
37778 (visit_nary_op): Add pattern matching for CSEing sign-changed
37779 or truncated operations with wider ones.
37781 2017-02-27 Richard Biener <rguenther@suse.de>
37783 PR tree-optimization/79690
37784 * tree-vect-stmts.c (vectorizable_store): Use vector type
37785 built from the DR with address-space.
37787 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
37789 * doc/invoke.texi (Optimize Options): Refine the description
37790 of asan-use-after-return.
37792 2017-02-25 Alan Modra <amodra@gmail.com>
37794 PR rtl-optimization/79584
37795 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37796 base, not ad->base_term, the reg within base. Remove assertion
37797 that ad->base == ad->base_term. Replace gen_int_mode using
37798 bogus mode with const0_rtx.
37800 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37802 PR middle-end/79396
37803 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37804 FMA_EXPR like tcc_binary or tcc_unary.
37806 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37809 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37811 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37812 (output_loc_operands): Handle DW_OP_call_ref and
37813 DW_OP_GNU_variable_value.
37814 (struct variable_value_struct): New type.
37815 (struct variable_value_hasher): Likewise.
37816 (variable_value_hash): New variable.
37817 (string_types): Remove.
37818 (copy_loc_descr): New function.
37819 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37820 (prepend_loc_descr_to_each): New function.
37821 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37822 instead of add_loc_descr_to_each if the first argument is single
37823 location list and the second has multiple.
37824 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37825 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37826 when looking for variable value which doesn't have other location info.
37827 (loc_list_from_tree): Formatting fix.
37828 (gen_array_type_die): Simplify DW_AT_string_length handling.
37829 (adjust_string_types): Remove.
37830 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37831 string_types. Call resolve_variable_values.
37832 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37833 (resolve_addr_in_expr): Likewise. Add A argument.
37834 (copy_deref_exprloc): Remove deref argument. Adjust for the
37835 original expression being DW_OP_GNU_variable_value with optionally
37836 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37837 optionally after it.
37838 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37839 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37840 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37841 (variable_value_hasher::hash, variable_value_hasher::equal): New
37843 (resolve_variable_value_in_expr, resolve_variable_value,
37844 resolve_variable_values, note_variable_value_in_expr,
37845 note_variable_value): New functions.
37846 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37849 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37852 * opts.h (handle_generated_option): Add GENERATED_P argument.
37853 * opts-common.c (handle_option): Adjust function comment.
37854 (handle_generated_option): Add GENERATED_P argument, pass it to
37856 (control_warning_option): Pass false to handle_generated_option
37858 * opts.c (maybe_default_option): Pass true to handle_generated_option
37860 * optc-gen.awk: Likewise.
37862 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37864 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37865 a REG, look at the REG it is a SUBREG of.
37866 (splitter for cmpeqsi_t): Ditto.
37868 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37870 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37871 the special USEs with the pattern of the insn, not the insn itself.
37873 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37876 * doc/invoke.texi: Document -mload-store-pairs.
37878 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37879 Sandra Loosemore <sandra@codesourcery.com>
37881 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37882 argument isn't a CONST_INT.
37883 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37884 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37885 (nios2_validate_compare): Bypass alternate compare logic if *op2
37886 is not a CONST_INT.
37887 (ldstwm_operation_p): Return false if first_base is not a REG or
37888 if first_offset is not a CONST_INT.
37890 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37892 * config/cris/cris.md: Use correct operand in a define_peephole2.
37894 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37896 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37898 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37900 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37901 this_insn if it is an INSN or JUMP_INSN.
37902 (force_offsettable): Look at base, not at addr.
37903 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37904 on things that aren't necessarily CONST_INTs.
37906 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37908 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37909 -mfpmath=sse is the default also for x86-32 targets with SSE2
37910 instruction set when @option{-ffast-math} is enabled
37912 2017-02-24 Jeff Law <law@redhat.com>
37914 PR rtl-optimizatoin/79286
37915 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37918 2017-02-24 Richard Biener <rguenther@suse.de>
37920 PR tree-optimization/79389
37921 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37924 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37926 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37927 function comment to reflect reality.
37928 (loop_exits_before_overflow): Fix typo in function description.
37930 2017-02-24 Richard Biener <rguenther@suse.de>
37932 PR tree-optimization/79389
37933 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37934 properly that a threading opportunity exists. Detect conditional
37935 copy/constant propagation opportunities.
37937 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37939 * config/visium/visium.md (type): Add trap.
37940 (b): New mode attribute.
37941 (*btst): Rename into...
37942 (*btst<mode>): ...this and adjust.
37943 (*cbranchsi4_btst_insn): Rename into...
37944 (*cbranch<mode>4_btst_insn): ...this and adjust.
37945 (trap): New define_insn.
37947 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37949 PR tree-optimization/79389
37950 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37951 (noce_reversed_cond_code): New function.
37952 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37953 reversed_comparison_code. Formatting fix.
37954 (noce_try_store_flag): Test rev_cond != NULL in addition to
37955 reversed_comparison_code.
37956 (noce_try_store_flag_constants): Likewise.
37957 (noce_try_store_flag_mask): Likewise.
37958 (noce_try_addcc): Use rev_cond if non-NULL instead of
37959 reversed_comparison_code.
37960 (noce_try_cmove_arith): Likewise. Formatting fixes.
37961 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37962 (noce_find_if_block): Initialize rev_cond.
37963 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37964 instead of false as last argument never attempt to reverse it
37967 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37969 PR tree-optimization/79663
37970 * tree-predcom.c (combine_chains): Process refs in reverse order
37971 only for ZERO length chains, and add explaining comment.
37973 2017-02-23 Jeff Law <law@redhat.com>
37975 PR tree-optimization/79578
37976 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37977 in call to operand_equal_p.
37979 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37982 * config/i386/cpuid.h: Fix another undefined behavior.
37984 2017-02-23 Richard Biener <rguenther@suse.de>
37986 PR tree-optimization/79683
37987 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37988 vector types for data-refs.
37990 2017-02-23 Martin Liska <mliska@suse.cz>
37992 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37994 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37996 PR middle-end/79665
37997 * internal-fn.c (get_range_pos_neg): Moved to ...
37998 * tree.c (get_range_pos_neg): ... here. No longer static.
37999 * tree.h (get_range_pos_neg): New prototype.
38000 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
38001 are known to be in between 0 and signed maximum inclusive, try to
38002 expand both unsigned and signed divmod and use the cheaper one from
38005 2017-02-22 Jeff Law <law@redhat.com>
38007 PR tree-optimization/79578
38008 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
38009 to compare base operands.
38011 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
38014 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
38015 gpc_reg_operand instead of fpr_reg_operand.
38017 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
38019 * config/mips/mips.c (mips_return_in_memory): Force FP
38020 vector types to be returned in memory for o32 ABI.
38022 2017-02-22 Jakub Jelinek <jakub@redhat.com>
38024 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
38025 instead of DW_TAG_member for static data member declarations and don't
38026 set no_linkage_name for static inline data members.
38027 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
38030 2017-02-22 Martin Liska <mliska@suse.cz>
38032 * doc/invoke.texi: Replace inequality signs with square brackets
38035 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
38038 * lra-constraints.c (simplify_operand_subreg): Handle
38039 WORD_REGISTER_OPERATIONS targets.
38041 2017-02-22 Jakub Jelinek <jakub@redhat.com>
38044 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
38045 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
38046 elimination by swapping fld*.
38048 2017-02-22 Richard Biener <rguenther@suse.de>
38050 PR tree-optimization/79673
38051 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
38052 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
38053 irrelevant address-space qualifiers and avoiding a
38054 ADDR_SPACE_CONVERT_EXPR from fold_convert.
38056 2017-02-22 Richard Biener <rguenther@suse.de>
38058 PR tree-optimization/79666
38059 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
38060 to not symbolically negate if that may introduce undefined
38063 2017-02-22 Martin Liska <mliska@suse.cz>
38066 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
38067 * data-streamer-out.c (streamer_write_gcov_count_stream):
38069 * value-prof.c (stream_out_histogram_value): Make assert more
38070 precise based on type of counter.
38072 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
38075 * config/i386/i386.md (standard_x87sse_constant_load splitter):
38076 Use nonimmediate_operand instead of memory_operand for operand 1.
38077 (float-extend standard_x87sse_constant_load splitter): Ditto.
38079 2017-02-21 Jeff Law <law@redhat.com>
38081 PR tree-optimization/79621
38082 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
38083 blocks with edges to themselves.
38085 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38088 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
38089 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
38090 Use gimple_call_builtin_p.
38093 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
38094 on temporarily removed DEBUG_INSNs.
38096 PR tree-optimization/79649
38097 * tree-loop-distribution.c (classify_partition): Give up on
38098 non-generic address space loads/stores.
38100 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
38102 * doc/loop.texi (Loop manipulation): Remove nonexistent
38103 tree_ssa_loop_version from the documentation.
38104 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
38106 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38109 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
38110 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
38111 * config/rs6000/rs6000.c: Include except.h.
38112 (rs6000_expand_split_stack_prologue): Call
38113 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
38115 2017-02-21 Martin Jambor <mjambor@suse.cz>
38118 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
38119 have been analyzed.
38121 2017-02-21 Martin Jambor <mjambor@suse.cz>
38123 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
38124 for backward compatibility only.
38125 * doc/invoke.texi (Option Summary): Remove all references to
38126 -fipa-cp-alignment.
38128 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
38132 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38134 * lra-constraints.c (curr_insn_transform): Handle
38135 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38137 2017-02-21 Martin Liska <mliska@suse.cz>
38139 * config/i386/i386.opt: Replace -masm-dialect with -masm.
38141 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
38143 PR translation/79638
38144 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
38146 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
38149 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
38150 (arm_function_ok_for_sibcall): Return false for an indirect call by
38151 descriptor if all the argument registers are used.
38152 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
38153 alignment of the function.
38155 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38157 PR tree-optimization/61441
38158 * simplify-rtx.c (simplify_const_unary_operation): For
38159 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
38160 the sNaN unmodified.
38162 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
38164 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
38165 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
38166 instead of SYSTEM_HEADER_DIR.
38168 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
38169 Martin Liška <mliska@suse.cz>
38171 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
38172 Fix typos and grammar, use active voice, and clarify.
38174 2017-02-20 Marek Polacek <polacek@redhat.com>
38176 PR middle-end/79537
38177 * gimplify.c (gimplify_expr): Handle unused *&&L;.
38180 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
38182 2017-02-20 Jakub Jelinek <jakub@redhat.com>
38185 * config/i386/i386.c (ix86_expand_builtin): Handle
38186 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
38187 ix86_builtins_isa[fcode].isa as a requirement of those
38188 flags and any other flag in the bitmask.
38189 (ix86_init_mmx_sse_builtins): Use 0 instead of
38190 ~OPTION_MASK_ISA_64BIT as mask.
38191 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
38192 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
38193 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
38194 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
38196 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38199 * lra-constraints.c (split_reg): Check requested split mode
38200 is supported by the register.
38202 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38204 * lra-constraints.c (simplify_operand_subreg): Remove early
38207 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38210 * lra-constraints.c (curr_insn_transform): Tighten condition
38211 for converting SUBREG reloads from OP_OUT to OP_INOUT.
38213 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38216 * lra-constraints.c (curr_insn_transform): Handle
38217 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38219 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
38222 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
38224 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
38226 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
38229 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
38232 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38234 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
38237 2017-02-18 Jakub Jelinek <jakub@redhat.com>
38239 * final.c (last_columnnum, override_columnnum): New variables.
38240 (final_start_function): Set last_columnnum, pass it to begin_prologue
38241 hook and pass 0 to dwarf2out_begin_prologue.
38242 (final_scan_insn): Update override_columnnum. Pass last_columnnum
38243 to source_line debug hook.
38244 (notice_source_line): Compute last_columnnum and for debug_column_info
38245 return true on column changes.
38246 * debug.h (struct gcc_debug_hooks): Add column argument to
38247 source_line and begin_prologue hooks.
38248 (debug_nothing_int_charstar_int_bool): Remove prototype.
38249 (debug_nothing_int_int_charstar,
38250 debug_nothing_int_int_charstar_int_bool): New prototypes.
38251 (dwarf2out_begin_prologue): Add column argument.
38252 * debug.c (do_nothing_debug_hooks): Adjust source_line and
38253 begin_prologue hooks.
38254 (debug_nothing_int_charstar_int_bool): Remove.
38255 (debug_nothing_int_int_charstar,
38256 debug_nothing_int_int_charstar_int_bool): New functions.
38257 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
38258 through to dwarf2out_source_line.
38259 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
38260 (dwarf2out_source_line): Add column argument, emit it if requested.
38261 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
38263 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38264 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38265 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
38266 through to dwarf2out_begin_prologue.
38267 (vmsdbgout_source_line): Add column argument, pass it through to
38268 dwarf2out_source_line.
38269 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
38270 dbxout_source_line caller.
38271 (dbxout_source_line): Add column argument.
38273 * common.opt (gno-column-info, gcolumn-info): New options.
38274 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
38275 (check_die): Also test for multiple DW_AT_decl_column attributes.
38276 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
38277 DW_AT_decl_column if requested.
38278 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
38280 (gen_variable_die): Likewise.
38281 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
38282 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
38285 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
38286 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
38287 (ix86_handle_option): Handle OPT_m3dnowa.
38288 * doc/invoke.texi (-m3dnowa): Document.
38289 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
38290 -m3dnowa instead of -m3dnow -march=athlon.
38293 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
38294 instead of gcc_assert for K, r and R code checks. Formatting fixes.
38296 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38299 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
38300 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
38301 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
38302 generator for vsx_xxpermdi_<mode>_be.
38303 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
38304 force big-endian semantics.
38305 (vsx_xxpermdi_<mode>_be): New define_expand with same
38306 implementation as previous version of vsx_xxpermdi_<mode>.
38308 2017-02-17 Jakub Jelinek <jakub@redhat.com>
38310 PR tree-optimization/79327
38311 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
38312 variable, its initialization and use.
38314 2017-02-17 Julia Koval <julia.koval@intel.com>
38316 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
38317 (OPTION_MASK_ISA_PKU_UNSET): New.
38318 (ix86_handle_option): Handle -mrdpid.
38319 * config/i386/cpuid.h (bit_RDPID): New.
38320 * config/i386/driver-i386.c (host_detect_local_cpu):
38321 Detect RDPID feature.
38322 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
38323 * config/i386/i386-c.c (ix86_target_macros_internal):
38325 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
38326 (ix86_valid_target_attribute_inner_p): Add "rdpid".
38327 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
38328 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
38329 * config/i386/i386.md (define_insn "rdpid"): New.
38330 * config/i386/i386.opt Add -mrdpid.
38331 * config/i386/immintrin.h (_rdpid_u32): New.
38333 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
38335 PR rtl-optimization/79541
38336 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
38337 instead of transforming it into USE.
38339 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
38341 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
38342 If HONOR_SNANS (SFmode) force the input to a register.
38343 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
38344 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
38345 an frsp or similar insn.
38347 2017-02-17 Martin Liska <mliska@suse.cz>
38349 PR rtl-optimization/79577
38350 * params.def (selsched-max-sched-times): Increase minimum to 1.
38352 2017-02-17 Martin Liska <mliska@suse.cz>
38354 PR rtl-optimization/79574
38355 * gcse.c (want_to_gcse_p): Prevent integer overflow.
38357 2017-02-17 Martin Liska <mliska@suse.cz>
38359 PR tree-optimization/79529
38360 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
38361 ssa_defined_default_def_p to handle cases which are implicitly
38363 * tree-ssa.c (ssa_defined_default_def_p): New function.
38364 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
38365 which are implicitly defined.
38366 * tree-ssa.h (ssa_defined_default_def_p): Declare.
38368 2017-02-17 Richard Biener <rguenther@suse.de>
38370 PR middle-end/79576
38371 * params.def (max-ssa-name-query-depth): Limit to 10.
38373 2017-02-17 Richard Biener <rguenther@suse.de>
38375 PR tree-optimization/79552
38376 * tree-ssa-structalias.c (visit_loadstore): Properly verify
38379 2017-02-17 Richard Biener <rguenther@suse.de>
38382 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
38384 2017-02-17 Marek Polacek <polacek@redhat.com>
38386 PR middle-end/79536
38387 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
38388 (fold_negate_expr): New wrapper.
38390 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
38392 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
38393 Correct terminology and de-emphasize pre-standard behavior.
38395 2017-02-16 Alan Modra <amodra@gmail.com>
38397 PR rtl-optimization/79286
38398 * ira.c (def_dominates_uses): New function.
38399 (update_equiv_regs): Don't create an equivalence for insns that
38400 may trap where the register def does not dominate the use.
38402 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
38404 PR rtl-optimization/78127
38405 * lra.c (lra): Call lra_eliminate before finish the loop after
38408 2017-02-16 Richard Biener <rguenther@suse.de>
38410 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
38412 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
38413 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
38414 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
38415 (isl_val_int_from_wi): New function.
38416 (extract_affine_gmp): Rename to ...
38417 (extract_affine_wi): ... this, take a widest_int.
38418 (extract_affine_int): Just wrap extract_affine_wi.
38419 (add_param_constraints): Use isl_val_int_from_wi.
38420 (add_loop_constraints): Likewise, and extract_affine_wi.
38422 2017-02-15 Jeff Law <law@redhat.com>
38424 PR middle-end/79521
38425 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
38426 ira_init_register_move_cost_if_necessary.
38428 2017-02-15 Martin Sebor <msebor@redhat.com>
38430 PR middle-end/32003
38431 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
38432 removed in a prior commit.
38434 2017-02-15 Bin Cheng <bin.cheng@arm.com>
38436 PR tree-optimization/79347
38437 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
38438 counters during peeling.
38440 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
38442 * Makefile.in (site.exp): Remove "set ISLVER".
38444 2017-02-15 Jakub Jelinek <jakub@redhat.com>
38447 * real.c (real_from_integer): Call real_convert even for decimal.
38449 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
38452 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
38454 2017-02-14 Andrew Pinski <apinski@cavium.com>
38456 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
38457 cores and change the partno/implementer to be correct.
38458 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
38459 the 'B" as the implementer.
38460 * config/aarch64/aarch64-tune.md: Regenerate.
38462 2017-02-14 Carl Love <cel@us.ibm.com>
38464 * config/rs6000/rs6000.c: Add case statement entry to make the
38465 xvcvuxdsp built-in argument unsigned.
38466 * config/rs6000/vsx.md: Fix the source and return operand types so they
38467 match the instruction definitions from the ISA document. Fix typo
38468 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
38471 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
38474 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
38475 member early_clobber_alts.
38476 * lra-lives.c (reg_early_clobber_p): New.
38477 (process_bb_lives): Use it.
38478 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
38479 (debug_operand_data): Initialize early_clobber_alts.
38480 (setup_operand_alternative): Set up early_clobber_alts.
38481 (collect_non_operand_hard_regs): Ditto. Pass early clobber
38482 alternatives to new_insn_reg.
38483 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
38485 (lra_update_insn_regno_info): Pass the new arg.
38487 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38489 PR middle-end/79505
38490 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
38491 (new_oacc_loop_raw): Don't clear already cleared fields.
38494 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
38495 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
38496 _mm512_prefetch_i64gather_ps): New inline functions and macros.
38498 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
38501 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
38503 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
38506 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
38507 the extra instruction to the right place to store 128-bit constant
38510 2017-02-14 Martin Sebor <msebor@redhat.com>
38512 PR middle-end/79448
38513 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
38514 warning for strings of unknown length.
38516 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
38518 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
38520 2017-02-14 Jeff Law <law@redhat.com>
38523 * ira-costs.c (scan_one_insn): Initialize register move costs
38524 for pseudos seen in USE/CLOBBER insns.
38526 PR tree-optimization/79095
38527 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
38528 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
38529 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
38530 if the operands are known to be not equal, then the resulting range
38532 (intersect_ranges): If the new range is ~[0,0] and the old range is
38533 wide, then prefer ~[0,0].
38534 * tree-vrp.c (overflow_comparison_p_1): New function.
38535 (overflow_comparison_p): New function.
38536 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
38537 if NAME is used in an overflow test.
38538 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
38539 overflow check that can be expressed as an equality test, then adjust
38540 ops to be that equality test.
38542 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38544 * config/s390/s390-builtin-types.def: Remove flags argument.
38545 * config/s390/s390.c (s390_init_builtins): Likewise.
38547 2017-02-14 Martin Liska <mliska@suse.cz>
38549 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
38550 vector. Fix trailing white spaces.
38552 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
38554 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
38557 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38559 PR rtl-optimization/68664
38560 * config/arm/arm.c (arm_sched_can_speculate_insn):
38561 New function. Declare prototype.
38562 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38564 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38566 PR rtl-optimization/68664
38567 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
38569 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38571 2017-02-14 Amit Pawar <amit.pawar@amd.com>
38573 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
38574 max skip bytes for function, loop and jump.
38576 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38578 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
38579 ABS_EXPR for gimple dump.
38581 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38584 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
38586 PR tree-optimization/79408
38587 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
38588 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
38589 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
38590 also if rhs1 is INTEGER_CST.
38592 2017-02-14 Richard Biener <rguenther@suse.de>
38594 PR middle-end/79432
38595 * tree-into-ssa.c (insert_phi_nodes): When the function can
38596 have abnormal edges rewrite SSA names with broken use-def
38597 dominance out of SSA and register them for PHI insertion.
38599 2017-02-13 Martin Sebor <msebor@redhat.com>
38601 PR middle-end/79496
38602 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
38603 clearing info.nowrite flag when snprintf size argument is a range.
38605 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38607 * cprop.c (cprop_jump): Add missing space in string literal.
38608 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
38609 (get_constraint_for_component_ref): Likewise.
38610 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
38611 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
38612 * lra-constraints.c (process_alt_operands): Likewise.
38613 * ipa-inline.c (inline_small_functions): Likewise.
38614 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
38615 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
38616 * trans-mem.c (diagnose_tm_1_op): Likewise.
38617 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
38618 (grid_parallel_clauses_gridifiable): Likewise.
38620 * config/nvptx/mkoffload.c (process): Add space in between
38623 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
38624 "MOD4_SSE_REGS" and "ALL_REGS".
38626 * spellcheck.c (test_data): Add , in between "foo" and "food".
38628 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38631 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
38632 boundary crossing check and subsequent code generation agree.
38634 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38636 * config/aarch64/aarch64.c (has_memory_op): Delete.
38637 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
38640 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38642 PR rtl-optimization/79388
38643 PR rtl-optimization/79450
38644 * combine.c (distribute_notes): When removing TEM_INSN for which
38645 corresponding dest has last value recorded, invalidate that last
38648 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38650 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
38651 of explicit '@'. Add missing assembly comment marker on branch costs
38654 2017-02-13 Nathan Sidwell <nathan@acm.org>
38656 * gengtype-lex.l (<in_struct>): Add '/'.
38658 2017-02-13 Martin Liska <mliska@suse.cz>
38661 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
38663 2017-02-13 Richard Biener <rguenther@suse.de>
38665 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
38667 * configure: Re-generate.
38668 * config.in: Likewise.
38669 * graphite-dependences.c: Simplify as if
38670 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
38671 * graphite-isl-ast-to-gimple.c: Likewise.
38672 * graphite-optimize-isl.c: Likewise.
38673 * graphite-poly.c: Likewise.
38674 * graphite-sese-to-poly.c: Likewise.
38675 * graphite.h: Likewise.
38676 * toplev.c: Include isl/version.h and use isl_version () for
38677 printing the ISL version.
38678 * doc/install.texi: Update ISL requirement.
38680 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38682 * doc/standards.texi (Standards): Update reference to
38685 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38687 * doc/extend.texi (Named Address Spaces): sourceware.org now
38689 * doc/install.texi (Binaries): Ditto.
38692 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38694 * doc/cpp.texi: Replace "stringify"/"stringification" with C
38695 standard terminology "stringize"/"stringizing" throughout.
38696 * doc/cppinternals.texi: Likewise.
38698 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38700 * doc/extend.texi: Fix some spelling mistakes and typos.
38701 * doc/invoke.texi: Likewise.
38703 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38706 * params.def (inline-min-speedup) Change from 10 to 8.
38708 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38710 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
38713 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38716 * ipa-inline-analysis.c (get_minimal_bb): New function.
38717 (record_modified): Use it.
38718 (remap_edge_change_prob): Handle also ancestor functions.
38720 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
38722 * doc/contrib.texi (Contributors): Remove broken link into
38723 the Mauve CVS repository.
38725 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38727 PR middle-end/79454
38728 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
38729 result computation whenever lhs doesn't have vector mode, not
38730 just when it has BLKmode.
38732 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38734 * doc/makefile.texi (profiledbootstrap): Refer to the
38735 installation instructions only in textual form.
38737 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38740 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
38742 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38744 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
38745 (Specific): Update mingw-w64 reference.
38747 (Specific): Remove broken link to Renesas RX processor.
38749 2017-02-10 Richard Biener <rguenther@suse.de>
38751 * toplev.c (process_options): Do not mention obsolete graphite
38752 options when printing sorry message about missing graphite support.
38753 Mention -floop-nest-optimize.
38755 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
38757 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
38758 (vtst_p16): Likewise.
38759 (vtstq_p8): Likewise.
38760 (vtstq_p16): Likewise.
38762 (vtstq_p64): Likewise.
38763 * config/arm/arm_neon.h (vgetq_lane_p64): New.
38764 (vset_lane_p64): New.
38765 (vsetq_lane_p64): New.
38767 2017-02-10 Jakub Jelinek <jakub@redhat.com>
38769 PR tree-optimization/79411
38770 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
38771 stmt operands are SSA_NAMEs used in abnormal phis.
38772 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
38775 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38778 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
38781 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38783 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
38785 2017-02-09 Jakub Jelinek <jakub@redhat.com>
38787 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38791 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38792 not arbitrary TREE_CONSTANT.
38795 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38796 "omp declare target link" attribute unless is_global_var.
38797 * omp-offload.c (find_link_var_op): Likewise.
38799 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38800 Chung-Lin Tang <cltang@codesourcery.com>
38802 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38804 (gimplify_adjust_omp_clauses): Don't delete TILE.
38805 (gimplify_omp_for): Deal with TILE.
38806 * internal-fn.c (expand_GOACC_TILE): New function.
38807 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38809 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38810 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38812 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38813 avoid DIV for outermost collapse var.
38814 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38815 Remove out of date comments, fix whitespace.
38816 * omp-general.c (omp_extract_for_data): Deal with tiling.
38817 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38818 adjust OLF_DIM_BASE value.
38819 (struct omp_for_data): Add tiling field.
38820 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38821 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38822 for auto loops. Remove default auto determining, moved to
38823 oacc_loop_fixed_partitions.
38824 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38825 stmts, add e_mask field.
38826 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38827 (oacc_thread_numbers): Use oacc_dim_call.
38828 (oacc_xform_tile): New.
38829 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38830 (finish_oacc_loop): Adjust for ifns vector.
38831 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38832 add case for GOACC_TILE fns.
38833 (oacc_loop_xform_loop): Delete.
38834 (oacc_loop_process): Iterate over call list directly, and add
38835 handling for GOACC_TILE fns.
38836 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38838 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38839 vector partitioning to outer loops. Assign 2 partitions to loops
38840 when available. Add TILE handling.
38841 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38842 (execite_oacc_device_lower): Process GOACC_TILE fns,
38843 ignore unknown specs.
38844 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38845 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38846 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38848 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38850 * configure.ac (ACX_BUGURL): Update.
38851 * configure: Regenerate.
38853 2017-02-09 Richard Biener <rguenther@suse.de>
38855 PR tree-optimization/69823
38856 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38857 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38859 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38861 * config/arc/arc-c.def: Add __NPS400__ definition.
38862 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38863 (TARGET_NPS400): Define.
38865 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38867 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38869 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38870 pointer, arch_info.
38871 (arc_cpu_types): Fill the arch_info field with a pointer into the
38872 arc_arch_types table.
38873 (arc_selected_cpu): Declare.
38874 * config/arc/arc.c (arc_selected_cpu): Make global.
38875 (arc_selected_arch): Delete.
38876 (arc_base_cpu): Delete.
38877 (arc_override_options): Remove references to deleted variables,
38878 update access to arch information.
38879 (ARC_OPT): Update access to arch information.
38880 (ARC_OPTX): Likewise.
38881 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38882 (TARGET_ARC600): Update access to arch information.
38883 (TARGET_ARC601): Likewise.
38884 (TARGET_ARC700): Likewise.
38885 (TARGET_EM): Likewise.
38886 (TARGET_HS): Likewise.
38887 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38890 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38893 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38894 condition/operands for integer GE/LE/GEU/LEU operations.
38896 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38898 PR translation/79397
38899 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38902 2017-02-08 Martin Jambor <mjambor@suse.cz>
38905 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38906 whether allocation happened.
38907 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38908 nothing was allocated.
38910 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38912 PR tree-optimization/79408
38913 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38914 constant, but SSA_NAME with a known integer range, use the minimum
38915 of that range instead of op1 to determine if modulo can be replaced
38916 with its first operand.
38918 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38920 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38922 2017-02-08 Richard Biener <rguenther@suse.de>
38924 PR tree-optimization/71824
38925 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38926 Check all loops contained in the merged region.
38928 2017-02-07 Andrew Pinski <apinski@cavium.com>
38930 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38932 2017-02-07 Andrew Pinski <apinski@cavium.com>
38934 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38935 (thunderxt88): Likewise.
38936 (thunderxt81): Disable LSE and change v8.1 to v8.
38937 (thunderxt83): Likewise.
38939 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38940 Richard Biener <rguenther@suse.de>
38942 PR middle-end/79399
38943 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38944 type from int to size_t.
38945 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38947 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38949 PR rtl-optimization/79386
38950 * cprop.c (bypass_conditional_jumps): Initialize
38951 bypass_last_basic_block already before splitting bbs after
38952 unconditional traps...
38953 (bypass_conditional_jumps): ... rather than here.
38956 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38957 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38958 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38959 fix -masm=intel patterns.
38961 2017-02-07 Richard Biener <rguenther@suse.de>
38963 PR tree-optimization/79256
38964 PR middle-end/79278
38965 * builtins.c (get_object_alignment_2): Use min_align_of_type
38966 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38967 and ADJUST_FIELD_ALIGN.
38969 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38971 * doc/tm.texi: Regenerate.
38972 * stor-layout.c (layout_decl): Adjust.
38973 (update_alignment_for_field): Likewise.
38974 (place_field): Likewise.
38975 (min_align_of_type): Likewise.
38976 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38977 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38978 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38979 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38980 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38981 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38982 * config/i386/i386.c (x86_field_alignment): Likewise.
38983 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38984 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38985 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38986 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38987 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38988 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38992 2017-01-30 Richard Biener <rguenther@suse.de>
38994 PR tree-optimization/79256
38995 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38996 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38999 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
39001 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
39002 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
39003 builtins to SImode and emit a zero-extend, if necessary.
39005 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
39007 * docs/invoke.texi (RISC-V Options): Alphabetize.
39009 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
39011 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
39014 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
39016 * config/riscv/riscv.c: New file.
39017 * common/config/riscv/riscv-common.c: Likewise.
39018 * config.gcc: Likewise.
39019 * config/riscv/constraints.md: Likewise.
39020 * config/riscv/elf.h: Likewise.
39021 * config/riscv/generic.md: Likewise.
39022 * config/riscv/linux.h: Likewise.
39023 * config/riscv/multilib-generator: Likewise.
39024 * config/riscv/peephole.md: Likewise.
39025 * config/riscv/pic.md: Likewise.
39026 * config/riscv/predicates.md: Likewise.
39027 * config/riscv/riscv-builtins.c: Likewise.
39028 * config/riscv/riscv-c.c: Likewise.
39029 * config/riscv/riscv-ftypes.def: Likewise.
39030 * config/riscv/riscv-modes.def: Likewise.
39031 * config/riscv/riscv-opts.h: Likewise.
39032 * config/riscv/riscv-protos.h: Likewise.
39033 * config/riscv/riscv.h: Likewise.
39034 * config/riscv/riscv.md: Likewise.
39035 * config/riscv/riscv.opt: Likewise.
39036 * config/riscv/sync.md: Likewise.
39037 * config/riscv/t-elf-multilib: Likewise.
39038 * config/riscv/t-linux: Likewise.
39039 * config/riscv/t-linux-multilib: Likewise.
39040 * config/riscv/t-riscv: Likewise.
39041 * configure.ac: Likewise.
39042 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
39043 Waterman as RISC-V maintainers.
39044 * doc/install.texi: Add RISC-V entries.
39045 * doc/invoke.texi: Add RISC-V options section.
39046 * doc/md.texi: Add RISC-V constraints section.
39047 * configure: Regenerated.
39049 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
39052 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
39053 false values to be constant vectors with all 0 or all 1 bits set.
39054 (vcondu<mode><mode>): Likewise.
39055 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
39057 (fpmask_comparison_operator): Update comment.
39058 (vecint_comparison_operator): New predicate.
39059 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
39060 vector conditionals when the true and false values are constant
39061 vectors with all 0 bits or all 1 bits set.
39063 2017-02-06 Martin Sebor <msebor@redhat.com>
39065 PR tree-optimization/79376
39066 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
39068 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
39070 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
39071 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
39072 to simplify split condition.
39074 2017-02-06 Jakub Jelinek <jakub@redhat.com>
39076 * omp-expand.c (oxpand_omp_atomic_fetch_op,
39077 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
39080 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
39082 PR rtl-optimization/68664
39083 * target.def (can_speculate_insn): New hook.
39084 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
39085 * doc/tm.texi: Regenerate.
39086 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
39087 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
39088 (rs6000_sched_can_speculate_insn): New function.
39090 2017-02-06 Jakub Jelinek <jakub@redhat.com>
39092 PR tree-optimization/79284
39093 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
39094 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
39095 vectorizable_mask_load_store, vectorizable_operation,
39096 vect_is_simple_cond, get_same_sized_vectype): Use it instead
39097 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
39098 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
39099 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
39100 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
39101 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
39102 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
39103 is_gimple_assign (stmt). Replace another such test with
39104 is_gimple_assign (stmt).
39106 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
39109 * config/avr/avr.c (rtl-iter.h): Include it.
39110 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
39111 (avr_legitimate_combined_insn): ...and implementation.
39113 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
39115 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
39116 * config/s390/s390.c (s390_const_operand_ok)
39117 (s390_canonicalize_comparison, s390_extract_part)
39118 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
39119 (s390_contiguous_bitmask_p, s390_rtx_costs)
39120 (legitimize_pic_address): Likewise.
39121 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
39122 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
39123 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
39124 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
39125 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
39127 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
39129 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
39130 REGNO($0) == REGNO($1).
39132 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39134 * config/s390/linux.h(SIZE_TYPE): Add comment.
39136 2017-02-06 Julian Brown <julian@codesourcery.com>
39137 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39138 Virendra Pathak <virendra.pathak@broadcom.com>
39140 * config/aarch64/aarch64-cores.def: Change the scheduler
39142 * config/aarch64/aarch64.md: Include thunderx2t99.md.
39143 * config/aarch64/thunderx2t99.md: New file.
39145 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39147 * doc/standards.texi (Go Language): Update link to language
39150 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
39152 * tree-eh.c (lower_resx): Sanitize profile.
39153 (cleanup_empty_eh_move_lp): Likewise.
39155 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
39158 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
39160 * cfgloopmanip.h (loop_version): Update prototype.
39161 * modulo-sched.c (sms_schedule): Update call of loop_version.
39162 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
39163 * tree-parloops.c (gen_parallel_loop): Likewise.
39164 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
39165 * tree-ssa-loop-split.c (split_loop): Likewise.
39166 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
39167 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
39169 2017-02-05 Martin Liska <mliska@suse.cz>
39172 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
39174 (print_operand_address): Initialize a struct to zero.
39176 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39178 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
39179 garbage collector only in textual form.
39181 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39183 * doc/extend.texi (x86 specific memory model extensions for
39184 transactional memory): Simplify a phrase.
39186 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
39189 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
39190 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
39191 (atomic_storedi_1): Likewise.
39193 2017-02-04 Jakub Jelinek <jakub@redhat.com>
39195 PR tree-optimization/79338
39196 * tree-parloops.c (gather_scalar_reductions): Don't call
39197 vect_analyze_loop_form for loop->inner before destroying loop's
39200 2017-02-03 Martin Sebor <msebor@redhat.com>
39202 PR tree-optimization/79327
39203 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
39204 when precision has resulted in leading zeros.
39205 (format_integer): Adjust the likely counter to assume an unknown
39206 argument that may be zero is non-zero.
39208 2017-02-03 Jason Merrill <jason@redhat.com>
39211 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
39212 avoid copying non-taken branch.
39214 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39216 PR tree-optimization/79340
39217 * tree-vect-loop.c (vectorizable_reduction): Release
39218 vec_defs elements after safe_splicing them into other vectors.
39221 PR tree-optimization/79327
39222 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
39223 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
39225 (format_integer): Use wide_int_to_tree instead of build_int_cst
39226 + to_?hwi. If argmin is NULL, just set argmin and argmax to
39227 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
39228 of shortest and longest sequence.
39230 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
39232 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
39233 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
39235 2017-02-03 Walter Lee <walt@tilera.com>
39238 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
39239 after initial stackframe link reg save.
39240 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
39242 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39245 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
39246 wu for stxssp alternative.
39248 2017-02-03 Martin Sebor <msebor@redhat.com>
39250 PR tree-optimization/79352
39251 * gimple-fold.c (get_range_strlen): Add argument.
39252 (get_range_strlen): Change return type to bool.
39253 (get_maxval_strlen): Pass in a dummy argument.
39254 * gimple-fold.h (get_range_strlen): Change return type to bool.
39255 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
39256 * tree.h (array_at_struct_end_p): Add argument.
39257 * tree.c (array_at_struct_end_p): Handle it.
39259 2017-02-03 Martin Liska <mliska@suse.cz>
39262 * multiple_target.c (create_dispatcher_calls): Redirect edge
39263 from a caller of a dispatcher.
39264 (expand_target_clones): Make the clones local.
39265 (ipa_target_clone): Do both target clones and resolvers.
39266 (ipa_dispatcher_calls): Remove the pass.
39267 (pass_dispatcher_calls::gate): Likewise.
39268 (make_pass_dispatcher_calls): Likewise.
39269 * passes.def (pass_target_clone): Put as very first IPA early
39272 2017-02-03 Martin Liska <mliska@suse.cz>
39274 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
39275 in case of a function with ifunc attribute.
39277 2017-02-03 Martin Liska <mliska@suse.cz>
39279 * cgraph.c (cgraph_node::dump): Dump function version info.
39280 * symtab.c (symtab_node::dump_base): Add missing new line.
39282 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39284 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
39285 (ifcombine_ifandif): Use it.
39287 2017-02-03 Martin Liska <mliska@suse.cz>
39289 * doc/invoke.texi: Document default value for
39290 use-after-scope-direct-emission-threshold.
39292 2017-02-03 Martin Liska <mliska@suse.cz>
39294 PR tree-optimization/79339
39295 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
39296 (format_floating): Likewise.
39298 2017-02-03 Martin Liska <mliska@suse.cz>
39301 * ipa-prop.c (ipa_node_params_t::insert): Remove current
39303 (ipa_node_params_t::remove): Likewise.
39304 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
39305 initialization from removed ipa_node_params_t::insert.
39306 (ipa_node_params::~ipa_node_params): Move from removed
39307 ipa_node_params_t::release.
39308 * symbol-summary.h (symbol_summary::m_released): New member.
39309 Do not release a summary twice. Do not allow to call finalizer
39310 for types of a summary that live in GGC memory.
39312 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39314 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
39317 2017-02-02 Martin Sebor <msebor@redhat.com>
39319 PR middle-end/79275
39320 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
39321 (format_string): Tighten up the range of output for non-constant
39322 strings and correct the expected range for wide non-constant strings.
39324 2017-02-02 Martin Sebor <msebor@redhat.com>
39326 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
39328 PR middle-end/32003
39329 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
39331 (-fdump-tree-@var): Add to index and document how to come up
39332 with pass-specific option and dump file names.
39333 (-fdump-passes): Clarify where to look for output.
39335 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39337 PR middle-end/77445
39338 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
39339 statistics of the analyzed path; allow threading for speed when
39340 any of BBs along the path are optimized for speed.
39342 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
39344 PR middle-end/78468
39345 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
39346 settings of the virtual registers.
39349 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
39351 * explow.c (get_dynamic_stack_size): Take known alignment of stack
39352 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
39355 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39357 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
39358 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
39360 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39362 * config/s390/s390.md: Add missing comments with the expanded
39364 * config/s390/vector.md: Likewise.
39365 * config/s390/vx-builtins.md: Likewise.
39367 2017-02-02 Jakub Jelinek <jakub@redhat.com>
39370 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
39371 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
39372 conditions on a single line.
39374 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39376 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39377 __S390_VX__ to __VX__.
39379 2017-02-01 Andrew Pinski <apinski@cavium.com>
39381 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
39382 stmt_info to record_stmt_cost.
39383 (vect_get_known_peeling_cost): Pass stmt_info if known to
39385 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
39386 cpu_vector_cost field into
39387 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
39388 field into vec_int_stmt_cost and vec_fp_stmt_cost.
39389 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
39390 splitting of scalar_stmt_cost and vec_stmt_cost.
39391 (thunderx_vector_cost): Likewise.
39392 (cortexa57_vector_cost): LIkewise.
39393 (exynosm1_vector_cost): Likewise.
39394 (xgene1_vector_cost): Likewise.
39395 (thunderx2t99_vector_cost): Improve after the splitting of the two
39397 (aarch64_builtin_vectorization_cost): Update for the splitting of
39398 scalar_stmt_cost and vec_stmt_cost.
39400 2017-02-01 Torvald Riegel <triegel@redhat.com>
39401 Richard Henderson <rth@redhat.com>
39403 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
39404 conditional on existance of a fast atomic load.
39405 * optabs-query.c (can_atomic_load_p): New function.
39406 * optabs-query.h (can_atomic_load_p): Declare it.
39407 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
39408 no fast atomic load is available for the particular size of access.
39409 (expand_atomic_compare_and_swap): Likewise.
39410 (expand_atomic_load): Likewise.
39411 (expand_atomic_store): Likewise.
39412 (expand_atomic_fetch_op): Likewise.
39413 * testsuite/lib/target-supports.exp
39414 (check_effective_target_sync_int_128): Remove x86 because it provides
39415 no fast atomic load.
39416 (check_effective_target_sync_int_128_runtime): Likewise.
39418 2017-02-01 Richard Biener <rguenther@suse.de>
39420 * graphite.c: Include tree-vectorizer.h for find_loop_location.
39421 (graphite_transform_loops): Provide opt-info for optimized nests.
39422 * tree-parloop.c (parallelize_loops): Provide opt-info for
39423 parallelized loops.
39425 2017-02-01 Richard Biener <rguenther@suse.de>
39427 PR middle-end/79315
39428 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
39429 was not set before.
39431 2017-02-01 Richard Biener <rguenther@suse.de>
39433 PR tree-optimization/71824
39434 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
39435 Verify the loops are valid in the merged SESE region.
39436 (scop_detection::can_represent_loop_1): Check analyzing the
39437 evolution of the number of iterations in the region succeeds.
39439 2017-01-31 Ian Lance Taylor <iant@golang.org>
39441 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
39442 REG_ARGS_SIZE note to 32-bit push insns and call insn.
39444 2017-01-31 David Malcolm <dmalcolm@redhat.com>
39446 PR preprocessor/79210
39447 * input.c (get_substring_ranges_for_loc): Replace line_width
39448 assertion with error-handling.
39450 2017-01-31 Richard Biener <rguenther@suse.de>
39452 PR tree-optimization/77318
39453 * graphite-sese-to-poly.c (extract_affine): Fix assert.
39454 (create_pw_aff_from_tree): Take loop parameter.
39455 (add_condition_to_pbb): Pass loop of the condition to
39456 create_pw_aff_from_tree.
39458 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39460 * config/s390/s390.c (s390_asan_shadow_offset): New function.
39461 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
39463 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
39467 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
39469 (convert_int_to_float128): Likewise.
39470 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
39471 (convert_int_to_float128): Likewise.
39472 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
39473 (UNSPEC_IEEE128_CONVERT): Likewise.
39474 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
39475 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
39476 Use local variables for IBM extended format.
39477 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
39478 (fix_trunc<mode>si2_fprs): Likewise.
39479 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
39480 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
39481 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
39482 to know that we can now have integers of all sizes in vector
39484 (fix<uns>_<mode>di2_hw): Likewise.
39485 (float<uns>_<mode>si2_hw): Likewise.
39486 (fix_<mode>si2_hw): Likewise.
39487 (fixuns_<mode>si2_hw): Likewise.
39488 (float<uns>_<mode>di2_hw): Likewise.
39489 (float_<mode>di2_hw): Likewise.
39490 (float_<mode>si2_hw): Likewise.
39491 (floatuns_<mode>di2_hw): Likewise.
39492 (floatuns_<mode>si2_hw): Likewise.
39493 (xscvqp<su>wz_<mode>): Delete, no longer used.
39494 (xscvqp<su>dz_<mode>): Likewise.
39495 (xscv<su>dqp_<mode>): Likewise.
39496 (ieee128_mfvsrd_64bit): Likewise.
39497 (ieee128_mfvsrd_32bit): Likewise.
39498 (ieee128_mfvsrwz): Likewise.
39499 (ieee128_mtvsrw): Likewise.
39500 (ieee128_mtvsrd_64bit): Likewise.
39501 (ieee128_mtvsrd_32bit): Likewise.
39503 2017-01-31 Martin Liska <mliska@suse.cz>
39506 * ipa-prop.c (ipa_free_all_node_params): Call release method
39507 instead of ~sumbol_summary to not to trigger double times
39510 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
39512 PR tree-optimization/71691
39513 * bitmap.h (class auto_bitmap): New.
39514 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
39515 is_maybe_undefined instead of ssa_undefined_value_p.
39517 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39519 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39520 __S390_ARCH_LEVEL__ to __ARCH__.
39522 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39524 PR tree-optimization/79267
39525 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
39526 if should_remove_lhs_p is true.
39528 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
39531 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
39532 (add_alignment_attribute): New.
39533 (base_type_die): Add alignment attribute.
39534 (subrange_type_die): Likewise.
39535 (modified_type_die): Likewise.
39536 (gen_array_type_die): Likewise.
39537 (gen_descr_array_type_die: Likewise.
39538 (gen_enumeration_type_die): Likewise.
39539 (gen_subprogram_die): Likewise.
39540 (gen_variable_die): Likewise.
39541 (gen_field_die): Likewise.
39542 (gen_ptr_to_mbr_type_die): Likewise.
39543 (gen_struct_or_union_type_die): Likewise.
39544 (gen_subroutine_type_die): Likewise.
39545 (gen_typedef_die): Likewise.
39546 (base_type_cmp): Compare alignment attribute.
39548 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39551 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
39552 (setb_unsigned) New pattern for setb with CCUNS.
39553 * config/rs6000/rs6000.c (expand_block_compare): Use a different
39554 subfc./subfe sequence to avoid overflow problems. Generate a
39555 shorter sequence with cmpld/setb for power9.
39556 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
39557 for generating subfc. instruction.
39558 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
39559 now uses this instruction.
39561 2017-01-30 Ian Lance Taylor <iant@google.com>
39564 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
39565 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
39567 2017-01-30 Martin Sebor <msebor@redhat.com>
39569 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
39570 Move constant to the right of a relational operator.
39571 (get_mpfr_format_length, format_character, format_string): Ditto.
39572 (should_warn_p, maybe_warn): Same.
39574 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
39576 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
39579 * asan.c (get_translation_unit_decl): Remove function.
39580 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
39582 2017-01-30 Martin Liska <mliska@suse.cz>
39584 PR gcov-profile/79259
39585 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
39586 -fprofile-generate.
39588 2017-01-30 Martin Liska <mliska@suse.cz>
39591 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
39592 Initialize variables with NULL value.
39594 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
39597 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
39599 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
39601 2017-01-30 Richard Biener <rguenther@suse.de>
39603 PR tree-optimization/79276
39604 * tree-vrp.c (process_assert_insertions): Properly adjust common
39605 when removing a duplicate.
39607 2017-01-30 Richard Biener <rguenther@suse.de>
39609 PR tree-optimization/79256
39610 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
39611 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
39613 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
39615 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
39618 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
39619 ("*r<noxa>sbg_<mode>_sll_bitmask")
39620 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
39621 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
39622 Use contiguous_bitmask_nowrap_operand.
39624 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39627 * config/rs6000/altivec.h (vec_xl): Revise #define.
39628 (vec_xst): Likewise.
39630 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
39632 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
39634 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
39636 PR rtl-optimization/79194
39637 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
39638 traps before call to bypass_conditional_jumps.
39640 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39642 PR tree-optimization/71374
39643 * lra-constraints.c (check_conflict_input_operands): New.
39644 (match_reload): Use it.
39646 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39649 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
39650 account to calculate conflict_set.
39652 2017-01-27 Bin Cheng <bin.cheng@arm.com>
39654 PR rtl-optimization/78559
39655 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
39656 other_insn in combine.
39658 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
39660 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
39661 uint16_type_node for BT_UINT16.
39663 2017-01-27 David Malcolm <dmalcolm@redhat.com>
39665 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
39666 "RTL Tests" to menu.
39667 (GIMPLE Tests): New node.
39668 (RTL Tests): New node.
39670 2017-01-27 Richard Biener <rguenther@suse.de>
39672 PR tree-optimization/79245
39673 * tree-loop-distribution.c (distribute_loop): Apply cost
39674 modeling also to detected patterns.
39676 2017-01-27 Richard Biener <rguenther@suse.de>
39678 PR tree-optimization/71433
39679 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
39680 (compare_assert_loc): New function.
39681 (process_assert_insertions): Sort and optimize assert locations
39682 to remove duplicates and push down identical assertions on
39683 edges to their destination block.
39685 2017-01-27 Richard Biener <rguenther@suse.de>
39687 PR tree-optimization/79244
39688 * tree-vrp.c (remove_range_assertions): Forcefully propagate
39689 out SSA names even if abnormal.
39691 2017-01-27 Jakub Jelinek <jakub@redhat.com>
39693 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
39694 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
39695 instead of MPFR_RNDN.
39697 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
39700 * arm.c (arm_option_override): Don't call build_target_option_node
39701 until after doing all option overrides.
39702 (arm_valid_target_attribute_tree): Likewise.
39704 2017-01-27 Martin Liska <mliska@suse.cz>
39706 * doc/invoke.texi (-fprofile-arcs): Document profiling support
39707 for {cd}tors and C++ {cd}tors.
39709 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
39711 * config/s390/s390.md ("*setmem_long_and")
39712 ("*setmem_long_and_31z"): Use zero_extend instead of and.
39714 2017-01-26 Martin Sebor <msebor@redhat.com>
39716 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
39719 2017-01-26 Martin Sebor <msebor@redhat.com>
39721 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
39722 HAVE_DFmode before using XFmode or DFmode.
39723 (parse_directive): Avoid using the z length modifier to avoid
39724 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
39726 PR middle-end/78703
39727 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
39728 to accept adjustment as an array.
39729 (get_int_range): New function.
39730 (struct directive): Make width and prec arrays.
39731 (directive::set_width, directive::set_precision): Call get_int_range.
39732 (format_integer, format_floating): Handle width and precision ranges.
39733 (format_string, parse_directive): Same.
39735 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39738 * dwarf2out.c (generate_skeleton_bottom_up): For children with
39739 comdat_type_p set, just clone them, but keep the children in the
39743 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
39744 which have direct callers with -fvar-tracking-assignments enabled
39746 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
39747 inside of type units.
39749 2017-01-26 Martin Sebor <msebor@redhat.com>
39751 PR middle-end/78703
39752 * gimple-ssa-sprintf.c (struct result_range): Add likely and
39754 (struct format_result): Replace number_chars, number_chars_min,
39755 and number_chars_max with a single member of struct result_range.
39757 (format_result::operator+=): Adjust.
39758 (struct fmtresult): Remove bounded. Handle likely and unlikely
39760 (fmtresult::adjust_for_width_or_precision): New function.
39761 (fmtresult:type_max_digits): New function.
39762 (bytes_remaining): Handle likely and unlikely counters.
39763 (min_bytes_remaining): Remove.
39764 (format_percent): Simplify.
39765 (format_integer, format_floating): Set likely and unlikely counters.
39766 (get_string_length, format_character, format_string): Same.
39767 (format_plain, should_warn_p): New function.
39768 (maybe_warn): Call should_warn_p. Update diagnostic messages
39769 and handle those for all directives, including plain strings.
39770 (format_directive): Handle likely and unlikely counters.
39771 Remove unnecessary quoting from diagnostics. Add an informational
39773 (add_bytes): Remove.
39774 (pass_sprintf_length::compute_format_length): Simplify.
39775 (try_substitute_return_value): Handle likely and unlikely counters.
39777 2017-01-26 Carl Love <cel@us.ibm.com>
39779 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
39780 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
39782 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
39785 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
39786 endianess for subregs into account.
39787 * lra-constraints.c (lra_constraints): Do risky transformations
39788 always on the first iteration.
39789 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39790 last_call_used_reg_set.
39791 (process_bb_lives): Define and use last_call_used_reg_set.
39792 * lra.c (lra): Always continue after lra_constraints on the first
39795 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39797 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39799 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39801 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39803 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39804 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39805 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39806 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39807 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39808 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39809 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39810 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39811 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39813 2017-01-26 Marek Polacek <polacek@redhat.com>
39816 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39817 for the third operand.
39819 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39821 PR middle-end/79236
39822 * omp-low.c (struct omp_context): Add simt_stmt field.
39823 (scan_omp_for): Return omp_context *.
39824 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39825 context to the _simt_ SIMD stmt.
39826 (lower_omp_for): For combined SIMD with sibling _simt_
39827 SIMD, make sure to use the same decls in _looptemp_
39828 clauses as in the sibling.
39830 2017-01-26 David Sherwood <david.sherwood@arm.com>
39832 PR middle-end/79212
39833 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39836 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39839 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39840 emit fld b; fld a; if possible.
39842 * brig-builtins.def: Update copyright years.
39843 * config/arm/arm_acle_builtins.def: Update copyright years.
39845 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39848 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39849 constraint instead of o for the stxsd instruction.
39851 2017-01-25 Carl Love <cel@us.ibm.com>
39853 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39854 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39856 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39858 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39860 2017-01-25 Richard Biener <rguenther@suse.de>
39862 PR tree-optimization/69264
39863 * target.def (vector_alignment_reachable): Improve documentation.
39864 * doc/tm.texi: Regenerate.
39865 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39867 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39868 earlier changes with respect to TYPE_USER_ALIGN.
39869 (vector_alignment_reachable_p): Likewise. Improve dumping.
39871 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39874 * config/arm/arm.md (xordi3): Force constant operand into a register
39877 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39879 * doc/invoke.texi (-fstore-merging): Correct default optimization
39880 levels at which it is enabled.
39881 (-O): Move -fstore-merging from list to...
39884 2017-01-25 Richard Biener <rguenther@suse.de>
39887 * omp-expand.c: Include debug.h.
39888 (expand_omp_taskreg): Make sure to generate early debug before
39889 outlining anything from a function.
39890 (expand_omp_target): Likewise.
39891 (grid_expand_target_grid_body): Likewise.
39893 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39896 * asan.c (get_translation_unit_decl): New function.
39897 (asan_add_global): Extract modules file name from globals
39898 TRANSLATION_UNIT_DECL name.
39900 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39903 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39904 for long calls with APCS frame and VFP.
39906 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39908 * cfg.c (original_copy_tables_initialized_p): New function.
39909 * cfg.h (original_copy_tables_initialized_p): New decl.
39910 * cfgrtl.c (relink_block_chain): Guard the call to
39911 free_original_copy_tables with a call to
39912 original_copy_tables_initialized_p.
39913 * cgraph.h (symtab_node::native_rtl_p): New decl.
39914 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39915 (symtab_node::needed_p): Don't assert for early assembly output
39916 for __RTL functions.
39917 (cgraph_node::finalize_function): Set "force_output" for __RTL
39919 (cgraph_node::analyze): Bail out early for __RTL functions.
39920 (analyze_functions): Update assertion to support __RTL functions.
39921 (cgraph_node::expand): Bail out early for __RTL functions.
39922 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39924 * function.h (struct function): Update comment for field
39926 * gimple-expr.c: Include "tree-pass.h".
39927 (gimple_has_body_p): Return false for __RTL functions.
39928 * Makefile.in (OBJS): Add run-rtl-passes.o.
39929 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39931 (gcc::pass_manager::get_clean_slate): New accessor.
39932 * passes.c: Include "insn-addr.h".
39933 (should_skip_pass_p): Add logging. Update logic for running
39934 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39935 property-provider override so it is only done for gimple passes.
39937 (skip_pass): New function.
39938 (execute_one_pass): Call skip_pass when skipping passes.
39939 * read-md.c (md_reader::read_char): Support filtering
39940 the input to a subset of line numbers.
39941 (md_reader::md_reader): Initialize fields
39942 m_first_line and m_last_line.
39943 (md_reader::read_file_fragment): New function.
39944 * read-md.h (md_reader::read_file_fragment): New decl.
39945 (md_reader::m_first_line): New field.
39946 (md_reader::m_last_line): New field.
39947 * read-rtl-function.c (function_reader::create_function): Only
39948 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39949 curr_properties. Set DECL_INITIAL to a dummy block.
39950 (read_rtl_function_body_from_file_range): New function.
39951 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39953 * run-rtl-passes.c: New file.
39954 * run-rtl-passes.h: New file.
39956 2017-01-24 Jeff Law <law@redhat.com>
39958 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39961 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39963 PR tree-optimization/79159
39964 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39965 (record_nonwrapping_iv): Improve boundary using above function if no
39966 value range information.
39968 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39969 Martin Jambor <mjambor@suse.cz>
39971 * brig-builtins.def: New file.
39972 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39973 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39974 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39975 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39976 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39977 * builtin-types.def (BT_INT8): New.
39978 (BT_INT16): Likewise.
39979 (BT_UINT8): Likewise.
39980 (BT_UINT16): Likewise.
39981 (BT_FN_ULONG): Likewise.
39982 (BT_FN_UINT_INT): Likewise.
39983 (BT_FN_UINT_ULONG): Likewise.
39984 (BT_FN_UINT_LONG): Likewise.
39985 (BT_FN_UINT_PTR): Likewise.
39986 (BT_FN_ULONG_PTR): Likewise.
39987 (BT_FN_INT8_FLOAT): Likewise.
39988 (BT_FN_INT16_FLOAT): Likewise.
39989 (BT_FN_UINT32_FLOAT): Likewise.
39990 (BT_FN_UINT16_FLOAT): Likewise.
39991 (BT_FN_UINT8_FLOAT): Likewise.
39992 (BT_FN_UINT64_FLOAT): Likewise.
39993 (BT_FN_UINT16_UINT32): Likewise.
39994 (BT_FN_UINT32_UINT16): Likewise.
39995 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39996 (BT_FN_INT_PTR_INT): Likewise.
39997 (BT_FN_UINT_PTR_UINT): Likewise.
39998 (BT_FN_LONG_PTR_LONG): Likewise.
39999 (BT_FN_ULONG_PTR_ULONG): Likewise.
40000 (BT_FN_VOID_UINT64_UINT64): Likewise.
40001 (BT_FN_UINT8_UINT8_UINT8): Likewise.
40002 (BT_FN_INT8_INT8_INT8): Likewise.
40003 (BT_FN_INT16_INT16_INT16): Likewise.
40004 (BT_FN_INT_INT_INT): Likewise.
40005 (BT_FN_UINT_FLOAT_UINT): Likewise.
40006 (BT_FN_FLOAT_UINT_UINT): Likewise.
40007 (BT_FN_ULONG_UINT_UINT): Likewise.
40008 (BT_FN_ULONG_UINT_PTR): Likewise.
40009 (BT_FN_ULONG_ULONG_ULONG): Likewise.
40010 (BT_FN_UINT_UINT_UINT): Likewise.
40011 (BT_FN_VOID_UINT_PTR): Likewise.
40012 (BT_FN_UINT_UINT_PTR: Likewise.
40013 (BT_FN_UINT32_UINT64_PTR): Likewise.
40014 (BT_FN_INT_INT_UINT_UINT): Likewise.
40015 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
40016 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
40017 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
40018 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
40019 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
40020 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
40021 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
40022 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
40023 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
40024 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
40025 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
40026 * doc/frontends.texi: List BRIG FE.
40027 * doc/install.texi (Testing): Add BRIG tesring requirements.
40028 * doc/invoke.texi (Overall Options): Mention BRIG.
40029 * doc/standards.texi (Standards): Doucment BRIG HSA version.
40031 2017-01-24 Richard Biener <rguenther@suse.de>
40033 PR translation/79208
40034 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
40036 2017-01-24 Martin Jambor <mjambor@suse.cz>
40039 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
40040 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
40041 and known_contexts.
40043 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
40045 PR middle-end/79123
40046 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
40047 casts from signed to unsigned really don't have a range.
40049 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
40051 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
40052 GMP_RNDx for compatiblity.
40054 2017-01-24 Martin Liska <mliska@suse.cz>
40057 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
40058 that would prevent us to call alloca with -1 as argument.
40060 2017-01-24 Jakub Jelinek <jakub@redhat.com>
40062 * dwarf2out.c (output_compilation_unit_header, output_file_names):
40063 Avoid -Wformat-security warning.
40065 2017-01-23 Andrew Pinski <apinski@cavium.com>
40067 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
40070 2017-01-23 Martin Sebor <msebor@redhat.com>
40072 PR middle-end/78703
40073 * gimple-ssa-sprintf.c (warn_level): New global.
40074 (format_integer): Use it here and throughout the rest of the file.
40075 Use the same switch to compute sign as base.
40076 (maybe_warn): New function.
40077 (format_directive): Factor out warnings into maybe_warn.
40078 Add debugging output. Use warn_level.
40079 (add_bytes): Use warn_level.
40080 (pass_sprintf_length::compute_format_length): Add debugging output.
40081 (try_substitute_return_value): Same.
40082 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
40084 PR middle-end/78703
40085 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
40086 (struct fmtresult, format_integer, format_floating): Adjust.
40087 (fmtresult::fmtresult): Set max correctly in two argument ctor.
40088 (get_string_length, format_string,format_directive): Same.
40089 (pass_sprintf_length::compute_format_length): Same.
40090 (try_substitute_return_value): Simplify slightly.
40092 PR middle-end/78703
40093 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
40094 (fmtresult::operator+=): Outlined.
40095 (struct fmtresult): Add ctors.
40096 (struct conversion_spec): Rename...
40097 (struct directive): ...to this. Add and remove data members.
40098 (directive::set_width, directive::set_precision): New functions.
40099 (format_percent): Use fmtresult ctor.
40100 (get_width_and_precision): Remove.
40101 (format_integer): Make naming changes. Avoid computing width and
40103 (format_floating): Same. Adjust indentation.
40104 (format_character, format_none): New functions.
40105 (format_string): Moved character handling to format_character.
40106 (format_directive): Remove arguments, change return type.
40107 (parse_directive): New function.
40108 (pass_sprintf_length::compute_format_length): Move directive
40109 parsing to parse_directive.
40111 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40113 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
40114 (assign_assembler_name_if_needed): ... this.
40115 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
40116 (assign_assembler_name_if_needed): ... this.
40117 (free_lang_data_in_cgraph): Adjust callers.
40118 * cgraphunit.c (cgraph_node::analyze): Likewise.
40119 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
40122 2017-01-23 Richard Biener <rguenther@suse.de>
40124 PR tree-optimization/79088
40125 PR tree-optimization/79188
40126 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
40127 resetting loop bounds after last path deletion. Reset loop
40128 bounds of the target loop, make code match the comments.
40129 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
40130 Make sure loops need no fixups.
40132 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
40134 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
40135 exponent support with double type for first argument.
40136 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
40137 type returned by __builtin_vec_extract_sig,
40138 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
40139 functions from "vector int" to "vector unsigned int" or from
40140 "vector long long int" to "vector unsigned long long int".
40141 Changed type returned by __builtin_vec_extract_exp,
40142 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
40143 functions from "vector int" to "vector unsigned int" or from
40144 "vector long long int" to "vector unsigned long long int".
40145 Changed return type of __builtin_vec_test_data_class,
40146 __builtin_vec_test_data_class_sp, and
40147 __builtin_vec_test_data_class_dp from "vector int" to
40148 "vector bool int" or from "vector long long int" to "vector bool
40149 long long int" and changed second argument type from "unsigned
40150 int" to "int". Added new overloaded function forms "vector float
40151 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
40152 "vector float __builtin_vec_insert_exp_sp (vector float, vector
40153 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
40154 double, vector unsigned long long int)" and "vector double
40155 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
40156 long int)". Changed return type of
40157 __builtin_scalar_test_data_class and
40158 __builtin_scalar_test_data_class_sp and
40159 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
40160 int" and changed second argument from "unsigned int" to "int".
40161 Changed type returned by __builtin_scalar_test_neg,
40162 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
40163 from "int" to "bool int". Added new overloaded function form
40164 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
40165 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
40166 exponent double-precision with floating point first argument.
40167 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
40168 documentation of scalar_test_data_class, scalar_test_neg,
40169 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
40170 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
40171 vec_test_data_class built-in functions to reflect refinements in
40172 their type signatures.
40174 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
40176 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
40178 (aarch64_elf_asm_destructor): Likewise.
40180 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
40182 PR rtl-optimization/78634
40183 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
40184 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
40185 * ifcvt.c (noce_try_cmove): Add missing cost check.
40187 PR rtl-optimization/71724
40188 * combine.c (if_then_else_cond): Look for situations where it is
40189 beneficial to undo the work of one of the recursive calls.
40191 2017-01-23 Bin Cheng <bin.cheng@arm.com>
40193 PR tree-optimization/70754
40194 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
40195 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
40196 combined stmt before it if not NULL.
40197 (combine_chains): Process refs reversely and compute dominance point
40200 2017-01-23 Martin Liska <mliska@suse.cz>
40202 PR tree-optimization/79196
40203 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
40204 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
40206 (strlen_optimize_stmt): Call the renamed function.
40208 2017-01-23 Michael Matz <matz@suse.de>
40210 PR tree-optimization/78384
40211 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
40213 2017-01-23 Richard Biener <rguenther@suse.de>
40215 PR tree-optimization/79186
40216 * tree-vrp.c (register_new_assert_for): Make sure we've seen
40217 both incoming edges before moving an assert.
40219 2017-01-23 Martin Jambor <mjambor@suse.cz>
40221 * ipa-prop.c (load_from_param_1): Removed.
40222 (load_from_unmodified_param): Bits from load_from_param_1 put back
40224 (load_from_param): Removed.
40225 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
40226 with stmt. Reverted back to use of load_from_unmodified_param.
40228 2017-01-23 Martin Jambor <mjambor@suse.cz>
40231 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
40232 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
40233 field a pointer to garbage collected vector, mark lattices and
40234 ipcp_orig_node with GTY((skip)).
40235 (ipa_get_param_count): Adjust to descriptors being a pointer.
40236 (ipa_get_param): Likewise.
40237 (ipa_get_type): Likewise.
40238 (ipa_get_param_move_cost): Likewise.
40239 (ipa_set_param_used): Likewise.
40240 (ipa_get_controlled_uses): Likewise.
40241 (ipa_set_controlled_uses): Likewise.
40242 (ipa_is_param_used): Likewise.
40243 (ipa_node_params_t): Move into garbage collector. New methods insert
40245 (ipa_node_params_sum): Annotate wth GTY(()).
40246 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
40248 (ipa_load_from_parm_agg): Adjust declaration.
40249 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
40250 * ipa-profile.c (ipa_profile): Likewise.
40251 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
40252 (ipa_populate_param_decls): Make descriptors parameter garbage
40254 (ipa_dump_param): Adjust to descriptors being a pointer.
40255 (ipa_alloc_node_params): Likewise.
40256 (ipa_initialize_node_params): Likewise.
40257 (load_from_param_1): Make descriptors parameter garbage collected.
40258 (load_from_unmodified_param): Likewise.
40259 (load_from_param): Likewise.
40260 (ipa_load_from_parm_agg): Likewise.
40261 (ipa_node_params::~ipa_node_params): Removed.
40262 (ipa_free_all_node_params): Remove call to delete operator.
40263 (ipa_node_params_t::insert): New.
40264 (ipa_node_params_t::remove): Likewise.
40265 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
40266 copy known_csts and known_contexts vectors.
40267 (ipa_read_node_info): Adjust to descriptors being a pointer.
40268 (ipcp_modif_dom_walker): Make m_descriptors field garbage
40270 (ipcp_transform_function): Make descriptors variable garbage
40273 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
40275 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
40276 * config/i386/avx512dqintrin.h: Ditto.
40277 * config/i386/avx512fintrin.h: Ditto.
40278 * config/i386/i386.c: Handle new builtins.
40279 * config/i386/i386-builtin.def: Add new builtins.
40280 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
40281 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
40283 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40284 Martin Liska <mliska@suse.cz>
40286 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
40287 * asan.c (asan_expand_poison_ifn): Support stores and use
40288 appropriate ASAN report function.
40289 * internal-fn.c (expand_ASAN_POISON_USE): New function.
40290 * internal-fn.def (ASAN_POISON_USE): Declare.
40291 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
40292 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
40293 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
40294 ASAN_POISON calls w/o LHS.
40295 * tree-ssa.c (execute_update_addresses_taken): Create clobber
40296 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
40297 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
40298 * gimplify.c (asan_poison_variables): Add attribute
40299 use_after_scope_memory to variables that really needs to live
40301 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
40302 having the attribute.
40304 2017-01-23 Martin Liska <mliska@suse.cz>
40306 * asan.c (create_asan_shadow_var): New function.
40307 (asan_expand_poison_ifn): Likewise.
40308 * asan.h (asan_expand_poison_ifn): New declaration.
40309 * internal-fn.c (expand_ASAN_POISON): Likewise.
40310 * internal-fn.def (ASAN_POISON): New builtin.
40311 * sanopt.c (pass_sanopt::execute): Expand
40312 asan_expand_poison_ifn.
40313 * tree-inline.c (copy_decl_for_dup_finish): Make function
40315 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
40316 * tree-ssa.c (is_asan_mark_p): New function.
40317 (execute_update_addresses_taken): Rewrite local variables
40318 (identified just by use-after-scope as addressable) into SSA.
40320 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40322 * doc/install.texi (Specific): opensource.apple.com uses https
40323 now. Remove trailing slash.
40325 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40327 * README.Portability: Remove note on an Irix compatibility issue.
40329 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
40331 * gcov.c (INCLUDE_ALGORITHM): Define.
40332 (INCLUDE_VECTOR): Define.
40333 No longer include <vector> and <algorithm> directly.
40335 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
40337 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
40339 * doc/invoke.texi (Code Gen Options): Ditto.
40341 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
40344 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
40346 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
40348 rtl-optimization/79125
40349 * cprop.c (local_cprop_pass): Handle cases where we make an
40350 unconditional trap.
40352 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
40356 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
40357 read from, for big endian.
40359 2017-01-20 Jiong Wang <jiong.wang@arm.com>
40361 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
40362 register pauth builtins for LP64 only.
40364 2017-01-20 Marek Polacek <polacek@redhat.com>
40367 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
40370 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40372 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
40374 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
40375 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
40376 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
40378 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40381 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
40382 in big-endian mode when they are not a single duplicated value.
40384 2017-01-20 Richard Biener <rguenther@suse.de>
40386 * BASE-VER: Bump to 7.0.1.
40388 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40390 * omp-low.c (omplow_simd_context): New struct. Use it...
40391 (lower_rec_simd_input_clauses): ...here and...
40392 (lower_rec_input_clauses): ...here to hold common data. Adjust all
40393 references to idx, lane, max_vf, is_simt.
40395 2017-01-20 Graham Markall <graham.markall@embecosm.com>
40397 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
40400 2017-01-20 Martin Jambor <mjambor@suse.cz>
40402 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
40403 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
40405 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
40406 (GTFILES): Rename hsa.c to hsa-common.c.
40407 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
40408 * hsa-dump.c: Likewise.
40409 * hsa-gen.c: Likewise.
40410 * hsa-regalloc.c: Likewise.
40411 * ipa-hsa.c: Likewise.
40412 * omp-expand.c: Likewise.
40413 * omp-low.c: Likewise.
40414 * toplev.c: Likewise.
40416 2017-01-20 Marek Polacek <polacek@redhat.com>
40419 * doc/invoke.texi: Document -Wduplicated-branches.
40420 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
40421 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
40422 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
40423 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
40424 return 0 only when not OEP_LEXICOGRAPHIC.
40425 (fold_build_cleanup_point_expr): Use the expression
40426 location when building CLEANUP_POINT_EXPR.
40427 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
40428 * tree.c (add_expr): Handle error_mark_node.
40430 2017-01-20 Martin Liska <mliska@suse.cz>
40433 * tree-profile.c (init_ic_make_global_vars): Do not call
40435 (gimple_init_gcov_profiler): Likewise.
40437 2017-01-20 Martin Liska <mliska@suse.cz>
40440 * cgraph.h (maybe_create_reference): Remove argument and
40442 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
40444 * ipa-cp.c (create_specialized_node): Likewise.
40445 * symtab.c (symtab_node::maybe_create_reference): Handle
40446 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
40448 2017-01-20 Martin Liska <mliska@suse.cz>
40450 * read-rtl-function.c (function_reader::create_function): Use
40451 build_decl instread of build_decl_stat.
40453 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
40455 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
40456 * config/i386/avx512dqintrin.h: Ditto.
40457 * config/i386/avx512fintrin.h: Ditto.
40458 * config/i386/i386-builtin-types.def: Add new types.
40459 * config/i386/i386.c: Handle new types.
40460 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
40461 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
40462 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
40463 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
40464 (__builtin_ia32_kshiftridi): New.
40465 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
40467 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
40471 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
40472 define to rs6000_init_stack_protect_guard.
40473 (rs6000_init_stack_protect_guard): New function.
40475 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40476 Yunqiang Su <yunqiang.su@imgtec.com>
40478 * config.gcc (supported_defaults): Add madd4.
40479 (with_madd4): Add validation.
40480 (all_defaults): Add madd4.
40481 * config/mips/mips.opt (mmadd4): New option.
40482 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40484 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40486 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
40487 (ISA_HAS_FUSED_MADD4): Likewise.
40488 * doc/invoke.texi (-mmadd4): Document the new option.
40489 * doc/install.texi (--with-madd4): Document the new option.
40491 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40493 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
40494 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
40495 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
40496 (aarch64_init_pauth_hint_builtins): New.
40497 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
40498 (aarch64_expand_builtin): Expand new builtins.
40500 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40502 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
40503 * combine-stack-adj.c (no_unhandled_cfa): Handle
40504 REG_CFA_TOGGLE_RA_MANGLE.
40505 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
40506 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
40507 info for return address signing.
40508 (aarch64_expand_epilogue): Likewise.
40510 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40512 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
40513 * config/aarch64/aarch64-protos.h
40514 (aarch64_return_address_signing_enabled): New declaration.
40515 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
40517 (aarch64_expand_prologue): Sign return address before it's pushed onto
40519 (aarch64_expand_epilogue): Authenticate return address fetched from
40521 (aarch64_override_options): Sanity check for ILP32 and ISA level.
40522 (aarch64_attributes): New function attributes for "sign-return-address".
40523 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
40524 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
40525 ("*do_return"): Generate combined instructions according to key index.
40526 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
40527 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
40529 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
40530 * config/aarch64/aarch64.opt (msign-return-address=): New.
40531 * doc/extend.texi (AArch64 Function Attributes): Documents
40532 "sign-return-address=".
40533 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
40535 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40537 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
40538 overall option summary.
40540 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40542 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
40543 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
40544 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
40545 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
40547 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
40549 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
40550 -mpower9-minmax by default for -mcpu=power9.
40551 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
40552 128-bit floating point.
40554 2017-01-20 Alan Modra <amodra@gmail.com>
40556 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
40557 optimizing for size.
40559 2017-01-20 Alan Modra <amodra@gmail.com>
40562 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
40563 for strcmp and strncmp from corresponding builtin decl.
40565 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40567 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
40568 instead of i386/rtems-64.h.
40569 * config/i386/rtems-64.h: Remove.
40571 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40575 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
40577 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
40579 2017-01-19 Tamar Christina <tamar.christina@arm.com>
40581 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
40582 Change int to HOST_WIDE_INT.
40583 * config/aarch64/aarch64-protos.h
40584 (aarch64_simd_gen_const_vector_dup): Likewise.
40585 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
40587 2017-01-19 David Malcolm <dmalcolm@redhat.com>
40589 * langhooks-def.h (lhd_type_for_size): New decl.
40590 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
40591 * langhooks.c (lhd_type_for_size): New function, taken from
40594 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
40596 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
40597 define_bypass for CR latency.
40598 (power9-cracked-alu): Update bypass latency and remove power9-branch.
40599 (power9-alu2): Add define_bypass for CR latency.
40601 (power9-mul): Update insn latency.
40602 (power9-mul-compare): Update insn latency, bypass latency and remove
40605 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40607 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
40609 * config/aarch64/aarch64.md
40610 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
40611 aarch64_nopcrelative_literal_loads.
40612 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
40614 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
40616 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
40617 TARGET_LOONGSON_3A.
40618 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
40620 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
40623 * config.gcc (supported_defaults): Add lxc1-sxc1.
40624 (with_lxc1_sxc1): Add validation.
40625 (all_defaults): Add lxc1-sxc1.
40626 * config/mips/mips.opt (mlxc1-sxc1): New option.
40627 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40629 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40630 __mips_no_lxc1_sxc1.
40631 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
40632 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
40633 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
40635 2017-01-19 Richard Biener <rguenther@suse.de>
40637 PR tree-optimization/72488
40638 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
40639 sure to restore SSA info.
40640 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
40642 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
40644 PR rtl-optimization/79121
40645 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
40646 of the inner type when shifting an extended value.
40648 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40651 * symtab.c (symtab_node::equal_address_to): Fix comparing of
40652 interposable aliases.
40654 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
40657 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
40658 Use the evmergelohi instruction.
40659 (mov_si<mode>_e500_subreg4_2_le): Likewise.
40660 (mov_sitf_e500_subreg8_2_be): Likewise.
40661 (mov_sitf_e500_subreg12_2_le): Likewise.
40662 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
40663 (mov_si<mode>_e500_subreg4_2_be): Likewise.
40664 (mov_sitf_e500_subreg8_2_le): Likewise.
40665 (mov_sitf_e500_subreg12_2_be): Likewise.
40667 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40669 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
40670 attribute from vecsimple to vecperm.
40671 (altivec_vbpermq2): Likewise.
40673 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40676 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
40678 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
40679 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
40680 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
40681 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
40682 case where N arg is SIZE_MAX.
40683 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
40684 (cmpstrsi): Add pattern.
40686 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
40688 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
40689 __builtin_vec_revb builtins.
40690 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
40691 built-in functions to support generation of the ISA 3.0 XXBR<x>
40692 vector byte reverse instructions.
40693 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
40694 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
40695 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
40696 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
40697 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
40698 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
40699 (P9V_BUILTIN_VEC_REVB): Likewise.
40700 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
40701 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
40702 (p9_xxbrq_v16qi): Likewise.
40703 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
40704 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
40705 (p9_xxbrh_v8hi): Likewise.
40706 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
40707 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
40708 vec_revb built-in functions.
40710 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
40712 PR rtl-optimization/78952
40713 * config/i386/i386.md (any_extract): New code iterator.
40714 (*insvqi_2): Use any_extract for source operand.
40715 (*insvqi_3): Use any_shiftrt for source operand.
40717 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
40719 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
40721 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
40723 2017-01-18 Matthias Klose <doko@ubuntu.com>
40725 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
40727 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40729 * config/rs6000/altivec.h (vec_bperm): Change #define.
40730 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
40731 (altivec_vbpermq2): New define_insn.
40732 (altivec_vbpermd): Likewise.
40733 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
40734 function interface.
40735 (VBPERMD): Likewise.
40736 (VBPERM): New polymorphic function interface.
40737 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
40738 Add entries for P9V_BUILTIN_VEC_VBPERM.
40739 * doc/extend.texi: Add interfaces for vec_bperm.
40741 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40743 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
40744 first letter of error messages.
40745 (s390_resolve_overloaded_builtin): Likewise.
40746 * config/s390/s390.c (s390_expand_builtin): Likewise.
40747 (s390_invalid_arg_for_unprototyped_fn): Likewise.
40748 (s390_valid_target_attribute_inner_p): Likewise.
40749 * config/s390/s390.md ("tabort"): Likewise.
40751 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
40753 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
40754 (ISA_AVOID_DIV_HILO): New macro.
40755 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
40756 (ISA_HAS_DDIV): Likewise.
40758 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40760 * doc/invoke.texi (fabi-version): Correct number of occurrences.
40762 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40764 * doc/invoke.texi (fabi-version): Spelling fix.
40766 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40769 * doc/invoke.texi (fabi-version): Mention mangling fix for
40772 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40775 * doc/invoke.texi (fabi-version): Document discriminator mangling.
40777 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
40780 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
40781 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
40783 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
40785 (stack_protect_test): Ditto.
40786 * config/rs6000/rs6000.opt (mstack-protector-guard=,
40787 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
40789 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40790 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40791 -mstack-protector-guard-offset=.
40792 (RS/6000 and PowerPC Options): Ditto.
40794 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40796 * config/i386/i386.h (MASK_CLASS_P): New define.
40797 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40798 there are no registers from different register sets also when
40799 mask registers are used. Update function comment.
40800 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40801 to (*k/*r) and (*k/*km) alternatives.
40803 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40805 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40806 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40807 (EH_RETURN_HANDLER_RTX): New define.
40808 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40809 Force frame pointer in EH return functions.
40810 (aarch64_expand_epilogue): Add barrier for eh_return.
40811 (aarch64_final_eh_return_addr): Remove.
40812 (aarch64_eh_return_handler_rtx): New function.
40813 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40815 (aarch64_eh_return_handler_rtx): New prototype.
40817 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40819 * config/rs6000/altivec.h (vec_rlmi): New #define.
40820 (vec_vrlnm): Likewise.
40821 (vec_rlnm): Likewise.
40822 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40823 (UNSPEC_VRLNM): Likewise.
40824 (VIlong): New mode iterator.
40825 (altivec_vrl<VI_char>mi): New define_insn.
40826 (altivec_vrl<VI_char>nm): Likewise.
40827 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40829 (VRLDNM): Likewise.
40830 (RLNM): New polymorphic function entry.
40831 (VRLWMI): New monomorphic function entry.
40832 (VRLDMI): Likewise.
40833 (RLMI): New polymorphic function entry.
40834 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40835 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40836 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40839 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40842 * dwarf2out.c (field_byte_offset): Restore the
40843 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40844 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40845 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40848 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40851 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40853 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40856 * dwarf2out.c (add_data_member_location_attribute): For constant
40857 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40858 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40859 DW_AT_byte_size attributes.
40861 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40863 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40864 after forcing to constant memory when the code model is medium.
40866 2017-01-17 Julia Koval <julia.koval@intel.com>
40869 * config/i386/avx512fintrin.h
40870 (_mm512_i32gather_ps): Change __addr type to void const*.
40871 (_mm512_mask_i32gather_ps): Ditto.
40872 (_mm512_i32gather_pd): Ditto.
40873 (_mm512_mask_i32gather_pd): Ditto.
40874 (_mm512_i64gather_ps): Ditto.
40875 (_mm512_mask_i64gather_ps): Ditto.
40876 (_mm512_i64gather_pd): Ditto.
40877 (_mm512_mask_i64gather_pd): Ditto.
40878 (_mm512_i32gather_epi32): Ditto.
40879 (_mm512_mask_i32gather_epi32): Ditto.
40880 (_mm512_i32gather_epi64): Ditto.
40881 (_mm512_mask_i32gather_epi64): Ditto.
40882 (_mm512_i64gather_epi32): Ditto.
40883 (_mm512_mask_i64gather_epi32): Ditto.
40884 (_mm512_i64gather_epi64): Ditto.
40885 (_mm512_mask_i64gather_epi64): Ditto.
40886 (_mm512_i32scatter_ps): Change __addr type to void*.
40887 (_mm512_mask_i32scatter_ps): Ditto.
40888 (_mm512_i32scatter_pd): Ditto.
40889 (_mm512_mask_i32scatter_pd): Ditto.
40890 (_mm512_i64scatter_ps): Ditto.
40891 (_mm512_mask_i64scatter_ps): Ditto.
40892 (_mm512_i64scatter_pd): Ditto.
40893 (_mm512_mask_i64scatter_pd): Ditto.
40894 (_mm512_i32scatter_epi32): Ditto.
40895 (_mm512_mask_i32scatter_epi32): Ditto.
40896 (_mm512_i32scatter_epi64): Ditto.
40897 (_mm512_mask_i32scatter_epi64): Ditto.
40898 (_mm512_i64scatter_epi32): Ditto.
40899 (_mm512_mask_i64scatter_epi32): Ditto.
40900 (_mm512_i64scatter_epi64): Ditto.
40901 (_mm512_mask_i64scatter_epi64): Ditto.
40902 * config/i386/avx512pfintrin.h
40903 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40904 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40905 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40906 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40907 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40908 (_mm512_prefetch_i32scatter_ps): Ditto.
40909 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40910 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40911 (_mm512_prefetch_i64scatter_pd): Ditto.
40912 (_mm512_prefetch_i64scatter_ps): Ditto.
40913 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40914 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40915 * config/i386/avx512vlintrin.h
40916 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40917 (_mm_mmask_i32gather_ps): Ditto.
40918 (_mm256_mmask_i32gather_pd): Ditto.
40919 (_mm_mmask_i32gather_pd): Ditto.
40920 (_mm256_mmask_i64gather_ps): Ditto.
40921 (_mm_mmask_i64gather_ps): Ditto.
40922 (_mm256_mmask_i64gather_pd): Ditto.
40923 (_mm_mmask_i64gather_pd): Ditto.
40924 (_mm256_mmask_i32gather_epi32): Ditto.
40925 (_mm_mmask_i32gather_epi32): Ditto.
40926 (_mm256_mmask_i32gather_epi64): Ditto.
40927 (_mm_mmask_i32gather_epi64): Ditto.
40928 (_mm256_mmask_i64gather_epi32): Ditto.
40929 (_mm_mmask_i64gather_epi32): Ditto.
40930 (_mm256_mmask_i64gather_epi64): Ditto.
40931 (_mm_mmask_i64gather_epi64): Ditto.
40932 (_mm256_i32scatter_ps): Change __addr type to void*.
40933 (_mm256_mask_i32scatter_ps): Ditto.
40934 (_mm_i32scatter_ps): Ditto.
40935 (_mm_mask_i32scatter_ps): Ditto.
40936 (_mm256_i32scatter_pd): Ditto.
40937 (_mm256_mask_i32scatter_pd): Ditto.
40938 (_mm_i32scatter_pd): Ditto.
40939 (_mm_mask_i32scatter_pd): Ditto.
40940 (_mm256_i64scatter_ps): Ditto.
40941 (_mm256_mask_i64scatter_ps): Ditto.
40942 (_mm_i64scatter_ps): Ditto.
40943 (_mm_mask_i64scatter_ps): Ditto.
40944 (_mm256_i64scatter_pd): Ditto.
40945 (_mm256_mask_i64scatter_pd): Ditto.
40946 (_mm_i64scatter_pd): Ditto.
40947 (_mm_mask_i64scatter_pd): Ditto.
40948 (_mm256_i32scatter_epi32): Ditto.
40949 (_mm256_mask_i32scatter_epi32): Ditto.
40950 (_mm_i32scatter_epi32): Ditto.
40951 (_mm_mask_i32scatter_epi32): Ditto.
40952 (_mm256_i32scatter_epi64): Ditto.
40953 (_mm256_mask_i32scatter_epi64): Ditto.
40954 (_mm_i32scatter_epi64): Ditto.
40955 (_mm_mask_i32scatter_epi64): Ditto.
40956 (_mm256_i64scatter_epi32): Ditto.
40957 (_mm256_mask_i64scatter_epi32): Ditto.
40958 (_mm_i64scatter_epi32): Ditto.
40959 (_mm_mask_i64scatter_epi32): Ditto.
40960 (_mm256_i64scatter_epi64): Ditto.
40961 (_mm256_mask_i64scatter_epi64): Ditto.
40962 (_mm_i64scatter_epi64): Ditto.
40963 (_mm_mask_i64scatter_epi64): Ditto.
40964 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40965 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40966 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40967 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40968 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40969 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40970 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40971 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40972 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40973 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40974 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40975 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40976 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40977 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40978 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40979 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40980 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40981 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40982 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40983 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40984 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40985 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40986 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40987 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40988 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40989 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40990 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40991 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40992 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40993 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40994 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40995 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40996 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40997 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40998 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40999 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
41000 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
41001 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
41002 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
41003 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
41004 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
41005 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
41006 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
41007 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
41008 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
41009 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
41010 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
41011 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
41012 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
41013 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
41014 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
41015 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
41016 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
41017 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
41018 definitions accordingly.
41020 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
41021 Kuan-Lin Chen <kuanlinchentw@gmail.com>
41024 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
41027 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
41030 * ira-conflicts.c (ira_build_conflicts): Update total conflict
41031 hard regs for inner regno.
41033 2017-01-17 Martin Liska <mliska@suse.cz>
41036 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
41037 assumption and add comment.
41039 2017-01-17 Nathan Sidwell <nathan@acm.org>
41041 * ipa-visibility.c (localize_node): New function, broken out of ...
41042 (function_and_variable_visibility): ... here. Call it.
41044 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
41046 PR middle-end/77445
41047 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
41048 correctly set frequency of oudgoing edge.
41049 (duplicate_thread_path): Fix profile updating.
41051 2017-01-17 Jakub Jelinek <jakub@redhat.com>
41054 * configure.ac: Add GCC_BASE_VER.
41055 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
41056 version from BASE-VER file.
41057 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
41058 (gcc.o): Depend on $(BASEVER).
41059 * common.opt (dumpfullversion): New option.
41060 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
41061 * doc/invoke.texi: Document -dumpfullversion.
41062 * doc/install.texi: Document --with-gcc-major-version-only.
41063 * configure: Regenerated.
41065 2017-01-17 Richard Biener <rguenther@suse.de>
41067 PR tree-optimization/71433
41068 * tree-vrp.c (register_new_assert_for): Merge same asserts
41069 on all incoming edges.
41070 (process_assert_insertions_for): Handle insertions at the
41073 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
41075 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
41076 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
41078 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
41081 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
41084 2017-01-17 Alan Modra <amodra@gmail.com>
41087 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
41088 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
41089 symbolic stack limit when pic.
41091 2017-01-16 Martin Sebor <msebor@redhat.com>
41093 PR tree-optimization/78608
41094 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
41096 2017-01-16 Jeff Law <law@redhat.com>
41099 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
41100 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
41101 for several include directories that may be relative to sysroot.
41102 * config/i386/x-mingw32 (gplus_includedir): Define.
41103 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
41104 (native_system_includedir): Likewise.
41105 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
41106 override if TARGET_SYSTEM_ROOT is defined.
41107 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
41109 PR tree-optimization/79090
41110 PR tree-optimization/33562
41111 PR tree-optimization/61912
41112 PR tree-optimization/77485
41113 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
41114 and computed trims into the dump file.
41116 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
41118 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
41120 2017-01-16 Jakub Jelinek <jakub@redhat.com>
41123 * gimplify.c (gimplify_init_constructor): If want_value and
41124 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
41128 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
41129 sequence. Formatting fixes.
41130 (doloop_optimize): Formatting fixes.
41133 * gcc.c (debug_level_greater_than_spec_func): New function.
41134 (static_spec_functions): Add debug-level-gt spec function.
41135 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
41137 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
41138 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
41139 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
41140 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
41141 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
41142 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
41144 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
41146 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
41147 QImode fixups to general and mask registers only.
41149 2017-01-16 Carl Love <cel@us.ibm.com>
41151 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
41152 for built-in functions
41153 vector signed char vec_nabs (vector signed char)
41154 vector signed short vec_nabs (vector signed short)
41155 vector signed int vec_nabs (vector signed int)
41156 vector signed long long vec_nabs (vector signed long long)
41157 vector float vec_nabs (vector float)
41158 vector double vec_nabs (vector double)
41159 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
41161 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
41162 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
41163 * doc/extend.texi: Update the documentation file for the new built-in
41166 2017-01-16 Martin Sebor <msebor@redhat.com>
41168 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
41171 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41173 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
41174 UNSPEC_VSX__XXSPLTD to require special splat handling.
41176 2017-01-16 David Malcolm <dmalcolm@redhat.com>
41179 * system.h: Poison strndup.
41181 2017-01-16 Alan Modra <amodra@gmail.com>
41184 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
41187 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
41189 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
41191 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
41193 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
41194 call recog here. Assert that INSN_CODE (insn) is non-negative.
41196 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41199 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
41201 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
41202 in the currently scheduled RTL fragment.
41204 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41206 PR rtl-optimization/78751
41207 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
41210 2017-01-14 Jeff Law <law@redhat.com>
41212 PR tree-optimization/79090
41213 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
41214 variable length stores.
41215 (compute_trims): Delete dead assignment to *trim_tail.
41216 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
41219 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
41221 PR rtl-optimization/78626
41222 PR rtl-optimization/78727
41223 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
41224 of a block, and split such blocks after everything else is finished.
41226 2017-01-14 Alan Modra <amodra@gmail.com>
41229 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
41230 target legitimate_combined_insn.
41231 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
41232 (rs6000_legitimate_combined_insn): New function.
41233 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
41235 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
41236 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
41237 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
41239 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
41241 * doc/frontends.texi (G++ and GCC): Remove references to Java.
41243 2017-01-13 Jeff Law <law@redhat.com>
41245 PR tree-optimization/33562
41246 PR tree-optimization/61912
41247 PR tree-optimization/77485
41248 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
41250 (delete_dead_assignment): Likewise.
41251 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
41252 statement to delete_dead_call and delete_dead_assignment.
41254 2017-01-13 David Malcolm <dmalcolm@redhat.com>
41257 * substring-locations.c (format_warning_va): Strengthen case 1 so
41258 that both endpoints of the substring must be within the format
41259 range for just the substring to be printed.
41261 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
41263 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
41264 * config/i386/i386.c (ix86_target_string): Add missing options
41265 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
41266 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
41267 flags_other and ix86_target_other to flags2_other. Display unknown
41269 (ix86_valid_target_attribute_inner_p): Add missing options and
41270 reorder options by implied ISAs, as in ix86_target_string.
41272 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41274 * hash-table.h (hash_table::too_empty_p): New function.
41275 (hash_table::expand): Use it.
41276 (hash_table::traverse): Likewise.
41277 (hash_table::empty_slot): Use sizeof (value_type) instead of
41278 sizeof (PTR) to convert bytes to elements. Shrink the table
41279 if the current size is excessive for the current number of
41282 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41284 * ira-costs.c (record_reg_classes): Break from the inner loop
41285 early once alt_fail is known to be true. Update outer loop
41286 handling accordingly.
41288 2017-01-13 Jeff Law <law@redhat.com>
41290 * tree-ssa-dse.c (decrement_count): New function.
41291 (increment_start_addr, maybe_trim_memstar_call): Likewise.
41292 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
41293 when we know the partially dead statement is a mem* function.
41295 PR tree-optimization/61912
41296 PR tree-optimization/77485
41297 * tree-ssa-dse.c: Include expr.h.
41298 (maybe_trim_constructor_store): New function.
41299 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
41301 PR tree-optimization/33562
41302 PR tree-optimization/61912
41303 PR tree-optimization/77485
41304 * doc/invoke.texi: Document new dse-max-object-size param.
41305 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
41306 * tree-ssa-dse.c: Include params.h.
41307 (dse_store_status): New enum.
41308 (initialize_ao_ref_for_dse): New, partially extracted from
41310 (valid_ao_ref_for_dse, normalize_ref): New.
41311 (setup_live_bytes_from_ref, compute_trims): Likewise.
41312 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
41313 (maybe_trim_partially_dead_store): Likewise.
41314 (maybe_trim_complex_store): Likewise.
41315 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
41316 Track what bytes live from the original store. Return tri-state
41317 for dead, partially dead or live.
41318 (dse_dom_walker): Add constructor, destructor and new private members.
41319 (delete_dead_call, delete_dead_assignment): New extracted from
41321 (dse_optimize_stmt): Make a member of dse_dom_walker.
41322 Use initialize_ao_ref_for_dse.
41324 PR tree-optimization/33562
41325 PR tree-optimization/61912
41326 PR tree-optimization/77485
41327 * sbitmap.h (bitmap_count_bits): Prototype.
41328 (bitmap_clear_range, bitmap_set_range): Likewise.
41329 * sbitmap.c (bitmap_clear_range): New function.
41330 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
41332 2017-01-13 Martin Liska <mliska@suse.cz>
41335 * function.c (set_cfun): Add new argument force.
41336 * function.h (set_cfun): Likewise.
41337 * ipa-inline-transform.c (inline_call): Use the function when
41338 strict alising from is dropped for function we inline to.
41340 2017-01-13 Richard Biener <rguenther@suse.de>
41342 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
41343 for dumping GIMPLE INTEGER_CSTs.
41345 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
41347 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
41348 to 201112L since C++17.
41350 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
41353 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
41354 if -fsanitize=kernel-address is present.
41356 2017-01-13 Richard Biener <rguenther@suse.de>
41358 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
41359 as _Literal ( type ) number in case usual suffixes do not
41360 preserve all information.
41362 2017-01-13 Richard Biener <rguenther@suse.de>
41364 PR tree-optimization/77283
41365 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
41366 and ssa-iterators.h.
41367 (is_feasible_trace): Implement a cost model based on joiner
41370 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
41373 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
41374 char or short to __float128/_Float128 directly.
41376 2017-01-12 Martin Sebor <msebor@redhat.com>
41378 to -Wformat-overflow.
41379 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
41380 (min_bytes_remaining): Same.
41381 (get_string_length): Same.
41382 (format_string): Same.
41383 (format_directive): Same.
41385 (pass_sprintf_length::handle_gimple_call): Same.
41387 2017-01-12 Jakub Jelinek <jakub@redhat.com>
41389 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
41390 info.nowrite calls with no lhs that can't throw. Return bool
41391 whether gsi_remove has been called or not.
41392 (pass_sprintf_length::handle_gimple_call): Return bool whether
41393 try_substitute_return_value called gsi_remove. Formatting fix.
41394 (pass_sprintf_length::execute): Don't use gsi_remove if
41395 handle_gimple_call returned true.
41398 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
41399 be removed due to side-effects, don't remove following barrier nor
41400 turn the successor edge into fallthru edge.
41402 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41405 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
41406 element-reversing loads and stores as not swappable.
41408 2017-01-12 Nathan Sidwell <nathan@acm.org>
41409 Nicolai Stange <nicstange@gmail.com>
41411 * combine.c (try_combine): Don't ignore result of overlap checking
41412 loop. Combine overlap & asm check into single loop.
41414 2017-01-12 Richard Biener <rguenther@suse.de>
41416 * tree-pretty-print.c (dump_generic_node): Provide -gimple
41417 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
41419 2017-01-12 Richard Biener <rguenther@suse.de>
41421 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
41422 and TS_TARGET_OPTION directly derive from TS_BASE.
41423 * tree-core.h (tree_optimization_option): Derive from tree_base.
41424 (tree_target_option): Likewise.
41426 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
41428 * config/i386/i386.c (memory_address_length): Increase len
41429 only when rip_relative_addr_p returns false.
41431 2017-01-11 Julia Koval <julia.koval@intel.com>
41433 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
41434 (OPTION_MASK_ISA_SGX_SET): New.
41435 (ix86_handle_option): Handle OPT_msgx.
41436 * config.gcc: Added sgxintrin.h.
41437 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
41438 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
41439 * config/i386/i386.c (ix86_target_string): Add -msgx.
41441 (ix86_option_override_internal): Handle new options.
41442 (ix86_valid_target_attribute_inner_p): Add sgx.
41443 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
41444 * config/i386/i386.opt: Add msgx.
41445 * config/i386/sgxintrin.h: New file.
41446 * config/i386/x86intrin.h: Add sgxintrin.h.
41448 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41451 * fold-const.c (maybe_nonzero_address): Return 1 for function
41453 (tree_single_nonzero_warnv_p): Don't handle function local objects
41457 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
41460 2017-01-11 David Malcolm <dmalcolm@redhat.com>
41463 * opts.c: Include "spellcheck.h"
41464 (struct string_fragment): New struct.
41465 (struct edit_distance_traits<const string_fragment &>): New
41467 (get_closest_sanitizer_option): New function.
41468 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
41470 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41472 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
41474 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
41475 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
41476 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
41477 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
41478 for initial die_offset if dwarf_split_debug_info.
41479 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
41480 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
41482 (output_skeleton_debug_sections): Formatting fix. Use
41483 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
41484 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
41486 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
41488 * config/arm/cortex-a53.md: Add bypasses for
41489 cortex_a53_r2f_cvt.
41490 (cortex_a53_r2f): Only use for transfers.
41491 (cortex_a53_f2r): Likewise.
41492 (cortex_a53_r2f_cvt): Add reservation for conversions.
41493 (cortex_a53_f2r_cvt): Likewise.
41495 2017-01-11 Tamar Christina <tamar.christina@arm.com>
41497 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
41498 to all inlined functions, change static to extern.
41500 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
41503 * config/arm/arm.c (legitimize_pic_address): Handle reference to
41505 (arm_assemble_integer): Likewise.
41507 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
41509 * config.gcc: Use new awk script to check CPU, FPU and architecture
41510 parameters for --with-... options.
41511 * config/arm/parsecpu.awk: New file
41512 * config/arm/arm-cpus.in: New file.
41513 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
41515 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
41517 * config/arm/t-arm: Update dependency rules.
41518 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
41519 of processing .def files.
41520 * config/arm/genopt.sh: Deleted.
41521 * config/arm/gentune.sh: Deleted.
41522 * config/arm/arm-cores.def: Deleted.
41523 * config/arm/arm-arches.def: Deleted.
41524 * config/arm/arm-fpus.def: Deleted.
41525 * config/arm/arm-tune.md: Regenerated.
41526 * config/arm/arm-tables.opt: Regenerated.
41527 * config/arm/arm-cpu.h: New generated file.
41528 * config/arm/arm-cpu-data.h: New generated file.
41529 * config/arm/arm-cpu-cdata.h: New generated file.
41531 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
41534 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
41536 (input_varpool_node): Unpack dynamically_initialized bit.
41538 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
41540 PR rtl-optimization/79032
41541 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
41542 the alignment of the adjusted memory reference against that of MODE,
41543 instead of the alignment of the original memory reference.
41545 2017-01-11 Martin Jambor <mjambor@suse.cz>
41547 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
41549 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
41550 decorated functions.
41552 2017-01-11 Richard Biener <rguenther@suse.de>
41554 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
41555 set range/nonnull info for PHI results. Do not set it on
41556 stmts marked for removal.
41558 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
41560 * expr.c (store_field): In the bitfield case, fetch the return value
41561 from the registers before applying a single big-endian adjustment.
41562 Always do a final load for a BLKmode value not larger than a word.
41564 2017-01-10 David Malcolm <dmalcolm@redhat.com>
41567 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41568 that we correctly handle column numbers greater than
41569 LINE_MAP_MAX_COLUMN_NUMBER.
41571 2017-01-10 Martin Sebor <msebor@redhat.com>
41573 PR middle-end/78245
41574 * gimple-ssa-sprintf.c (get_destination_size): Call
41575 {init,fini}object_sizes.
41576 * tree-object-size.c (addr_object_size): Adjust.
41577 (pass_through_call): Adjust.
41578 (pass_object_sizes::execute): Adjust.
41579 * tree-object-size.h (fini_object_sizes): Declare.
41581 2017-01-10 Martin Sebor <msebor@redhat.com>
41583 PR tree-optimization/78775
41584 * builtins.c (get_size_range): Move...
41585 * calls.c: ...to here.
41586 (alloc_max_size): Accept zero argument.
41587 (operand_signed_p): Remove.
41588 (maybe_warn_alloc_args_overflow): Call get_size_range.
41589 * calls.h (get_size_range): Declare.
41591 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
41593 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
41594 from TI's devices.csv file as of September 2016.
41595 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
41597 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
41599 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
41600 * doc/invoke.texi: Likewise.
41601 * doc/md.texi: Likewise.
41602 * doc/objc.texi: Likewise.
41604 2017-01-10 Joshua Conner <joshconner@google.com>
41606 * config/arm/fuchsia-elf.h: New file.
41607 * config/fuchsia.h: New file.
41608 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
41609 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
41611 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
41613 2016-01-10 Richard Biener <rguenther@suse.de>
41615 PR tree-optimization/79034
41616 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
41617 Propagate out degenerate PHIs in the joiner.
41619 2017-01-10 Martin Liska <mliska@suse.cz>
41621 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
41622 (sort_congruence_classes_by_decl_uid): Likewise.
41623 (sort_congruence_class_groups_by_decl_uid): Likewise.
41624 (sem_item_optimizer::merge_classes): Sort class, groups in these
41625 classes and members in the groups by DECL_UID of declarations.
41626 This would make merge operations stable.
41628 2017-01-10 Martin Liska <mliska@suse.cz>
41630 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
41631 usage of m_classes_vec.
41632 (sem_item_optimizer::~sem_item_optimizer): Likewise.
41633 (sem_item_optimizer::get_group_by_hash): Likewise.
41634 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
41635 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
41636 (sem_item_optimizer::verify_classes): Likewise.
41637 (sem_item_optimizer::process_cong_reduction): Likewise.
41638 (sem_item_optimizer::dump_cong_classes): Likewise.
41639 (sem_item_optimizer::merge_classes): Likewise.
41640 * ipa-icf.h (congruence_class_hash): Rename from
41641 congruence_class_group_hash. Remove declaration of m_classes_vec.
41643 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
41645 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
41646 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
41647 * config.gcc: Add avx512vpopcntdqintrin.h.
41648 * config/i386/avx512vpopcntdqintrin.h: New.
41649 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
41650 * config/i386/i386-builtin-types.def: Add new types.
41651 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
41652 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
41653 __builtin_ia32_vpopcountq_v8di_mask): New.
41654 * config/i386/i386-c.c (ix86_target_macros_internal): Define
41655 __AVX512VPOPCNTDQ__.
41656 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
41657 (PTA_AVX512VPOPCNTDQ): Define.
41658 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
41659 TARGET_AVX512VPOPCNTDQ_P): Define.
41660 * config/i386/i386.opt: Add mavx512vpopcntdq.
41661 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
41662 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
41664 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41666 PR middle-end/77484
41667 * predict.def (PRED_CALL): Set to 67.
41669 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
41671 * expr.c (store_field): In the bitfield case, if the value comes from
41672 a function call and is of an aggregate type returned in registers, do
41673 not modify the field mode; extract the value in all cases if the mode
41674 is BLKmode and the size is not larger than a word.
41676 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
41679 * config/i386/cpuid.h: Fix undefined behavior.
41681 2017-01-04 Jeff Law <law@redhat.com>
41683 PR tree-optimization/79007
41684 PR tree-optimization/67955
41685 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
41686 conservative for pt.null when flag_non_call_exceptions is on.
41688 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41690 PR translation/79019
41691 PR translation/79020
41692 * params.def (PARAM_INLINE_MIN_SPEEDUP,
41693 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
41694 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
41696 * config/avr/avr.opt (maccumulate-args): Likewise.
41697 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
41698 * common.opt (freport-bug): Likewise.
41699 * cif-code.def (CIF_FINAL_ERROR): Likewise.
41700 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
41701 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
41702 translatable string.
41703 * config/i386/i386.c (function_value_32): Likewise.
41704 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
41705 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
41707 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
41708 * common/config/msp430/msp430-common.c (msp430_handle_option):
41710 * symtab.c (symtab_node::verify_base): Likewise.
41711 * opts.c (set_debug_level): Likewise.
41712 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
41713 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
41714 missing whitespace to translatable strings.
41715 * config/avr/avr.md (bswapsi2): Fix typo in comment.
41716 * config/sh/superh.h: Likewise.
41717 * config/i386/xopintrin.h: Likewise.
41718 * config/i386/znver1.md: Likewise.
41719 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
41720 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
41721 * double-int.h (struct double_int): Likewise.
41722 * double-int.c (div_and_round_double): Likewise.
41723 * wide-int.cc: Likewise.
41724 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
41725 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
41726 * cfgcleanup.c (crossjumps_occured): Renamed to ...
41727 (crossjumps_occurred): ... this.
41728 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
41731 PR tree-optimization/78899
41732 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
41733 returning bool return struct loop *, NULL for failure and the new
41735 (versionable_outer_loop_p): Don't version outer loop if it has
41736 dont_vectorized bit set.
41737 (tree_if_conversion): When versioning outer loop, ensure
41738 tree_if_conversion is performed also on the inner loop of the
41739 non-vectorizable outer loop copy.
41740 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
41741 LOOP_VECTORIZED in inner loop of the scalar outer loop and
41742 prevent vectorization of it.
41743 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
41744 the outer loop vectorization of the non-scalar version is attempted
41745 before vectorization of the inner loop in scalar version. If
41746 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
41747 vectorization of its inner loop.
41748 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
41749 has 2 inner loops, rename also on edges from bb whose single pred
41750 is outer_loop->header. Fix typo in function comment.
41752 2017-01-09 Martin Sebor <msebor@redhat.com>
41755 * asan.c (asan_emit_stack_protection): Increase local buffer size
41756 to avoid snprintf truncation warning.
41758 2017-01-09 Andrew Pinski <apinski@cavium.com>
41760 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
41761 to reference thunderx2t99 for the tuning structure
41762 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
41764 (thunderx2t99_extra_costs): This.
41765 * config/aarch64/aarch64-tune.md: Regenerate.
41766 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
41767 (vulcan_addrcost_table): This.
41768 (vulcan_regmove_cost): Rename to ...
41769 (thunderx2t99_regmove_cost): This.
41770 (vulcan_vector_cost): Rename to ...
41771 (thunderx2t99_vector_cost): this.
41772 (vulcan_branch_cost): Rename to ...
41773 (thunderx2t99_branch_cost): This.
41774 (vulcan_tunings): Rename to ...
41775 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
41776 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
41778 2017-01-09 Martin Jambor <mjambor@suse.cz>
41782 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
41783 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
41784 (propagate_vr_accross_jump_function): Use the above function for all
41785 value range computations for pass-through jump functions and type
41786 converasion from explicit value range values.
41787 (ipcp_propagate_stage): Do not attempt to deduce types of formal
41788 parameters from TYPE_ARG_TYPES.
41789 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41790 (ipa_write_node_info): Stream type of the actual argument.
41791 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41793 2017-01-09 Martin Liska <mliska@suse.cz>
41796 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41797 (lookup_compiler): Do not show error message with have_E.
41799 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41801 PR tree-optimization/78938
41802 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41803 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41804 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41805 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41808 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41810 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41813 2017-01-09 Richard Biener <rguenther@suse.de>
41815 PR tree-optimization/78997
41816 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41817 name condition properly.
41819 2017-01-09 Richard Biener <rguenther@suse.de>
41822 * dwarf2out.c (is_cxx): New overload with context.
41823 (is_naming_typedef_decl): Use it.
41825 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41827 * invoke.texi (Option Summary): Correct spacing in option lists
41828 and add line breaks to fix over-long lines.
41830 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41832 PR middle-end/17660
41834 * extend.texi (Common Variable Attributes): Add xref to GCC
41835 Internals manual to explain mode attribute keywords.
41837 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41840 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41841 and Preprocessor Options.
41842 (Options for Linking): Document -pthread here....
41843 (RS/6000 and PowerPC Options): ...not here.
41844 (Solaris 2 Options): ...or here.
41845 * doc/cppopts.texi: Document -pthread.
41847 2017-01-08 Martin Sebor <msebor@redhat.com>
41849 PR middle-end/77708
41850 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41851 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41852 New member functions.
41853 (format_directive): Used them.
41855 (pass_sprintf_length::handle_gimple_call): Same.
41856 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41857 to avoid truncation for any argument.
41858 (extract_affine_mul): Same.
41859 * tree.c (get_file_function_name): Same.
41861 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41863 PR middle-end/77484
41864 * predict.def (PRED_INDIR_CALL): Set to 86.
41866 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41868 PR preprocessor/54124
41869 * doc/cppopts.texi: Reformat -d subtable to list the full name
41870 of the options. Add cross-reference to the docs for the general
41871 compiler -d options.
41872 * doc/invoke.texi (Developer Options): Add cross-reference to the
41873 preprocessor-specific -d option documentation.
41875 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41877 PR preprocessor/13498
41878 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41879 redudant material, and reflect new command-line options.
41880 (System Headers): Likewise.
41882 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41884 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41885 -isystem, and -idirafter. Copy-edit.
41886 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41887 default for -ftrack-macro-expansion. Delete obsolete and
41888 badly-formatted implementation details about -fdebug-cpp output.
41889 * doc/cppwarnopts.texi: Copy-edit.
41891 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41894 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41895 that the transition from a max line width >= 1<<10 to narrower
41896 lines works correctly.
41898 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41900 * doc/options.texi (PerFunction): New.
41901 * opt-functions.awk (switch_flags): Map both Optimization and
41902 PerFunction to CL_OPTIMIZATION.
41903 * opth-gen.awk: Test for PerFunction flag along with
41905 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41906 it only when the latter is present. Skip those that don't in
41907 the hash function generator.
41908 * common.opt (fvar-tracking): Mark as PerFunction instead of
41910 (fvar-tracking-assignments): Likewise.
41911 (fvar-tracking-assignments-toggle): Likewise.
41912 (fvar-tracking-uninit): Likewise.
41914 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41916 PR translation/79018
41917 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41920 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41923 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41924 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41925 TARGET_LONG_JUMP_TABLE_OFFSETS.
41926 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41927 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41928 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41929 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41930 * config/m68k/m68k.md (tablejump expander): Likewise.
41931 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41932 TARGET_LONG_JUMP_TABLE_OFFSETS.
41933 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41934 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41936 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41937 David Holsgrove <david.holsgrove@xilinx.com>
41939 * common/config/microblaze/microblaze-common.c
41940 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41941 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41943 * config/microblaze/microblaze.c (microblaze_must_save_register)
41944 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41946 (microblaze_eh_return): New function.
41947 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41948 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41949 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41950 * config/microblaze/microblaze.md (eh_return): New pattern.
41952 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41954 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41955 GCC_DIAGNOSTIC_STRINGIFY): Define.
41957 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41959 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41961 * config/arm/arm.md (<mcrr>): New.
41963 * config/arm/arm.c (arm_arch5te): New.
41964 (arm_option_override): Set arm_arch5te.
41965 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41967 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41968 (arm_mcrr_qualifiers): ... this. New.
41969 (MRRC_QUALIFIERS): Define to...
41970 (arm_mrrc_qualifiers): ... this. New.
41971 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41973 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41974 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41975 (MRRCI, mrrc, MRRC): New.
41976 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41977 VUNSPEC_MRRC2): New.
41979 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41981 * config/arm/arm.md (<mcr>): New.
41983 * config/arm/arm.c (arm_coproc_builtin_available): Add
41984 support for mcr, mrc, mcr2 and mrc2.
41985 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41986 (arm_mcr_qualifiers): ... this. New.
41987 (MRC_QUALIFIERS): Define to ...
41988 (arm_mrc_qualifiers): ... this. New.
41989 (MCR_QUALIFIERS): Define to ...
41990 (arm_mcr_qualifiers): ... this. New.
41991 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41993 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41994 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41995 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41996 VUNSPEC_MRC2): New.
41998 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
42000 * config/arm/arm.md (*ldc): New.
42004 * config/arm/arm.c (arm_coproc_builtin_available): Add
42005 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
42006 (arm_coproc_ldc_stc_legitimate_address): New.
42007 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
42008 'qualifier_const_pointer'.
42009 (LDC_QUALIFIERS): Define to...
42010 (arm_ldc_qualifiers): ... this. New.
42011 (STC_QUALIFIERS): Define to...
42012 (arm_stc_qualifiers): ... this. New.
42013 * config/arm/arm-protos.h
42014 (arm_coproc_ldc_stc_legitimate_address): New.
42015 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
42016 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
42017 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
42018 stc2, stcl, stc2l): New.
42019 * config/arm/constraints.md (Uz): New.
42020 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
42021 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
42022 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
42023 VUNSPEC_STC2L): New.
42025 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
42027 * config/arm/arm.md (<cdp>): New.
42028 * config/arm/arm.c (neon_const_bounds): Rename this ...
42029 (arm_const_bounds): ... this.
42030 (arm_coproc_builtin_available): New.
42031 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
42032 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
42033 (CDP_QUALIFIERS): Define to...
42034 (arm_cdp_qualifiers): ... this. New.
42036 (arm_expand_builtin_args): Add case for 6 arguments.
42037 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
42038 (arm_const_bounds): ... this.
42039 (arm_coproc_builtin_available): New.
42040 * config/arm/arm_acle.h (__arm_cdp): New.
42042 * config/arm/arm_acle_builtins.def (cdp): New.
42044 * config/arm/iterators.md (CDPI,CDP,cdp): New.
42045 * config/arm/neon.md: Rename all 'neon_const_bounds' to
42046 'arm_const_bounds'.
42047 * config/arm/types.md (coproc): New.
42048 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
42049 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
42050 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
42051 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
42053 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
42055 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
42056 (UBINOP_QUALIFIERS): New.
42058 (acle_builtin_data): New. Change comment.
42059 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
42060 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
42061 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
42062 arm_acle_builtins.def.
42063 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
42064 (arm_init_acle_builtins): New.
42065 (CRC32_BUILTIN): Remove.
42066 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
42067 crc32cb, crc32ch and crc32cw.
42068 (arm_init_crc32_builtins): Remove.
42069 (arm_init_builtins): Use arm_init_acle_builtins rather
42070 than arm_init_crc32_builtins.
42071 (arm_expand_acle_builtin): New.
42072 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
42073 * config/arm/arm_acle_builtins.def: New.
42075 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
42077 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
42078 (arm_builtin_datum): ... this.
42079 (arm_init_neon_builtin): Rename to ...
42080 (arm_init_builtin): ... this. Add a new parameters PREFIX
42081 and USE_SIG_IN_NAME.
42082 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
42083 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
42084 'arm_builtin_datum'.
42085 (arm_init_vfp_builtins): Likewise.
42086 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
42087 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
42088 (arm_expand_neon_args): Rename to ...
42089 (arm_expand_builtin_args): ... this. Rename builtin_arg
42090 enum values and differentiate between ARG_BUILTIN_MEMORY
42091 and ARG_BUILTIN_NEON_MEMORY.
42092 (arm_expand_neon_builtin_1): Rename to ...
42093 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
42094 values, arm_expand_builtin_args and add bool parameter NEON.
42095 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
42096 (arm_expand_vfp_builtin): Likewise.
42097 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
42099 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42101 PR middle-end/77484
42102 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
42103 * predict.c (tree_estimate_probability_bb): Reverse direction of
42104 polymorphic call predictor.
42106 2017-01-06 David Malcolm <dmalcolm@redhat.com>
42108 * passes.c (execute_one_pass): Split out pass-skipping logic into...
42109 (determine_pass_name_match): ...this new function and...
42110 (should_skip_pass_p): ...this new function.
42112 2017-01-06 Nathan Sidwell <nathan@acm.org>
42114 * ipa-visibility.c (function_and_variable_visibility): Reformat
42115 comments and long lines. Remove extrneous if.
42116 * symtab.c (symtab_node::make_decl_local): Fix code format.
42117 (symtab_node::set_section_for_node): Fix comment typo.
42119 2017-01-06 Martin Liska <mliska@suse.cz>
42122 * lra-constraints.c: Rename invariant to lra_invariant.
42123 * predict.c (set_even_probabilities): Initialize e to NULL.
42125 2017-01-05 Martin Sebor <msebor@redhat.com>
42127 PR tree-optimization/78910
42128 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
42129 (format_integer): Correct off-by-one error in the handling
42130 of precision with negative numbers in signed conversions..
42132 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
42134 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
42136 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42138 PR tree-optimization/71016
42139 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
42140 factor_out_conditional_conversion. Formatting fix.
42141 (factor_out_conditional_conversion): Add cond_stmt argument.
42142 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
42143 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
42146 2017-01-05 David Malcolm <dmalcolm@redhat.com>
42148 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
42149 read-rtl-function.o, and selftest-rtl.o.
42150 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
42151 (selftest::aarch64_test_loading_full_dump): New function.
42152 (selftest::aarch64_run_selftests): New function.
42153 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
42154 selftest::aarch64_run_selftests.
42155 * config/i386/i386.c
42156 (selftest::ix86_test_loading_dump_fragment_1): New function.
42157 (selftest::ix86_test_loading_call_insn): New function.
42158 (selftest::ix86_test_loading_full_dump): New function.
42159 (selftest::ix86_test_loading_unspec): New function.
42160 (selftest::ix86_run_selftests): Call the new functions.
42161 * emit-rtl.c (maybe_set_max_label_num): New function.
42162 * emit-rtl.h (maybe_set_max_label_num): New decl.
42163 * function.c (instantiate_decls): Guard call to
42164 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
42165 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
42167 * gensupport.c (gen_reader::gen_reader): Pass "false"
42168 for new "compact" param of rtx_reader.
42169 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
42170 rather than an empty string for NULL strings.
42171 * read-md.c: Potentially include config.h rather than bconfig.h.
42172 Wrap include of errors.h with #ifdef GENERATOR_FILE.
42173 (have_error): New global, copied from errors.c.
42174 (md_reader::read_name): Rename to...
42175 (md_reader::read_name_1): ...this, adding "out_loc" param,
42176 and converting "missing name or number" to returning false, rather
42178 (md_reader::read_name): Reimplement in terms of read_name_1.
42179 (md_reader::read_name_or_nil): New function.
42180 (md_reader::read_string): Handle "(nil)" by returning NULL.
42181 (md_reader::md_reader): Add new param "compact".
42182 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
42183 (md_reader::read_file): New method.
42184 * read-md.h (md_reader::md_reader): Add new param "compact".
42185 (md_reader::read_file): New method.
42186 (md_reader::is_compact): New accessor.
42187 (md_reader::read_name): Convert return type from void to file_location.
42188 (md_reader::read_name_or_nil): New decl.
42189 (md_reader::read_name_1): New decl.
42190 (md_reader::m_compact): New field.
42191 (noop_reader::noop_reader): Pass "false" for new "compact" param
42193 (rtx_reader::rtx_reader): Add new "compact" param.
42194 (rtx_reader::read_rtx_operand): Make virtual and convert return
42195 type from void to rtx.
42196 (rtx_reader::read_until): New decl.
42197 (rtx_reader::handle_any_trailing_information): New virtual function.
42198 (rtx_reader::postprocess): New virtual function.
42199 (rtx_reader::finalize_string): New virtual function.
42200 (rtx_reader::m_in_call_function_usage): New field.
42201 (rtx_reader::m_reuse_rtx_by_id): New field.
42202 * read-rtl-function.c: New file.
42203 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
42204 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
42205 (selftest::verify_three_block_rtl_cfg): New decl.
42206 * read-rtl-function.h: New file.
42207 * read-rtl.c: Potentially include config.h rather than bconfig.h.
42208 For host, include function.h, memmodel.h, and emit-rtl.h.
42209 (one_time_initialization): New function.
42210 (struct compact_insn_name): New struct.
42211 (compact_insn_names): New array.
42212 (find_code): Handle insn codes in compact dumps.
42213 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
42214 (bind_subst_iter_and_attr): Likewise.
42215 (add_condition_to_string): Likewise.
42216 (add_condition_to_rtx): Likewise.
42217 (apply_attribute_uses): Likewise.
42218 (add_current_iterators): Likewise.
42219 (apply_iterators): Likewise.
42220 (initialize_iterators): Guard usage of apply_subst_iterator with
42221 #ifdef GENERATOR_FILE.
42222 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
42223 (md_reader::read_mapping): Likewise.
42224 (add_define_attr_for_define_subst): Likewise.
42225 (add_define_subst_attr): Likewise.
42226 (read_subst_mapping): Likewise.
42227 (check_code_iterator): Likewise.
42228 (rtx_reader::read_rtx): Likewise. Move one-time initialization
42230 (one_time_initialization): New function.
42231 (rtx_reader::read_until): New method.
42232 (read_flags): New function.
42233 (parse_reg_note_name): New function.
42234 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
42235 Handle reuse_rtx ids.
42236 Wrap iterator lookup within #ifdef GENERATOR_FILE.
42237 Add parsing support for RTL dumps, mirroring the special-cases in
42238 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
42239 values, and calling handle_any_trailing_information.
42240 (rtx_reader::read_rtx_operand): Convert return type from void
42241 to rtx, returning return_rtx. Handle case 'e'. Call
42242 finalize_string on XSTR and XTMPL fields.
42243 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
42244 "(nil)" values were omitted. Call the postprocess vfunc on the
42246 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
42247 class ctor. Initialize m_in_call_function_usage. Call
42248 one_time_initialization.
42249 * rtl-tests.c (selftest::test_uncond_jump): Call
42250 set_new_first_and_last_insn.
42251 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
42252 * selftest-rtl.c: New file.
42253 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
42254 (selftest::get_insn_by_uid): New decl.
42255 * selftest-run-tests.c (selftest::run_tests): Call
42256 read_rtl_function_c_tests.
42257 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
42258 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
42261 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
42263 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
42264 operands in a special way. Assert that pos+len <= mode precision.
42266 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42268 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
42269 3 argument Alias with unlimited for the negative form.
42270 (fno-vect-cost-model): Removed.
42272 2017-01-05 Martin Liska <mliska@suse.cz>
42274 * hsa-gen.c (gen_hsa_divmod): New function.
42275 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
42277 2017-01-05 Martin Liska <mliska@suse.cz>
42280 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
42283 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42285 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
42286 small constant length operands.
42288 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42290 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
42291 between loop iterations.
42293 2017-01-05 Martin Liska <mliska@suse.cz>
42296 * gimplify.c (gimplify_decl_expr): Compare to
42297 asan_poisoned_variables instread of checking flags.
42298 (gimplify_target_expr): Likewise.
42299 (gimplify_expr): Likewise.
42300 (gimplify_function_tree): Conditionally initialize
42301 asan_poisoned_variables.
42303 2017-01-04 Jeff Law <law@redhat.com>
42305 PR tree-optimizatin/78812
42306 * rtl.h (contains_mem_rtx_p): Prototype.
42307 * ifcvt.c (containts_mem_rtx_p): Move from here to...
42308 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
42309 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
42310 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
42311 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
42313 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42315 * input.c (assert_char_at_range): Default-initialize actual_range.
42317 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42319 * df-scan.c (df_ref_create_structure): Make regno unsigned,
42320 to match the caller.
42322 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42324 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
42325 insns after final jump in test to emit dummy move.
42327 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42329 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
42330 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
42332 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42334 * multiple_target.c (create_dispatcher_calls): Init e_next.
42335 * tree-ssa-loop-split.c (split_loop): Init border.
42336 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
42339 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
42344 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
42345 (altivec_register_operand): Do not return true if the operand
42346 contains a SUBREG mixing SImode and SFmode.
42347 (vsx_register_operand): Likewise.
42348 (vsx_reg_sfsubreg_ok): New predicate.
42349 (vfloat_operand): Do not return true if the operand contains a
42350 SUBREG mixing SImode and SFmode.
42351 (vint_operand): Likewise.
42352 (vlogical_operand): Likewise.
42353 (gpc_reg_operand): Likewise.
42354 (int_reg_operand): Likewise.
42355 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
42356 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
42357 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
42359 (rs6000_emit_move_si_sf_subreg): New helper function.
42360 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
42361 fixup SUBREGs involving SImode and SFmode.
42362 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
42363 numbers for the new peephole2 optimization.
42364 (peephole2 for SFmode unions): New peephole2 to optimize cases in
42365 the GLIBC math library that do AND/IOR/XOR operations on single
42366 precision floating point.
42367 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
42368 target macros to say whether we need to avoid SUBREGs mixing
42370 (TARGET_ALLOW_SF_SUBREG): Likewise.
42371 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
42372 (UNSPEC_SI_FROM_SF): Likewise.
42373 (iorxor): Change spacing.
42374 (and_ior_xor): New iterator for AND, IOR, and XOR.
42375 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
42376 (movdi_from_sf_zero_ext): Likewise.
42377 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
42378 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
42379 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
42380 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
42381 (fms<mode>4): Likewise.
42382 (fnma<mode>4): Likewise.
42383 (fnms<mode>4): Likewise.
42384 (nfma<mode>4): Likewise.
42385 (nfms<mode>4): Likewise.
42387 2017-01-04 Marek Polacek <polacek@redhat.com>
42390 * doc/invoke.texi: Document -Wpointer-compare.
42392 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42394 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
42397 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
42398 descriptions for -gdwarf-5 and emit them as uleb128 instead of
42401 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
42404 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
42405 documentation of the powerpc_popcntb_ok attribute.
42406 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
42407 code to issue warning messages if a requested CPU configuration is
42408 not supported by the binary (assembler and loader) toolchain.
42409 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
42410 made to define a built-in function that has been disabled.
42411 (paired_init_builtins): Add assertion to prevent ICE if attempt is
42412 made to define a built-in function that has been disabled.
42413 (altivec_init_builtins): Add comment explaining why definition
42414 of the DST built-in functions is not preceded by an assertion
42415 check. Add assertions to prevent ICE if attempts are made to
42416 define an altivec predicate or an abs* built-in function that has
42418 (htm_init_builtins): Add comment explaining why definition of the
42419 htm built-in functions is not preceded by an assertion check.
42421 2017-01-04 Jeff Law <law@redhat.com>
42423 PR tree-optimizatin/67955
42424 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
42425 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
42426 the points-to solution does not include pt_null. Use DECL_PT_UID
42429 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
42431 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
42432 Use gen_int_mode instead of gen_lopwart for const_int operands.
42434 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42436 PR tree-optimization/71563
42437 * match.pd: Simplify X << Y into X if Y is known to be 0 or
42438 out of range value - has low bits known to be zero.
42440 2017-01-04 Alan Modra <amodra@gmail.com>
42442 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
42443 * configure: Regenerate.
42444 * config.in: Regenerate.
42446 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42449 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
42450 a substring of the message, but strcmp with the whole message. Ifdef
42451 ENABLE_NLS, translate the message first using dgettext.
42453 2017-01-03 Jeff Law <law@redhat.com>
42455 PR tree-optimizatin/78856
42456 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
42457 (mark_threaded_blocks): Remove code to truncate thread paths that
42458 cross multiple loop headers. Instead invalidate the cached loop
42459 iteration information and handle case of a thread path walking
42460 into an irreducible region.
42462 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
42465 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
42466 assertions. Add support for doing the signbit if the IEEE 128-bit
42467 floating point value is in a GPR.
42468 * config/rs6000/rs6000.md (Fsignbit): Delete.
42469 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
42470 Update the length attribute if the value is in a GPR.
42471 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
42472 the sign or zero extension instruction, since the value is always 0/1.
42473 (signbit<mode>2_dm2): Delete using <Fsignbit>.
42476 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
42477 extracting SImode to a GPR register so that we can generate a
42478 store, limit the vector to be in a traditional Altivec register
42479 for the vextuwrx instruction.
42481 2017-01-03 Ian Lance Taylor <iant@google.com>
42483 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
42485 2017-01-03 Martin Sebor <msebor@redhat.com>
42487 PR tree-optimization/78696
42488 * gimple-ssa-sprintf.c (format_floating): Correct handling of
42489 precision. Use MPFR for %f for greater fidelity. Correct handling
42491 (pass_sprintf_length::compute_format_length): Set width and precision
42492 specified by asrerisk to void_node for vararg functions.
42493 (try_substitute_return_value): Adjust dump output.
42495 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
42497 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
42499 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
42501 * doc/invoke.texi (SPARC options): Document -mlra as the default.
42502 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
42503 -mlra/-mno-lra was passed to the compiler.
42505 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
42507 PR rtl-optimization/65618
42508 * emit-rtl.c (try_split): Move initialization of "before" and
42509 "after" to just before the call to emit_insn_after_setloc.
42511 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
42513 * doc/md.texi (Standard Names): Remove reference to Java frontend.
42515 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
42517 * dwarf2out.c (gen_enumeration_type_die): When
42518 -gno-strict-dwarf, add a DW_AT_encoding attribute.
42520 2017-01-03 Jakub Jelinek <jakub@redhat.com>
42522 PR tree-optimization/78965
42523 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
42524 Change first argument from const call_info & to call_info &. For %n
42525 set info.nowrite to false.
42527 PR middle-end/78901
42528 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
42529 possibly throwing calls.
42531 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
42532 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
42533 and fns handling, rather than in a separate case SSA_NAME.
42535 2017-01-02 Jeff Law <law@redhat.com>
42537 * config/darwin-driver.c (darwin_driver_init): Const-correctness
42538 fixes for first_period and second_period variables.
42540 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
42543 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
42544 (*insvqi_1): New insn pattern.
42545 (*insvqi_1_mem_rex64): Ditto.
42546 (*insvqi_2): Ditto.
42547 (*insvqi_3): Rename from *insvqi.
42549 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
42551 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
42553 * doc/cfg.texi (Edges): Remove reference to Java.
42554 (Maintaining the CFG): Ditto.
42556 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42558 PR middle-end/77674
42559 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
42560 transparent aliases.
42562 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42564 PR middle-end/77484
42565 * predict.def (PRED_CALL): Update hitrate.
42566 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
42567 * predict.c (tree_estimate_probability_bb): Split CALL predictor
42568 into direct/indirect/polymorphic variants.
42570 2017-01-01 Jakub Jelinek <jakub@redhat.com>
42572 Update copyright years.
42574 * gcc.c (process_command): Update copyright notice dates.
42575 * gcov-dump.c (print_version): Ditto.
42576 * gcov.c (print_version): Ditto.
42577 * gcov-tool.c (print_version): Ditto.
42578 * gengtype.c (create_file): Ditto.
42579 * doc/cpp.texi: Bump @copying's copyright year.
42580 * doc/cppinternals.texi: Ditto.
42581 * doc/gcc.texi: Ditto.
42582 * doc/gccint.texi: Ditto.
42583 * doc/gcov.texi: Ditto.
42584 * doc/install.texi: Ditto.
42585 * doc/invoke.texi: Ditto.
42587 Copyright (C) 2017 Free Software Foundation, Inc.
42589 Copying and distribution of this file, with or without modification,
42590 are permitted in any medium without royalty provided the copyright
42591 notice and this notice are preserved.