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 * combine.c (can_change_dest_mode): Handle polynomial
7 * expmed.c (store_bit_field_1): Likewise.
8 * expr.c (store_constructor): Likewise.
9 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
10 and polynomial REGMODE_NATURAL_SIZE.
11 (gen_lowpart_common): Likewise.
12 * reginfo.c (record_subregs_of_mode): Likewise.
13 * rtlanal.c (read_modify_subreg_p): Likewise.
15 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
16 Alan Hayward <alan.hayward@arm.com>
17 David Sherwood <david.sherwood@arm.com>
19 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
22 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23 Alan Hayward <alan.hayward@arm.com>
24 David Sherwood <david.sherwood@arm.com>
26 * match.pd: Cope with polynomial numbers of vector elements.
28 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29 Alan Hayward <alan.hayward@arm.com>
30 David Sherwood <david.sherwood@arm.com>
32 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
33 in a POINTER_PLUS_EXPR.
35 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
36 Alan Hayward <alan.hayward@arm.com>
37 David Sherwood <david.sherwood@arm.com>
39 * omp-simd-clone.c (simd_clone_subparts): New function.
40 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
41 (ipa_simd_modify_function_body): Likewise.
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 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
48 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
49 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
50 (expand_vector_condition, vector_element): Likewise.
51 (subparts_gt): New function.
52 (get_compute_type): Use subparts_gt.
53 (count_type_subparts): Delete.
54 (expand_vector_operations_1): Use subparts_gt instead of
57 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
58 Alan Hayward <alan.hayward@arm.com>
59 David Sherwood <david.sherwood@arm.com>
61 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
62 (vect_compile_time_alias): ...this new function. Do the calculation
63 on poly_ints rather than trees.
64 (vect_prune_runtime_alias_test_list): Update call accordingly.
66 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
67 Alan Hayward <alan.hayward@arm.com>
68 David Sherwood <david.sherwood@arm.com>
70 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
72 (vect_schedule_slp_instance): Likewise.
74 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
75 Alan Hayward <alan.hayward@arm.com>
76 David Sherwood <david.sherwood@arm.com>
78 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
79 constant and extern definitions for variable-length vectors.
80 (vect_get_constant_vectors): Note that the number of units
81 is known to be constant.
83 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
84 Alan Hayward <alan.hayward@arm.com>
85 David Sherwood <david.sherwood@arm.com>
87 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
88 of units as polynomial. Choose between WIDE and NARROW based
91 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
92 Alan Hayward <alan.hayward@arm.com>
93 David Sherwood <david.sherwood@arm.com>
95 * tree-vect-stmts.c (simd_clone_subparts): New function.
96 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
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_call): Treat the number of
103 vectors as polynomial. Use build_index_vector for
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 (get_load_store_type): Treat the number of
111 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
112 for variable-length vectors.
113 (vectorizable_mask_load_store): Treat the number of units as
114 polynomial, asserting that it is constant if the condition has
115 already been enforced.
116 (vectorizable_store, vectorizable_load): Likewise.
118 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
119 Alan Hayward <alan.hayward@arm.com>
120 David Sherwood <david.sherwood@arm.com>
122 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
123 of units as polynomial. Punt if we can't tell at compile time
124 which vector contains the final result.
126 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
127 Alan Hayward <alan.hayward@arm.com>
128 David Sherwood <david.sherwood@arm.com>
130 * tree-vect-loop.c (vectorizable_induction): Treat the number
131 of units as polynomial. Punt on SLP inductions. Use an integer
132 VEC_SERIES_EXPR for variable-length integer reductions. Use a
133 cast of such a series for variable-length floating-point
136 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
137 Alan Hayward <alan.hayward@arm.com>
138 David Sherwood <david.sherwood@arm.com>
140 * tree.h (build_index_vector): Declare.
141 * tree.c (build_index_vector): New function.
142 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
143 of units as polynomial, forcibly converting it to a constant if
144 vectorizable_reduction has already enforced the condition.
145 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
146 to create a {1,2,3,...} vector.
147 (vectorizable_reduction): Treat the number of units as polynomial.
148 Choose vectype_in based on the largest scalar element size rather
149 than the smallest number of units. Enforce the restrictions
152 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
153 Alan Hayward <alan.hayward@arm.com>
154 David Sherwood <david.sherwood@arm.com>
156 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
157 number of units as polynomial.
159 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
160 Alan Hayward <alan.hayward@arm.com>
161 David Sherwood <david.sherwood@arm.com>
163 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
164 * target.def (autovectorize_vector_sizes): Return the vector sizes
165 by pointer, using vector_sizes rather than a bitmask.
166 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
167 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
168 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
170 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
171 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
172 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
173 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
174 * omp-general.c (omp_max_vf): Likewise.
175 * omp-low.c (omp_clause_aligned_alignment): Likewise.
176 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
177 * tree-vect-loop.c (vect_analyze_loop): Likewise.
178 * tree-vect-slp.c (vect_slp_bb): Likewise.
179 * doc/tm.texi: Regenerate.
180 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
182 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
183 the vector size as a poly_uint64 rather than an unsigned int.
184 (current_vector_size): Change from an unsigned int to a poly_uint64.
185 (get_vectype_for_scalar_type): Update accordingly.
186 * tree.h (build_truth_vector_type): Take the size and number of
187 units as a poly_uint64 rather than an unsigned int.
188 (build_vector_type): Add a temporary overload that takes
189 the number of units as a poly_uint64 rather than an unsigned int.
190 * tree.c (make_vector_type): Likewise.
191 (build_truth_vector_type): Take the number of units as a poly_uint64
192 rather than an unsigned int.
194 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
195 Alan Hayward <alan.hayward@arm.com>
196 David Sherwood <david.sherwood@arm.com>
198 * target.def (get_mask_mode): Take the number of units and length
199 as poly_uint64s rather than unsigned ints.
200 * targhooks.h (default_get_mask_mode): Update accordingly.
201 * targhooks.c (default_get_mask_mode): Likewise.
202 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
203 * doc/tm.texi: Regenerate.
205 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
206 Alan Hayward <alan.hayward@arm.com>
207 David Sherwood <david.sherwood@arm.com>
209 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
210 * omp-general.c (omp_max_vf): Likewise.
211 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
212 (expand_omp_simd): Handle polynomial safelen.
213 * omp-low.c (omplow_simd_context): Add a default constructor.
214 (omplow_simd_context::max_vf): Change from int to poly_uint64.
215 (lower_rec_simd_input_clauses): Update accordingly.
216 (lower_rec_input_clauses): Likewise.
218 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
219 Alan Hayward <alan.hayward@arm.com>
220 David Sherwood <david.sherwood@arm.com>
222 * tree-vectorizer.h (vect_nunits_for_cost): New function.
223 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
224 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
225 (vect_analyze_slp_cost): Likewise.
226 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
227 (vect_model_load_cost): Likewise.
229 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
230 Alan Hayward <alan.hayward@arm.com>
231 David Sherwood <david.sherwood@arm.com>
233 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
234 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
235 from an unsigned int * to a poly_uint64_pod *.
236 (calculate_unrolling_factor): New function.
237 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
239 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
240 Alan Hayward <alan.hayward@arm.com>
241 David Sherwood <david.sherwood@arm.com>
243 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
244 from an unsigned int to a poly_uint64.
245 (_loop_vec_info::slp_unrolling_factor): Likewise.
246 (_loop_vec_info::vectorization_factor): Change from an int
248 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
249 (vect_get_num_vectors): New function.
250 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
251 (vect_get_num_copies): Use vect_get_num_vectors.
252 (vect_analyze_data_ref_dependences): Change max_vf from an int *
253 to an unsigned int *.
254 (vect_analyze_data_refs): Change min_vf from an int * to a
256 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
257 than an unsigned HOST_WIDE_INT.
258 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
259 (vect_analyze_data_ref_dependence): Change max_vf from an int *
260 to an unsigned int *.
261 (vect_analyze_data_ref_dependences): Likewise.
262 (vect_compute_data_ref_alignment): Handle polynomial vf.
263 (vect_enhance_data_refs_alignment): Likewise.
264 (vect_prune_runtime_alias_test_list): Likewise.
265 (vect_shift_permute_load_chain): Likewise.
266 (vect_supportable_dr_alignment): Likewise.
267 (dependence_distance_ge_vf): Take the vectorization factor as a
268 poly_uint64 rather than an unsigned HOST_WIDE_INT.
269 (vect_analyze_data_refs): Change min_vf from an int * to a
271 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
272 vfm1 as a poly_uint64 rather than an int. Make the same change
273 for the returned bound_scalar.
274 (vect_gen_vector_loop_niters): Handle polynomial vf.
275 (vect_do_peeling): Likewise. Update call to
276 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
277 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
279 * tree-vect-loop.c (vect_determine_vectorization_factor)
280 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
281 (vect_get_known_peeling_cost): Likewise.
282 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
283 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
284 (vect_transform_loop): Likewise. Use the lowest possible VF when
285 updating the upper bounds of the loop.
286 (vect_min_worthwhile_factor): Make static. Return an unsigned int
288 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
289 polynomial unroll factors.
290 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
291 (vect_make_slp_decision): Likewise.
292 (vect_supported_load_permutation_p): Likewise, and polynomial
294 (vect_analyze_slp_cost): Handle polynomial vf.
295 (vect_slp_analyze_node_operations): Likewise.
296 (vect_slp_analyze_bb_1): Likewise.
297 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
298 than an unsigned HOST_WIDE_INT.
299 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
300 (vectorizable_load): Handle polynomial vf.
301 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
303 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
305 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
306 Alan Hayward <alan.hayward@arm.com>
307 David Sherwood <david.sherwood@arm.com>
309 * match.pd: Handle bit operations involving three constants
310 and try to fold one pair.
312 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
314 * tree-vect-loop-manip.c: Include gimple-fold.h.
315 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
316 niters_maybe_zero parameters. Handle other cases besides a step of 1.
317 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
318 Add a path that uses a step of VF instead of 1, but disable it
320 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
321 and niters_no_overflow parameters. Update calls to
322 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
323 Create a new SSA name if the latter choses to use a ste other
324 than zero, and return it via niters_vector_mult_vf_var.
325 * tree-vect-loop.c (vect_transform_loop): Update calls to
326 vect_do_peeling, vect_gen_vector_loop_niters and
327 slpeel_make_loop_iterate_ntimes.
328 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
329 (vect_gen_vector_loop_niters): Update declarations after above changes.
331 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
333 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
334 128-bit round to integer instructions.
335 (ceil<mode>2): Likewise.
336 (btrunc<mode>2): Likewise.
337 (round<mode>2): Likewise.
339 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
341 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
342 unaligned VSX load/store on P8/P9.
343 (expand_block_clear): Allow the use of unaligned VSX
346 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
348 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
350 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
351 swap associated with both a load and a store.
353 2018-01-02 Andrew Waterman <andrew@sifive.com>
355 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
356 * config/riscv/riscv.md (clear_cache): Use it.
358 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
360 * web.c: Remove out-of-date comment.
362 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
364 * expr.c (fixup_args_size_notes): Check that any existing
365 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
366 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
367 (emit_single_push_insn): ...here.
369 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
371 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
372 (const_vector_encoded_nelts): New function.
373 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
374 (const_vector_int_elt, const_vector_elt): Declare.
375 * emit-rtl.c (const_vector_int_elt_1): New function.
376 (const_vector_elt): Likewise.
377 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
380 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
382 * expr.c: Include rtx-vector-builder.h.
383 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
384 directly on the tree encoding.
385 (const_vector_from_tree): Likewise.
386 * optabs.c: Include rtx-vector-builder.h.
387 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
388 sequence of "u" values.
389 * vec-perm-indices.c: Include rtx-vector-builder.h.
390 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
391 directly on the vec_perm_indices encoding.
393 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
395 * doc/rtl.texi (const_vector): Describe new encoding scheme.
396 * Makefile.in (OBJS): Add rtx-vector-builder.o.
397 * rtx-vector-builder.h: New file.
398 * rtx-vector-builder.c: Likewise.
399 * rtl.h (rtx_def::u2): Add a const_vector field.
400 (CONST_VECTOR_NPATTERNS): New macro.
401 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
402 (CONST_VECTOR_DUPLICATE_P): Likewise.
403 (CONST_VECTOR_STEPPED_P): Likewise.
404 (CONST_VECTOR_ENCODED_ELT): Likewise.
405 (const_vec_duplicate_p): Check for a duplicated vector encoding.
406 (unwrap_const_vec_duplicate): Likewise.
407 (const_vec_series_p): Check for a non-duplicated vector encoding.
408 Say that the function only returns true for integer vectors.
409 * emit-rtl.c: Include rtx-vector-builder.h.
410 (gen_const_vec_duplicate_1): Delete.
411 (gen_const_vector): Call gen_const_vec_duplicate instead of
412 gen_const_vec_duplicate_1.
413 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
414 (gen_const_vec_duplicate): Use rtx_vector_builder.
415 (gen_const_vec_series): Likewise.
416 (gen_rtx_CONST_VECTOR): Likewise.
417 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
418 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
419 Build a new vector rather than modifying a CONST_VECTOR in-place.
420 (handle_special_swappables): Update call accordingly.
421 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
422 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
423 Build a new vector rather than modifying a CONST_VECTOR in-place.
424 (handle_special_swappables): Update call accordingly.
426 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
428 * simplify-rtx.c (simplify_const_binary_operation): Use
429 CONST_VECTOR_ELT instead of XVECEXP.
431 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
433 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
434 the selector elements to be different from the data elements
435 if the selector is a VECTOR_CST.
436 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
437 ssizetype for the selector.
439 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
441 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
442 before testing each element individually.
443 * tree-vect-generic.c (lower_vec_perm): Likewise.
445 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
447 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
448 * selftest-run-tests.c (selftest::run_tests): Call it.
449 * vector-builder.h (vector_builder::operator ==): New function.
450 (vector_builder::operator !=): Likewise.
451 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
452 (vec_perm_indices::all_from_input_p): New function.
453 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
454 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
455 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
456 instead of reading the VECTOR_CST directly. Detect whether both
457 vector inputs are the same before constructing the vec_perm_indices,
458 and update the number of inputs argument accordingly. Use the
459 utility functions added above. Only construct sel2 if we need to.
461 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
463 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
464 the broadcast of the low byte.
465 (expand_mult_highpart): Use an explicit encoding for the permutes.
466 * optabs-query.c (can_mult_highpart_p): Likewise.
467 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
468 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
469 (vectorizable_bswap): Likewise.
470 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
471 explicit encoding for the power-of-2 permutes.
472 (vect_permute_store_chain): Likewise.
473 (vect_grouped_load_supported): Likewise.
474 (vect_permute_load_chain): Likewise.
476 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
478 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
479 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
480 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
481 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
482 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
483 (vect_gen_perm_mask_any): Likewise.
485 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
487 * int-vector-builder.h: New file.
488 * vec-perm-indices.h: Include int-vector-builder.h.
489 (vec_perm_indices): Redefine as an int_vector_builder.
490 (auto_vec_perm_indices): Delete.
491 (vec_perm_builder): Redefine as a stand-alone class.
492 (vec_perm_indices::vec_perm_indices): New function.
493 (vec_perm_indices::clamp): Likewise.
494 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
495 (vec_perm_indices::new_vector): New function.
496 (vec_perm_indices::new_expanded_vector): Update for new
497 vec_perm_indices class.
498 (vec_perm_indices::rotate_inputs): New function.
499 (vec_perm_indices::all_in_range_p): Operate directly on the
500 encoded form, without computing elided elements.
501 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
502 encoding. Update for new vec_perm_indices class.
503 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
504 the given vec_perm_builder.
505 (expand_vec_perm_var): Update vec_perm_builder constructor.
506 (expand_mult_highpart): Use vec_perm_builder instead of
507 auto_vec_perm_indices.
508 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
509 vec_perm_indices instead of auto_vec_perm_indices. Use a single
510 or double series encoding as appropriate.
511 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
512 vec_perm_indices instead of auto_vec_perm_indices.
513 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
514 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
515 (vect_permute_store_chain): Likewise.
516 (vect_grouped_load_supported): Likewise.
517 (vect_permute_load_chain): Likewise.
518 (vect_shift_permute_load_chain): Likewise.
519 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
520 (vect_transform_slp_perm_load): Likewise.
521 (vect_schedule_slp_instance): Likewise.
522 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
523 (vectorizable_mask_load_store): Likewise.
524 (vectorizable_bswap): Likewise.
525 (vectorizable_store): Likewise.
526 (vectorizable_load): Likewise.
527 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
528 vec_perm_indices instead of auto_vec_perm_indices. Use
529 tree_to_vec_perm_builder to read the vector from a tree.
530 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
531 vec_perm_builder instead of a vec_perm_indices.
532 (have_whole_vector_shift): Use vec_perm_builder and
533 vec_perm_indices instead of auto_vec_perm_indices. Leave the
534 truncation to calc_vec_perm_mask_for_shift.
535 (vect_create_epilog_for_reduction): Likewise.
536 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
537 from auto_vec_perm_indices to vec_perm_indices.
538 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
539 instead of changing individual elements.
540 (aarch64_vectorize_vec_perm_const): Use new_vector to install
541 the vector in d.perm.
542 * config/arm/arm.c (expand_vec_perm_d::perm): Change
543 from auto_vec_perm_indices to vec_perm_indices.
544 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
545 instead of changing individual elements.
546 (arm_vectorize_vec_perm_const): Use new_vector to install
547 the vector in d.perm.
548 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
549 Update vec_perm_builder constructor.
550 (rs6000_expand_interleave): Likewise.
551 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
552 (rs6000_expand_interleave): Likewise.
554 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
556 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
557 to qimode could truncate the indices.
558 * optabs.c (expand_vec_perm_var): Likewise.
560 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
562 * Makefile.in (OBJS): Add vec-perm-indices.o.
563 * vec-perm-indices.h: New file.
564 * vec-perm-indices.c: Likewise.
565 * target.h (vec_perm_indices): Replace with a forward class
567 (auto_vec_perm_indices): Move to vec-perm-indices.h.
568 * optabs.h: Include vec-perm-indices.h.
569 (expand_vec_perm): Delete.
570 (selector_fits_mode_p, expand_vec_perm_var): Declare.
571 (expand_vec_perm_const): Declare.
572 * target.def (vec_perm_const_ok): Replace with...
573 (vec_perm_const): ...this new hook.
574 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
575 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
576 * doc/tm.texi: Regenerate.
577 * optabs.def (vec_perm_const): Delete.
578 * doc/md.texi (vec_perm_const): Likewise.
579 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
580 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
581 expand_vec_perm for constant permutation vectors. Assert that
582 the mode of variable permutation vectors is the integer equivalent
583 of the mode that is being permuted.
584 * optabs-query.h (selector_fits_mode_p): Declare.
585 * optabs-query.c: Include vec-perm-indices.h.
586 (selector_fits_mode_p): New function.
587 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
588 is defined, instead of checking whether the vec_perm_const_optab
589 exists. Use targetm.vectorize.vec_perm_const instead of
590 targetm.vectorize.vec_perm_const_ok. Check whether the indices
591 fit in the vector mode before using a variable permute.
592 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
593 vec_perm_indices instead of an rtx.
594 (expand_vec_perm): Replace with...
595 (expand_vec_perm_const): ...this new function. Take the selector
596 as a vec_perm_indices rather than an rtx. Also take the mode of
597 the selector. Update call to shift_amt_for_vec_perm_mask.
598 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
599 Use vec_perm_indices::new_expanded_vector to expand the original
600 selector into bytes. Check whether the indices fit in the vector
601 mode before using a variable permute.
602 (expand_vec_perm_var): Make global.
603 (expand_mult_highpart): Use expand_vec_perm_const.
604 * fold-const.c: Includes vec-perm-indices.h.
605 * tree-ssa-forwprop.c: Likewise.
606 * tree-vect-data-refs.c: Likewise.
607 * tree-vect-generic.c: Likewise.
608 * tree-vect-loop.c: Likewise.
609 * tree-vect-slp.c: Likewise.
610 * tree-vect-stmts.c: Likewise.
611 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
613 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
614 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
615 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
616 (aarch64_vectorize_vec_perm_const): ...this new function.
617 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
618 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
619 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
620 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
621 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
622 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
623 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
625 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
626 check for NEON modes.
627 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
628 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
629 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
630 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
632 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
633 the old VEC_PERM_CONST conditions.
634 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
635 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
636 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
637 (ia64_vectorize_vec_perm_const_ok): Merge into...
638 (ia64_vectorize_vec_perm_const): ...this new function.
639 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
640 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
641 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
642 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
643 * config/mips/mips.c (mips_expand_vec_perm_const)
644 (mips_vectorize_vec_perm_const_ok): Merge into...
645 (mips_vectorize_vec_perm_const): ...this new function.
646 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
647 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
648 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
649 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
650 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
651 (rs6000_expand_vec_perm_const): Delete.
652 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
654 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
655 (altivec_expand_vec_perm_const_le): Take each operand individually.
656 Operate on constant selectors rather than rtxes.
657 (altivec_expand_vec_perm_const): Likewise. Update call to
658 altivec_expand_vec_perm_const_le.
659 (rs6000_expand_vec_perm_const): Delete.
660 (rs6000_vectorize_vec_perm_const_ok): Delete.
661 (rs6000_vectorize_vec_perm_const): New function.
662 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
663 an element count and rtx array.
664 (rs6000_expand_extract_even): Update call accordingly.
665 (rs6000_expand_interleave): Likewise.
666 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
667 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
668 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
669 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
670 (rs6000_expand_vec_perm_const): Delete.
671 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
672 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
673 (altivec_expand_vec_perm_const_le): Take each operand individually.
674 Operate on constant selectors rather than rtxes.
675 (altivec_expand_vec_perm_const): Likewise. Update call to
676 altivec_expand_vec_perm_const_le.
677 (rs6000_expand_vec_perm_const): Delete.
678 (rs6000_vectorize_vec_perm_const_ok): Delete.
679 (rs6000_vectorize_vec_perm_const): New function. Remove stray
680 reference to the SPE evmerge intructions.
681 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
682 an element count and rtx array.
683 (rs6000_expand_extract_even): Update call accordingly.
684 (rs6000_expand_interleave): Likewise.
685 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
686 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
688 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
690 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
692 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
693 vector mode and that that mode matches the mode of the data
695 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
696 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
697 directly using expand_vec_perm_1 when forcing selectors into
699 (expand_vec_perm_var): New function, split out from expand_vec_perm.
701 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
703 * optabs-query.h (can_vec_perm_p): Delete.
704 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
705 * optabs-query.c (can_vec_perm_p): Split into...
706 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
707 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
708 particular selector is valid.
709 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
710 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
711 (vect_grouped_load_supported): Likewise.
712 (vect_shift_permute_load_chain): Likewise.
713 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
714 (vect_transform_slp_perm_load): Likewise.
715 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
716 (vectorizable_bswap): Likewise.
717 (vect_gen_perm_mask_checked): Likewise.
718 * fold-const.c (fold_ternary_loc): Likewise. Don't take
719 implementations of variable permutation vectors into account
720 when deciding which selector to use.
721 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
722 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
723 with a false third argument.
724 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
725 to test whether the constant selector is valid and can_vec_perm_var_p
726 to test whether a variable selector is valid.
728 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
730 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
731 * optabs-query.c (can_vec_perm_p): Likewise.
732 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
733 instead of vec_perm_indices.
734 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
735 (vect_gen_perm_mask_checked): Likewise,
736 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
737 (vect_gen_perm_mask_checked): Likewise,
739 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
741 * optabs-query.h (qimode_for_vec_perm): Declare.
742 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
743 (qimode_for_vec_perm): ...this new function.
744 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
746 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
748 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
749 does not have a conditional at the top.
751 2018-01-02 Richard Biener <rguenther@suse.de>
753 * ipa-inline.c (big_speedup_p): Fix expression.
755 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
758 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
761 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
765 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
766 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
767 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
768 cond_taken_branch_cost 3->4.
770 2018-01-01 Jakub Jelinek <jakub@redhat.com>
772 PR tree-optimization/83581
773 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
774 TODO_cleanup_cfg if any changes have been made.
777 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
778 convert_modes if target mode has the right side, but different mode
782 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
783 last argument when extracting from CONCAT. If either from_real or
784 from_imag is NULL, use expansion through memory. If result is not
785 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
786 the parts directly to inner mode, if even that fails, use expansion
790 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
791 check for bswap in mode rather than HImode and use that in expand_unop
794 2017-12-31 Jakub Jelinek <jakub@redhat.com>
797 * config/i386/i386.c (ix86_attribute_table): Remove excess
798 initializer for "shared" attribute.
800 2017-12-30 Tom de Vries <tom@codesourcery.com>
803 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
805 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
806 offload_funcs entries that no longer have a corresponding cgraph_node.
807 Mark the remaining ones as DECL_PRESERVE_P.
808 (output_lto): Call prune_offload_funcs.
810 2017-12-30 Jakub Jelinek <jakub@redhat.com>
812 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
813 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
814 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
815 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
816 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
817 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
818 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
819 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
820 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
821 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
822 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
823 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
824 vaesenc_<mode>, vpclmulqdq_<mode>,
825 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
827 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
829 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
830 _Float<N> and _Float<N>X variants for rounding built-in
832 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
833 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
834 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
835 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
836 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
837 * builtins.c (mathfn_built_in_2): Likewise.
838 * internal-fn.def (CEIL): Likewise.
840 (NEARBYINT): Likewise.
844 * convert.c (convert_to_integer_1): Likewise.
845 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
846 (integer_valued_real_call_p): Likewise.
847 * fold-const-call.c (fold_const_call_ss): Likewise.
848 * gencfn-macros.c (print_case_cfn): Change CFN and operator
849 printers to take a const char * suffix instead of a bool.
850 (print_define_operator_list): Likewise.
851 (fltall_suffixes): New list of suffixes, that include the
852 traditional suffixes as well as all of the _Float<N> and
854 (main): For _Float<N> and _Float<N>X functions, emit both
855 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
856 has the _Float<N> and _Float<N>X case names or operators. The
857 <name>_ALL variant has both the traditional and the
858 _Float<N>/_Float<N>X case names or operators.
859 * match.pd (COPYSIGN optimizations): Provide optimizations for
860 _Float<N> and _Float<N>X types where possible.
861 (MIN/MAX optimizations): Likewise.
862 (sqrt optimizations): Likewise.
863 (rounding optimizations): Likewise.
865 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
867 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
868 gen_int_mode rather than GEN_INT.
869 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
871 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
873 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
874 instead of CONSTANT_P.
875 (gen_vec_series): Likewise.
876 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
878 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
880 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
883 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
885 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
886 (aarch64_ld1x2<VDC:mode>): Likewise.
887 (aarch64_simd_ld1<mode>_x2): Likewise.
888 (aarch64_simd_ld1<mode>_x2): Likewise.
889 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
890 (vld1_s8_x2): Likewise.
891 (vld1_u16_x2): Likewise.
892 (vld1_s16_x2): Likewise.
893 (vld1_u32_x2): Likewise.
894 (vld1_s32_x2): Likewise.
895 (vld1_u64_x2): Likewise.
896 (vld1_s64_x2): Likewise.
897 (vld1_f16_x2): Likewise.
898 (vld1_f32_x2): Likewise.
899 (vld1_f64_x2): Likewise.
900 (vld1_p8_x2): Likewise.
901 (vld1_p16_x2): Likewise.
902 (vld1_p64_x2): Likewise.
903 (vld1q_u8_x2): Likewise.
904 (vld1q_s8_x2): Likewise.
905 (vld1q_u16_x2): Likewise.
906 (vld1q_s16_x2): Likewise.
907 (vld1q_u32_x2): Likewise.
908 (vld1q_s32_x2): Likewise.
909 (vld1q_u64_x2): Likewise.
910 (vld1q_s64_x2): Likewise.
911 (vld1q_f16_x2): Likewise.
912 (vld1q_f32_x2): Likewise.
913 (vld1q_f64_x2): Likewise.
914 (vld1q_p8_x2): Likewise.
915 (vld1q_p16_x2): Likewise.
916 (vld1q_p64_x2): Likewise.
918 2017-12-27 Martin Liska <mliska@suse.cz>
920 PR tree-optimization/83552
921 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
922 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
924 2017-12-27 Tom de Vries <tom@codesourcery.com>
926 * config/nvptx/nvptx.c (nvptx_option_override): Disable
927 -gstatement-frontiers.
929 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
931 PR rtl-optimization/83513
932 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
933 before priority comparison.
935 2017-12-25 Jakub Jelinek <jakub@redhat.com>
938 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
939 ix86_isa_flags2 to ix86_isa_flags.
940 * config/i386/i386-c.c (ix86_target_macros_internal): Test
941 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
942 isa_flags rather than isa_flags2.
943 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
944 and -mavx512bitalg from isa2_opts to isa_opts.
945 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
946 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
947 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
949 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
950 OPTION_MASK_ISA_AVX512F ored with another option similarly to
951 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
952 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
953 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
954 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
955 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
956 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
957 builtins from bdesc_args2 to bdesc_args section.
958 (__builtin_ia32_compressstoreuqi512_mask,
959 __builtin_ia32_compressstoreuhi512_mask,
960 __builtin_ia32_compressstoreuqi256_mask,
961 __builtin_ia32_expandloadqi512_mask,
962 __builtin_ia32_expandloadqi512_maskz,
963 __builtin_ia32_expandloadhi512_mask,
964 __builtin_ia32_expandloadhi512_maskz,
965 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
966 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
967 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
968 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
969 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
970 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
971 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
972 __builtin_ia32_vpshldv_v32hi_maskz,
973 __builtin_ia32_vpopcountb_v64qi_mask,
974 __builtin_ia32_vpopcountw_v32hi_mask,
975 __builtin_ia32_vpshufbitqmb512_mask,
976 __builtin_ia32_vpshufbitqmb256_mask): Add
977 " | OPTION_MASK_ISA_AVX512BW".
978 (__builtin_ia32_expandloadqi256_mask,
979 __builtin_ia32_expandloadqi256_maskz,
980 __builtin_ia32_vpopcountb_v32qi_mask): Add
981 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
982 (__builtin_ia32_expandloadhi256_mask,
983 __builtin_ia32_expandloadhi256_maskz,
984 __builtin_ia32_expandloadqi128_mask,
985 __builtin_ia32_expandloadqi128_maskz,
986 __builtin_ia32_expandloadhi128_mask,
987 __builtin_ia32_expandloadhi128_maskz,
988 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
989 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
990 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
991 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
992 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
993 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
994 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
995 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
996 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
997 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
998 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
999 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
1000 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
1001 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
1002 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
1003 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
1004 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
1005 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
1006 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
1007 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
1008 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
1009 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
1010 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
1011 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
1012 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
1013 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
1014 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
1015 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
1016 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
1017 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
1018 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
1019 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
1020 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
1021 __builtin_ia32_vpopcountw_v8hi_mask): Add
1022 " | OPTION_MASK_ISA_AVX512VL".
1023 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1024 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
1025 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
1026 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
1027 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
1028 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
1029 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
1030 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
1031 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
1032 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
1033 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
1034 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
1035 _mm_mask_bitshuffle_epi64_mask): Likewise.
1036 * common/config/i386/i386-common.c
1037 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
1038 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
1039 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1040 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
1041 OPTION_MASK_ISA_AVX512BITALG_UNSET.
1042 (OPTION_MASK_ISA2_AVX512F_UNSET,
1043 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
1044 (ix86_handle_option): For -mno-general-regs-only, clear from
1045 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
1046 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
1047 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
1048 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
1049 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
1050 for moving from ix86_isa_flags2 to ix86_isa_flags.
1052 2017-12-23 Jakub Jelinek <jakub@redhat.com>
1055 * fold-const.c (struct contains_label_data): New type.
1056 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
1057 inside of a SWITCH_BODY seen during the walk.
1058 (contains_label_p): Use walk_tree instead of
1059 walk_tree_without_duplicates, prepare data for contains_label_1 and
1062 2017-12-22 Jakub Jelinek <jakub@redhat.com>
1065 * tree-iterator.c (alloc_stmt_list): Start with cleared
1066 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
1070 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
1071 enable avx512f explicitly in #pragma GCC target.
1072 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
1073 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
1074 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
1075 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
1076 __builtin_ia32_vpdpbusds_v8si_mask,
1077 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
1078 __builtin_ia32_vpdpbusds_v4si_mask,
1079 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
1080 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
1081 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
1082 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
1083 __builtin_ia32_vpdpwssds_v8si_mask,
1084 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
1085 __builtin_ia32_vpdpwssds_v4si_mask,
1086 __builtin_ia32_vpdpwssds_v4si_maskz): Use
1087 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
1088 just OPTION_MASK_ISA_AVX512VNNI.
1090 2017-12-22 Martin Jambor <mjambor@suse.cz>
1093 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
1096 2017-12-22 Julia Koval <julia.koval@intel.com>
1097 Sebastian Peryt <sebastian.peryt@intel.com>
1099 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
1100 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
1101 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
1102 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
1103 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
1104 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
1105 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
1106 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
1107 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
1108 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
1109 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
1110 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
1111 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
1112 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
1113 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
1114 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
1115 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
1116 _mm256_maskz_popcnt_epi64): New intrinsics.
1117 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
1118 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
1119 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
1120 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
1121 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
1122 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
1123 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
1124 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
1125 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
1126 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
1127 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
1128 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
1129 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
1130 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
1131 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
1132 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
1133 __builtin_ia32_vpshufbitqmb256_mask,
1134 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
1135 * config/i386/i386-c.c (__AVX512BITALG__): New.
1136 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
1137 (ix86_valid_target_attribute_inner_p): Ditto.
1138 (ix86_expand_args_builtin): Handle new types.
1139 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
1140 * config/i386/i386.opt: Add -mavx512bitalg.
1141 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
1142 avx512bitalgintrin.h.
1143 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
1144 (vpopcount<mode><mask_name>): Add more types.
1145 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
1146 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
1148 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1150 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
1151 Or in OPTION_MASK_ISA_AVX512F_SET.
1152 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1153 OPTION_MASK_ISA_AVX512VNNI_UNSET.
1154 (ix86_handle_option): Adjust for
1155 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
1156 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
1158 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
1159 isa_flag instead of isa_flag2.
1160 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
1161 isa_opts2 to isa_opts.
1162 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
1165 2017-12-22 Mike Stump <mikestump@comcast.net>
1166 Eric Botcazou <ebotcazou@adacore.com>
1168 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
1170 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1173 * config/i386/i386.c (ix86_function_arg_boundary): Return
1174 PARM_BOUNDARY for TYPE_EMPTY_P types.
1177 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
1178 if navail is >= dir.len.
1180 2017-12-21 Steve Ellcey <sellcey@cavium.com>
1182 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
1185 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1187 PR rtl-optimization/80747
1188 PR rtl-optimization/83512
1189 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
1190 succ edge from ENTRY, copy partition from e->dest to the newly
1192 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
1193 ENTRY, use BB_PARTITION of its successor block as current_partition.
1194 Don't copy partition when splitting succ edge from ENTRY.
1196 PR tree-optimization/83523
1197 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
1198 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
1199 (convert_mult_to_fma): Likewise.
1201 PR tree-optimization/83521
1202 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
1203 gimple_build_assign without code on result of
1204 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
1205 a VIEW_CONVERT_EXPR.
1207 2017-12-21 Andrew Pinski <apinski@cavium.com>
1208 Steve Ellcey <sellcey@cavium.com>
1210 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
1211 multi-arch for ilp32.
1213 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
1216 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
1217 constraints to operand 2.
1218 (*ashl<mode>3_mask_1): Ditto.
1219 (*<shift_insn><mode>3_mask): Ditto.
1220 (*<shift_insn><mode>3_mask_1): Ditto.
1221 (*<rotate_insn><mode>3_mask): Ditto.
1222 (*<rotate_insn><mode>3_mask_1): Ditto.
1224 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
1226 * reorg.c (make_return_insns): Reemit each insn with its own location.
1228 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1230 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
1231 generation for cases where splatting a value is not useful.
1232 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1233 across a vec_duplicate and a paradoxical subreg forming a vector
1234 mode to a vec_concat.
1236 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1238 * combine.c (simplify_set): Do not transform subregs to zero_extends
1239 if the destination is not a scalar int mode.
1241 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
1244 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
1245 shared zero if the input has overflowed.
1247 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1249 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
1250 support for Cortex-A55 and Cortex-A75.
1252 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1254 * common/config/arm/arm-common.c (compare_opt_names): Add function
1255 comment. Use strcmp instead of manual loop.
1257 2017-12-21 Martin Liska <mliska@suse.cz>
1259 PR gcov-profile/83509
1260 * gcov-dump.c (dump_gcov_file): Do not read info about
1261 support_unexecuted_blocks for gcda files.
1263 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1265 PR rtl-optimization/82973
1266 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
1267 (valid_for_const_vector_p): ... this.
1268 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
1269 (valid_for_const_vector_p): ... this. Adjust function comment.
1270 (gen_vec_duplicate): Adjust caller.
1271 * optabs.c (expand_vector_broadcast): Likewise.
1272 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
1273 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
1275 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
1276 instead of GET_CODE == CONST_FIXED.
1277 (simplify_subreg): Use CONST_FIXED_P macro instead of
1278 GET_CODE == CONST_FIXED.
1281 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
1282 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
1283 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
1284 (ix86_option_override_internal): Adjust for
1285 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
1286 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
1287 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
1288 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
1289 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
1290 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
1291 def_builtin2 for CET builtins.
1292 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
1293 up formatting in IX86_BUILTIN_RDPID code.
1294 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
1295 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
1296 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
1297 ix86_isa_flags2 to ix86_isa_flags.
1298 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
1300 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
1301 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
1302 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
1304 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
1305 Or in OPTION_MASK_ISA_AVX512F_SET.
1306 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1307 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
1308 (ix86_handle_option): Adjust for
1309 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
1310 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
1312 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1313 Alan Hayward <alan.hayward@arm.com>
1314 David Sherwood <david.sherwood@arm.com>
1316 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
1317 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1318 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
1319 Track polynomial offsets.
1321 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1322 Alan Hayward <alan.hayward@arm.com>
1323 David Sherwood <david.sherwood@arm.com>
1325 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1326 Treat drb->init as a poly_int. Fail if its misalignment wrt
1327 vector_alignment isn't known.
1329 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1330 Alan Hayward <alan.hayward@arm.com>
1331 David Sherwood <david.sherwood@arm.com>
1333 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
1335 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
1336 (vect_loop_versioning): Take the loop versioning threshold as a
1338 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
1339 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
1340 versioning_threshold.
1341 (vect_analyze_loop_2): Compute the loop versioning threshold
1342 whenever loop versioning is needed, and store it in the new
1343 field rather than combining it with the cost model threshold.
1344 (vect_transform_loop): Update call to vect_loop_versioning.
1345 Try to combine the loop versioning and cost thresholds here.
1347 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1348 Alan Hayward <alan.hayward@arm.com>
1349 David Sherwood <david.sherwood@arm.com>
1351 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
1352 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
1353 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
1354 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
1355 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
1356 (group_compare_offset): Update accordingly.
1357 (split_small_address_groups_p): Likewise.
1358 (record_use): Take addr_offset as a poly_uint64 rather than
1359 an unsigned HOST_WIDE_INT.
1360 (strip_offset): Return the offset as a poly_uint64 rather than
1361 an unsigned HOST_WIDE_INT.
1362 (record_group_use, split_address_groups): Track polynomial offsets.
1363 (add_iv_candidate_for_use): Likewise.
1364 (addr_offset_valid_p): Take the offset as a poly_int64 rather
1365 than a HOST_WIDE_INT.
1366 (strip_offset_1): Return the offset as a poly_int64 rather than
1369 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1370 Alan Hayward <alan.hayward@arm.com>
1371 David Sherwood <david.sherwood@arm.com>
1373 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
1374 rather than a HOST_WIDE_INT.
1375 * tree.c (get_binfo_at_offset): Likewise.
1377 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1378 Alan Hayward <alan.hayward@arm.com>
1379 David Sherwood <david.sherwood@arm.com>
1381 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
1382 rather than a HOST_WIDE_INT.
1383 * tree-sra.c (build_ref_for_offset): Likewise.
1385 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1386 Alan Hayward <alan.hayward@arm.com>
1387 David Sherwood <david.sherwood@arm.com>
1389 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
1391 * tree.c (mem_ref_offset): Likewise.
1392 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
1393 * builtins.c (get_object_alignment_2): Likewise.
1394 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
1395 * gimple-fold.c (get_base_constructor): Likewise.
1396 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
1397 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1399 * ipa-polymorphic-call.c
1400 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1401 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
1402 (get_ancestor_addr_info): Likewise.
1403 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
1404 * match.pd: Likewise.
1405 * tree-data-ref.c (dr_analyze_innermost): Likewise.
1406 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1407 * tree-eh.c (tree_could_trap_p): Likewise.
1408 * tree-object-size.c (addr_object_size): Likewise.
1409 * tree-ssa-address.c (copy_ref_info): Likewise.
1410 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
1411 (indirect_refs_may_alias_p): Likewise.
1412 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1413 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
1414 (non_rewritable_mem_ref_base): Likewise.
1415 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1416 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1417 * varasm.c (decode_addr_const): Likewise.
1419 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1420 Alan Hayward <alan.hayward@arm.com>
1421 David Sherwood <david.sherwood@arm.com>
1423 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
1424 offsets for MEM_REFs.
1426 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1427 Alan Hayward <alan.hayward@arm.com>
1428 David Sherwood <david.sherwood@arm.com>
1430 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
1431 as a poly_uint64 rather than an unsigned int.
1432 * tree-ssanames.c (adjust_ptr_info_misalignment): 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 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
1440 (decode_addr_const): Update accordingly.
1442 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1443 Alan Hayward <alan.hayward@arm.com>
1444 David Sherwood <david.sherwood@arm.com>
1446 * tree.h (bit_field_size, bit_field_offset): New functions.
1447 * hsa-gen.c (gen_hsa_addr): Use them.
1448 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
1449 (simplify_vector_constructor): Likewise.
1450 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1451 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
1452 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
1453 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
1456 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1457 Alan Hayward <alan.hayward@arm.com>
1458 David Sherwood <david.sherwood@arm.com>
1460 * expr.h (emit_group_load, emit_group_load_into_temps)
1461 (emit_group_store): Take the size as a poly_int64 rather than an int.
1462 * expr.c (emit_group_load_1, emit_group_load): Likewise.
1463 (emit_group_load_into_temp, emit_group_store): Likewise.
1465 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1466 Alan Hayward <alan.hayward@arm.com>
1467 David Sherwood <david.sherwood@arm.com>
1469 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
1470 an unsigned int to a poly_uint64.
1471 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
1472 sizes as poly_uint64s rather than unsigned ints.
1473 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
1476 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1477 Alan Hayward <alan.hayward@arm.com>
1478 David Sherwood <david.sherwood@arm.com>
1480 * lra-constraints.c (emit_inc): Change inc_amount from an int
1483 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1484 Alan Hayward <alan.hayward@arm.com>
1485 David Sherwood <david.sherwood@arm.com>
1487 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
1489 (add_stack_var, stack_var_cmp, partition_stack_vars)
1490 (dump_stack_var_partition): Update accordingly.
1491 (alloc_stack_frame_space): Take the size as a poly_int64 rather
1492 than a HOST_WIDE_INT.
1493 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
1494 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
1495 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
1496 rather than a HOST_WIDE_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 * target.def (return_pops_args): Treat both the input and output
1503 sizes as poly_int64s rather than HOST_WIDE_INTS.
1504 * targhooks.h (default_return_pops_args): Update accordingly.
1505 * targhooks.c (default_return_pops_args): Likewise.
1506 * doc/tm.texi: Regenerate.
1507 * emit-rtl.h (incoming_args): Change pops_args, size and
1508 outgoing_args_size from int to poly_int64_pod.
1509 * function.h (expr_status): Change x_pending_stack_adjust and
1510 x_stack_pointer_delta from int to poly_int64.
1511 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
1512 (ARGS_SIZE_RTX): Update accordingly.
1513 * calls.c (highest_outgoing_arg_in_use): Change from int to
1515 (stack_usage_watermark, stored_args_watermark): New variables.
1516 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
1517 (emit_call_1): Change the stack_size and rounded_stack_size
1518 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
1520 (save_fixed_argument_area): Check stack_usage_watermark.
1521 (initialize_argument_information): Change old_pending_adj from
1522 a HOST_WIDE_INT * to a poly_int64_pod *.
1523 (compute_argument_block_size): Return the size as a poly_int64
1525 (finalize_must_preallocate): Track polynomial argument sizes.
1526 (compute_argument_addresses): Likewise.
1527 (internal_arg_pointer_based_exp): Track polynomial offsets.
1528 (mem_overlaps_already_clobbered_arg_p): Rename to...
1529 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
1530 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1531 Check stored_args_used_watermark.
1532 (load_register_parameters): Update accordingly.
1533 (check_sibcall_argument_overlap_1): Likewise.
1534 (combine_pending_stack_adjustment_and_call): Take the unadjusted
1535 args size as a poly_int64 rather than an int. Return a bool
1536 indicating whether the optimization was possible and return
1537 the new adjustment by reference.
1538 (check_sibcall_argument_overlap): Track polynomail argument sizes.
1539 Update stored_args_watermark.
1540 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
1541 (expand_call): Likewise. Maintain stack_usage_watermark and
1542 stored_args_watermark. Update calls to
1543 combine_pending_stack_adjustment_and_call.
1544 (emit_library_call_value_1): Handle polynomial argument sizes.
1545 Call stack_region_maybe_used_p and mark_stack_region_used.
1546 Maintain stack_usage_watermark.
1547 (store_one_arg): Likewise. Update call to
1548 mem_overlaps_already_clobbered_arg_p.
1549 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
1551 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
1552 * config/microblaze/microblaze.c (microblaze_function_prologue):
1554 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
1555 TARGET_RETURN_POPS_ARGS interface.
1556 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
1558 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
1559 * config/i386/i386.c (ix86_return_pops_args): Update for new
1560 TARGET_RETURN_POPS_ARGS interface.
1561 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
1562 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
1563 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
1564 TARGET_RETURN_POPS_ARGS interface.
1565 * config/vax/vax.c (vax_return_pops_args): Likewise.
1566 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
1567 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
1568 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
1569 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1570 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1571 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
1572 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1573 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1574 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
1575 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
1576 and x_stack_pointer_delta from int to poly_int64.
1577 * dojump.c (do_pending_stack_adjust): Update accordingly.
1578 * explow.c (allocate_dynamic_stack_space): Handle polynomial
1579 stack_pointer_deltas.
1580 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
1581 (pad_to_arg_alignment): Track polynomial offsets.
1582 (assign_parm_find_stack_rtl): Likewise.
1583 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
1584 * toplev.c (output_stack_usage): Update reference to
1585 current_function_pushed_stack_size.
1587 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1588 Alan Hayward <alan.hayward@arm.com>
1589 David Sherwood <david.sherwood@arm.com>
1591 * function.c (in_arg_offset, var_offset, dynamic_offset)
1592 (out_arg_offset, cfa_offset): Change from int to poly_int64.
1593 (instantiate_new_reg): Return the new offset as a poly_int64_pod
1594 rather than a HOST_WIDE_INT.
1595 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
1596 (instantiate_virtual_regs_in_insn): Likewise.
1598 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1599 Alan Hayward <alan.hayward@arm.com>
1600 David Sherwood <david.sherwood@arm.com>
1602 * rtl.h (get_args_size, add_args_size_note): New functions.
1603 (find_args_size_adjust): Return a poly_int64 rather than a
1605 (fixup_args_size_notes): Likewise. Make the same change to the
1606 end_args_size parameter.
1607 * rtlanal.c (get_args_size, add_args_size_note): New functions.
1608 * builtins.c (expand_builtin_trap): Use add_args_size_note.
1609 * calls.c (emit_call_1): Likewise.
1610 * explow.c (adjust_stack_1): Likewise.
1611 * cfgcleanup.c (old_insns_match_p): Update use of
1612 find_args_size_adjust.
1613 * combine.c (distribute_notes): Track polynomial arg sizes.
1614 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
1615 end_true_args_size, beg_delay_args_size and end_delay_args_size
1616 from HOST_WIDE_INT to poly_int64.
1617 (add_cfi_args_size): Take the args_size as a poly_int64 rather
1618 than a HOST_WIDE_INT.
1619 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
1620 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
1621 polynomial arg sizes.
1622 * emit-rtl.c (try_split): Use get_args_size.
1623 * recog.c (peep2_attempt): Likewise.
1624 * reload1.c (reload_as_needed): Likewise.
1625 * expr.c (find_args_size_adjust): Return the adjustment as a
1626 poly_int64 rather than a HOST_WIDE_INT.
1627 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
1628 to a poly_int64 and change the return type in the same way.
1629 (emit_single_push_insn): Track polynomial arg sizes.
1631 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1632 Alan Hayward <alan.hayward@arm.com>
1633 David Sherwood <david.sherwood@arm.com>
1635 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
1636 from HOST_WIDE_INT to poly_int64.
1637 * expr.c (push_block, emit_push_insn): Likewise.
1639 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1640 Alan Hayward <alan.hayward@arm.com>
1641 David Sherwood <david.sherwood@arm.com>
1643 * function.h (frame_space): Change start and length from HOST_WIDE_INT
1645 (get_frame_size): Return the size as a poly_int64 rather than a
1647 (frame_offset_overflow): Take the offset as a poly_int64 rather
1648 than a HOST_WIDE_INT.
1649 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1650 (assign_stack_temp): Likewise for the size.
1651 * function.c (get_frame_size): Return a poly_int64 rather than
1653 (frame_offset_overflow): Take the offset as a poly_int64 rather
1654 than a HOST_WIDE_INT.
1655 (try_fit_stack_local): Take the start, length and size as poly_int64s
1656 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
1657 rather than a HOST_WIDE_INT.
1658 (add_frame_space): Take the start and end as poly_int64s rather than
1660 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1661 (assign_stack_temp): Likewise for the size.
1662 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
1664 (find_temp_slot_from_address): Handle polynomial offsets.
1665 (combine_temp_slots): Likewise.
1666 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
1668 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
1669 poly_int64 rather than a HOST_WIDE_INT.
1670 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
1671 than a HOST_WIDE_INT.
1672 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
1673 polynomial frame offsets.
1674 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
1675 as a poly_int64 rather than an int.
1676 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
1677 * config/v850/v850-protos.h (compute_frame_size): Likewise.
1678 * config/v850/v850.c (compute_frame_size): Likewise.
1679 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
1680 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
1681 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
1682 * config/pa/pa.c (pa_compute_frame_size): Likewise.
1683 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
1684 rather than a HOST_WIDE_INT.
1685 * explow.c (get_dynamic_stack_base): Likewise.
1686 * final.c (final_start_function): Use the constant lower bound
1687 of the frame size for -Wframe-larger-than.
1688 * ira.c (do_reload): Adjust for new get_frame_size return type.
1689 * lra.c (lra): Likewise.
1690 * reload1.c (reload): Likewise.
1691 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
1692 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
1693 * rtlanal.c (get_initial_register_offset): Return the offset as
1694 a poly_int64 rather than a HOST_WIDE_INT.
1696 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1697 Alan Hayward <alan.hayward@arm.com>
1698 David Sherwood <david.sherwood@arm.com>
1700 * reload1.c (elim_table): Change initial_offset, offset and
1701 previous_offset from HOST_WIDE_INT to poly_int64_pod.
1702 (offsets_at): Change the target array's element type from
1703 HOST_WIDE_INT to poly_int64_pod.
1704 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
1705 (elimination_costs_in_insn, update_eliminable_offsets)
1706 (verify_initial_elim_offsets, set_offsets_for_label)
1707 (init_eliminable_invariants): Update after above changes.
1709 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1710 Alan Hayward <alan.hayward@arm.com>
1711 David Sherwood <david.sherwood@arm.com>
1713 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
1714 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
1715 (decomposition): Change start and end from HOST_WIDE_INT
1717 (decompose, immune_p): Update accordingly.
1718 (find_inc_amount): Return a poly_int64 rather than an int.
1719 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
1722 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1723 Alan Hayward <alan.hayward@arm.com>
1724 David Sherwood <david.sherwood@arm.com>
1726 * tree.h (get_inner_reference): Return the bitsize and bitpos
1727 as poly_int64_pods rather than HOST_WIDE_INT.
1728 * fold-const.h (ptr_difference_const): Return the pointer difference
1729 as a poly_int64_pod rather than a HOST_WIDE_INT.
1730 * expr.c (get_inner_reference): Return the bitsize and bitpos
1731 as poly_int64_pods rather than HOST_WIDE_INT.
1732 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
1734 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
1735 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
1736 (optimize_bit_field_compare): Update call to get_inner_reference.
1737 (decode_field_reference): Likewise.
1738 (fold_unary_loc): Track polynomial offsets and sizes.
1739 (split_address_to_core_and_offset): Return the bitpos as a
1740 poly_int64_pod rather than a HOST_WIDE_INT.
1741 (ptr_difference_const): Likewise for the pointer difference.
1742 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
1743 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
1744 * dbxout.c (dbxout_expand_expr): Likewise.
1745 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
1746 (loc_list_from_tree_1, fortran_common): Likewise.
1747 * gimple-laddress.c (pass_laddress::execute): Likewise.
1748 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
1749 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
1750 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1751 * tree-affine.c (tree_to_aff_combination): Likewise.
1752 (get_inner_reference_aff): Likewise.
1753 * tree-data-ref.c (split_constant_offset_1): Likewise.
1754 (dr_analyze_innermost): Likewise.
1755 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
1756 * tree-sra.c (ipa_sra_check_caller): Likewise.
1757 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1758 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
1759 (instrument_bool_enum_load, instrument_object_size): Likewise.
1760 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
1761 to get_inner_reference.
1762 * hsa-gen.c (gen_hsa_addr): Likewise.
1763 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
1764 * tsan.c (instrument_expr): Likewise.
1765 * match.pd: Update call to ptr_difference_const.
1767 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1768 Alan Hayward <alan.hayward@arm.com>
1769 David Sherwood <david.sherwood@arm.com>
1771 * fold-const.c (fold_comparison): Track sizes and offsets as
1772 poly_int64s rather than HOST_WIDE_INTs when folding address
1775 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1776 Alan Hayward <alan.hayward@arm.com>
1777 David Sherwood <david.sherwood@arm.com>
1779 * expr.h (get_bit_range): Return the bitstart and bitend as
1780 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1781 as a poly_int64 rather than a HOST_WIDE_INT.
1782 * expr.c (get_bit_range): Likewise.
1783 (expand_assignment): Update call accordingly.
1784 * fold-const.c (optimize_bit_field_compare): Likewise.
1786 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1787 Alan Hayward <alan.hayward@arm.com>
1788 David Sherwood <david.sherwood@arm.com>
1790 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1791 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1792 as an optional final parameter.
1793 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1794 (aarch64_legitimate_address_p): Likewise.
1795 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1796 instead of an rtx code.
1797 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1798 (aarch64_legitimate_address_hook_p): Likewise.
1799 (aarch64_print_ldpstp_address): Likewise.
1800 (aarch64_print_operand_address): Likewise.
1801 (aarch64_address_cost): Likewise.
1802 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1803 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1804 (aarch64_mem_pair_lanes_operand): Likewise.
1806 2017-12-20 Richard Biener <rguenther@suse.de>
1808 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1809 update_stmt_if_modified.
1811 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1813 PR tree-optimization/83491
1814 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1815 before walking uses. Improve coding style and comments.
1817 2017-12-20 Tom de Vries <tom@codesourcery.com>
1819 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1821 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1824 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1826 (pass_ipa_free_fn_summary): Add small_p private data member,
1827 initialize to false in the ctor.
1828 (pass_ipa_free_fn_summary::clone,
1829 pass_ipa_free_fn_summary::set_pass_param,
1830 pass_ipa_free_fn_summary::gate): New methods.
1831 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1832 | TODO_dump_symtab if small_p.
1833 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1834 entry and add another instance of the pass with false parm after
1836 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1837 ipa_free_fn_summary here.
1839 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1841 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1843 2017-12-20 Martin Sebor <msebor@redhat.com>
1846 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1847 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1848 (expand_builtin_strncmp): Same.
1850 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1853 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1854 allowed debug stmts before labels.
1855 (expand_gimple_basic_block): Likewise.
1856 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1857 * gimple-iterator.h (gsi_after_labels): Likewise.
1858 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1859 rename reused variable, and simplify using gsi_move_before.
1860 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1861 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1862 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1863 (gimple_verify_flow_info, gimple_block_label): Likewise.
1864 (make_blocks): Move debug markers after adjacent labels.
1865 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1866 allowed debug insns outside blocks.
1867 * df-scan.c (df_insn_delete): Likewise.
1868 * lra-constraints.c (update_ebb_live_info): Likewise.
1869 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1870 (vt_initialize, delete_vta_debug_insns): Likewise.
1871 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1873 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1874 Alan Hayward <alan.hayward@arm.com>
1875 David Sherwood <david.sherwood@arm.com>
1877 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1878 (round_up_to_byte_boundary): Likewise.
1879 * expr.h (get_bit_range): Add temporary shim.
1880 * gimple-ssa-store-merging.c (store_operand_info): Change the
1881 bitsize, bitpos, bitregion_start and bitregion_end fields from
1882 unsigned HOST_WIDE_INT to poly_uint64.
1883 (merged_store_group): Likewise load_align_base.
1884 (compatible_load_p, compatible_load_p): Update accordingly.
1885 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1886 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1887 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1888 bitregion_start and bitregion_end as poly_uint64s rather than
1889 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1890 (handled_load): Take the bitsize, bitpos,
1891 bitregion_start and bitregion_end as poly_uint64s rather than
1892 unsigned HOST_WIDE_INTs.
1893 (pass_store_merging::process_store): Update call to
1894 mem_valid_for_store_merging.
1896 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1897 Alan Hayward <alan.hayward@arm.com>
1898 David Sherwood <david.sherwood@arm.com>
1900 * builtins.c (get_object_alignment_2): Track polynomial offsets
1901 and sizes. Update the alignment handling.
1903 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1904 Alan Hayward <alan.hayward@arm.com>
1905 David Sherwood <david.sherwood@arm.com>
1907 * tree.h (get_inner_reference): Add a version that returns the
1908 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1909 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1910 the case in which bitpos is not associated with the first byte.
1912 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1913 Alan Hayward <alan.hayward@arm.com>
1914 David Sherwood <david.sherwood@arm.com>
1916 * tree-affine.h (get_inner_reference_aff): Return the size as a
1918 * tree-affine.c (get_inner_reference_aff): Likewise.
1919 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1920 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1922 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1923 Alan Hayward <alan.hayward@arm.com>
1924 David Sherwood <david.sherwood@arm.com>
1926 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1927 HOST_WIDE_INT rather than a poly_int64.
1929 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1930 Alan Hayward <alan.hayward@arm.com>
1931 David Sherwood <david.sherwood@arm.com>
1933 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1934 HOST_WIDE_INT to poly_int64_pod.
1935 (perform_symbolic_merge): Update accordingly.
1936 (bswap_replace): Likewise.
1938 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1939 Alan Hayward <alan.hayward@arm.com>
1940 David Sherwood <david.sherwood@arm.com>
1942 * tree-affine.h (aff_tree::offset): Change from widest_int
1944 (wide_int_ext_for_comb): Delete.
1945 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1946 constants as poly_widest_int rather than widest_int.
1947 (aff_combination_constant_multiple_p): Return the multiplier
1948 as a poly_widest_int.
1949 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1951 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1952 version static and add an overload for poly_widest_int.
1953 (aff_combination_const, aff_combination_add_cst)
1954 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1955 the constants as poly_widest_int rather than widest_int.
1956 (tree_to_aff_combination): Generalize INTEGER_CST case to
1958 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1959 (aff_combination_add_product, aff_combination_mult): Handle
1961 (aff_combination_constant_multiple_p): Return the multiplier
1962 as a poly_widest_int.
1963 * tree-predcom.c (determine_offset): Return the offset as a
1965 (split_data_refs_to_components, suitable_component_p): Update
1967 (valid_initializer_p): Update call to
1968 aff_combination_constant_multiple_p.
1969 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1970 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1971 as a poly_int64 rather than a HOST_WIDE_INT.
1972 (get_address_cost): Handle polynomial offsets.
1973 (iv_elimination_compare_lt): Likewise.
1974 (rewrite_use_nonlinear_expr): Likewise.
1976 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1977 Alan Hayward <alan.hayward@arm.com>
1978 David Sherwood <david.sherwood@arm.com>
1980 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1981 as a poly_int64_pod rather than a HOST_WIDE_INT.
1982 (get_addr_base_and_unit_offset): Likewise.
1983 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1984 (get_addr_base_and_unit_offset): Likewise.
1985 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1986 to poly_int64 in example.
1987 * fold-const.c (fold_binary_loc): Update call to
1988 get_addr_base_and_unit_offset.
1989 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1990 (maybe_canonicalize_mem_ref_addr): Likewise.
1991 (gimple_fold_stmt_to_constant_1): Likewise.
1992 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1994 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1995 * match.pd: Likewise.
1996 * omp-low.c (lower_omp_target): Likewise.
1997 * tree-sra.c (build_ref_for_offset): Likewise.
1998 (build_debug_ref_for_model): Likewise.
1999 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
2000 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
2001 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
2002 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
2003 (constant_pointer_difference): Likewise.
2004 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
2005 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
2006 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
2007 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
2008 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
2009 (set_ssa_val_to): Likewise.
2010 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
2011 (maybe_diag_stxncpy_trunc): Likewise.
2012 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
2013 * tree.c (build_simple_mem_ref_loc): Likewise.
2014 (array_at_struct_end_p): Likewise.
2016 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2017 Alan Hayward <alan.hayward@arm.com>
2018 David Sherwood <david.sherwood@arm.com>
2020 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
2021 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
2022 (get_ref_base_and_extent_hwi): Declare.
2023 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
2024 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
2025 (get_ref_base_and_extent_hwi): New function.
2026 * cfgexpand.c (expand_debug_expr): Update call to
2027 get_ref_base_and_extent.
2028 * dwarf2out.c (add_var_loc_to_decl): Likewise.
2029 * gimple-fold.c (get_base_constructor): Return the offset as a
2030 poly_int64_pod rather than a HOST_WIDE_INT.
2031 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
2032 * ipa-polymorphic-call.c
2033 (ipa_polymorphic_call_context::set_by_invariant)
2034 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
2035 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
2036 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
2037 rather than get_ref_base_and_extent.
2038 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
2039 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
2040 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
2042 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
2043 call to get_ref_base_and_extent.
2044 * tree-sra.c (create_access, get_access_for_expr): Likewise.
2045 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
2046 (stmt_kills_ref_p): Likewise.
2047 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2048 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
2050 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
2051 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
2052 when calling native_encode_expr.
2053 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
2054 call to get_ref_base_and_extent.
2055 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
2056 get_ref_base_and_extent.
2057 * var-tracking.c (track_expr_p): Likewise.
2059 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2060 Alan Hayward <alan.hayward@arm.com>
2061 David Sherwood <david.sherwood@arm.com>
2063 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
2064 HOST_WIDE_INT to poly_int64_pod.
2065 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
2066 polynomail parameter offsets.
2068 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2069 Alan Hayward <alan.hayward@arm.com>
2070 David Sherwood <david.sherwood@arm.com>
2072 * gengtype.c (main): Handle poly_int64_pod.
2073 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
2074 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
2075 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
2076 from HOST_WIDE_INT to poly_int64_pod.
2077 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
2078 (copy_cfa): New function.
2079 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
2080 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
2081 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
2082 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
2083 (initial_return_save): Treat offsets as poly_ints.
2084 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
2085 offset is nonconstant.
2086 (reg_save): Take the offset as a poly_int64. Fall back to
2087 DW_CFA_expression for nonconstant offsets.
2088 (queue_reg_save): Take the offset as a poly_int64.
2089 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
2091 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2092 Alan Hayward <alan.hayward@arm.com>
2093 David Sherwood <david.sherwood@arm.com>
2095 * rtl.h (operand_subword, operand_subword_force): Take the offset
2096 as a poly_uint64 an unsigned int.
2097 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
2099 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2100 Alan Hayward <alan.hayward@arm.com>
2101 David Sherwood <david.sherwood@arm.com>
2103 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
2104 'p' format code. Use INT_LIST rather than SUBREG as the example of
2105 a code with an XINT and an XEXP. Remove the implication that
2106 accessing an rtx field using XINT is expected to work.
2107 * rtl.def (SUBREG): Change format from "ei" to "ep".
2108 * rtl.h (rtunion::rt_subreg): New field.
2109 (XCSUBREG): New macro.
2110 (SUBREG_BYTE): Use it.
2111 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
2112 Update constructor accordingly.
2113 (subreg_shape::operator ==): Update accordingly.
2114 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
2115 than an unsigned int.
2116 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
2117 a poly_uint64 rather than an unsigned int.
2118 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
2119 than an unsigned int.
2120 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
2121 (subreg_size_highpart_offset): Return a poly_uint64 rather than
2122 an unsigned int. Take the sizes as poly_uint64s.
2123 (subreg_offset_from_lsb): Return a poly_uint64 rather than
2124 an unsigned int. Take the shift as a poly_uint64 rather than
2126 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
2127 as a poly_uint64 rather than an unsigned int.
2128 (simplify_subreg_regno): Likewise.
2129 (byte_lowpart_offset): Return the memory offset as a poly_int64
2131 (subreg_memory_offset): Likewise. Take the subreg offset as a
2132 poly_uint64 rather than an unsigned int.
2133 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
2134 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
2135 poly_uint64 rather than an unsigned int.
2136 * rtl.c (rtx_format): Describe 'p' in comment.
2137 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
2138 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
2139 offset as a poly_uint64 rather than an unsigned int.
2140 (byte_lowpart_offset): Return the memory offset as a poly_int64
2142 (subreg_memory_offset): Likewise. Take the subreg offset as a
2143 poly_uint64 rather than an unsigned int.
2144 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
2145 mode sizes as poly_uint64s rather than unsigned ints. Return a
2146 poly_uint64 rather than an unsigned int.
2147 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
2148 (copy_insn_1): Handle 'p'.
2149 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
2150 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
2151 an unsigned int. Return the shift in the same way.
2152 (subreg_lsb): Return the shift as a poly_uint64 rather than an
2154 (subreg_size_offset_from_lsb): Take the sizes and shift as
2155 poly_uint64s rather than unsigned ints. Return the offset as
2157 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
2158 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
2160 * rtlhash.c (add_rtx): Handle 'p'.
2161 * genemit.c (gen_exp): Likewise.
2162 * gengenrtl.c (type_from_format, gendef): Likewise.
2163 * gensupport.c (subst_pattern_match, get_alternatives_number)
2164 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
2165 (subst_dup): Likewise.
2166 * gengtype.c (adjust_field_rtx_def): Likewise.
2167 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
2168 (match_pattern_2): Likewise.
2169 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
2170 (rtx_test::subreg_field): New function.
2171 (operator ==, safe_to_hoist_p, transition_parameter_type)
2172 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
2173 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
2174 * genpeep.c (match_rtx): Likewise.
2175 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
2176 (rtx_writer::print_rtx_operand): Handle 'p'.
2177 (print_value): Handle SUBREG.
2178 * read-rtl.c (apply_int_iterator): Likewise.
2179 (rtx_reader::read_rtx_operand): Handle 'p'.
2180 * alias.c (rtx_equal_for_memref_p): Likewise.
2181 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
2182 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
2184 * calls.c (expand_call): Likewise.
2185 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
2186 (make_extraction, gen_lowpart_for_combine): Likewise.
2187 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
2189 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
2190 rather than an unsigned int. Treat subreg offsets as poly_ints.
2191 (exp_equiv_p): Handle 'p'.
2192 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
2193 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
2194 * dse.c (find_shift_sequence): Likewise.
2195 * dwarf2out.c (rtl_for_decl_location): Likewise.
2196 * expmed.c (extract_low_bits): Likewise.
2197 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
2198 (expand_expr_real_2): Likewise.
2199 * final.c (alter_subreg): Likewise.
2200 (leaf_renumber_regs_insn): Handle 'p'.
2201 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
2202 Treat subreg offsets as poly_ints.
2203 * fwprop.c (forward_propagate_and_simplify): Likewise.
2204 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
2205 * ira.c (get_subreg_tracking_sizes): Likewise.
2206 * ira-conflicts.c (go_through_subreg): Likewise.
2207 * ira-lives.c (process_single_reg_class_operands): Likewise.
2208 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
2209 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
2210 as a poly_uint64 rather than an unsigned int.
2211 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
2212 subreg offsets as poly_ints.
2213 * lra-constraints.c (operands_match_p): Handle 'p'.
2214 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
2215 * lra-spills.c (assign_mem_slot): Likewise.
2216 * postreload.c (move2add_valid_value_p): Likewise.
2217 * recog.c (general_operand, indirect_operand): Likewise.
2218 * regcprop.c (copy_value, maybe_mode_change): Likewise.
2219 (copyprop_hardreg_forward_1): Likewise.
2220 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
2221 (record_subregs_of_mode): Likewise.
2222 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
2223 * reload.c (operands_match_p): Handle 'p'.
2224 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
2225 * reload1.c (alter_reg, choose_reload_regs): Likewise.
2226 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
2227 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
2228 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
2229 (simplify_const_poly_int_tests<N>::run): Likewise.
2230 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
2231 a poly_uint64 rather than an unsigned int.
2232 * valtrack.c (debug_lowpart_subreg): Likewise.
2233 * var-tracking.c (var_lowpart): Likewise.
2234 (loc_cmp): Handle 'p'.
2236 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2237 Alan Hayward <alan.hayward@arm.com>
2238 David Sherwood <david.sherwood@arm.com>
2240 * ira.c (get_subreg_tracking_sizes): New function.
2241 (init_live_subregs): Take an integer size rather than a register.
2242 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
2243 to init_live_subregs.
2245 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2246 Alan Hayward <alan.hayward@arm.com>
2247 David Sherwood <david.sherwood@arm.com>
2249 * expr.c (store_constructor_field): Change bitsize from a
2250 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
2251 HOST_WIDE_INT to a poly_int64.
2252 (store_constructor): Change size from a HOST_WIDE_INT to
2254 (store_field): Likewise bitsize and bitpos.
2256 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2257 Alan Hayward <alan.hayward@arm.com>
2258 David Sherwood <david.sherwood@arm.com>
2260 * expmed.h (store_bit_field): Change bitregion_start and
2261 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
2262 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
2263 (store_bit_field_1, store_integral_bit_field, store_bit_field)
2264 (store_fixed_bit_field, store_split_bit_field): Likewise.
2265 * expr.c (store_constructor_field, store_field): Likewise.
2266 (optimize_bitfield_assignment_op): Likewise. Make the same change
2267 to bitsize and bitpos.
2268 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
2269 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
2270 same change in the constructor arguments.
2271 (get_best_mode): Change bitregion_start and bitregion_end from
2272 unsigned HOST_WIDE_INT to poly_uint64.
2273 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2274 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
2276 (bit_field_mode_iterator::next_mode): Update for new types
2277 of m_bitregion_start and m_bitregion_end.
2278 (get_best_mode): Change bitregion_start and bitregion_end from
2279 unsigned HOST_WIDE_INT to poly_uint64.
2281 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2282 Alan Hayward <alan.hayward@arm.com>
2283 David Sherwood <david.sherwood@arm.com>
2285 * rtl.h (simplify_gen_subreg): Add a temporary overload that
2286 accepts poly_uint64 offsets.
2287 * expmed.h (extract_bit_field): Take bitsize and bitnum as
2288 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2289 * expmed.c (lowpart_bit_field_p): Likewise.
2290 (extract_bit_field_as_subreg): New function, split out from...
2291 (extract_bit_field_1): ...here. Take bitsize and bitnum as
2292 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
2293 extractions, check that BITSIZE matches the size of the extracted
2294 value and that BITNUM is an exact multiple of that size.
2295 If all else fails, try forcing the value into memory if
2296 BITNUM is variable, and adjusting the address so that the
2297 offset is constant. Split the part that can only handle constant
2298 bitsize and bitnum out into...
2299 (extract_integral_bit_field): ...this new function.
2300 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
2301 rather than unsigned HOST_WIDE_INTs.
2303 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2304 Alan Hayward <alan.hayward@arm.com>
2305 David Sherwood <david.sherwood@arm.com>
2307 * expmed.h (store_bit_field): Take bitsize and bitnum as
2308 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2309 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
2310 that returns the byte size.
2311 (store_bit_field_1): Take bitsize and bitnum as
2312 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
2313 to simple_mem_bitfield_p. Split the part that can only handle
2314 constant bitsize and bitnum out into...
2315 (store_integral_bit_field): ...this new function.
2316 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
2317 than unsigned HOST_WIDE_INTs.
2318 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
2320 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2321 Alan Hayward <alan.hayward@arm.com>
2322 David Sherwood <david.sherwood@arm.com>
2324 * lra-int.h (lra_reg): Change offset from int to poly_int64.
2325 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
2327 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2328 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
2329 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
2330 offset as a poly_int64 rather than an int.
2331 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
2332 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2333 * lra-constraints.c (equiv_address_substitution): Track offsets
2335 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
2336 (curr_insn_transform): Handle the new form of sp_offset.
2337 * lra-eliminations.c (lra_elim_table): Change previous_offset
2338 and offset from HOST_WIDE_INT to poly_int64.
2339 (print_elim_table, update_reg_eliminate): Update accordingly.
2340 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
2341 (get_elimination): Update accordingly.
2342 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
2343 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2344 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
2345 poly_int64 offsets generally.
2346 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
2347 (mark_not_eliminable, init_elimination): Update accordingly.
2348 (remove_reg_equal_offset_note): Return a bool and pass the new
2349 offset back by pointer as a poly_int64.
2350 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
2351 rather than a HOST_WIDE_INT.
2352 (do_remat): Track offsets poly_int64s.
2353 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
2355 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2356 Alan Hayward <alan.hayward@arm.com>
2357 David Sherwood <david.sherwood@arm.com>
2359 * rtl.h (mem_attrs): Add a default constructor. Change size and
2360 offset from HOST_WIDE_INT to poly_int64.
2361 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
2362 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
2363 (widen_memory_access): Take the sizes and offsets as poly_int64s
2364 rather than HOST_WIDE_INTs.
2365 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
2366 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
2368 (adjust_offset_for_component_ref): Change the offset from a
2369 HOST_WIDE_INT to a poly_int64.
2370 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
2371 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
2372 * dce.c (find_call_stack_args): Likewise.
2373 * dse.c (record_store): Likewise.
2374 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
2375 * print-rtl.c (rtx_writer::print_rtx): Likewise.
2376 * read-rtl-function.c (test_loading_mem): Likewise.
2377 * rtlanal.c (may_trap_p_1): Likewise.
2378 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
2379 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
2380 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
2381 (mem_attrs::mem_attrs): New function.
2382 (set_mem_attributes_minus_bitpos): Change bitpos from a
2383 HOST_WIDE_INT to poly_int64.
2384 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
2385 (clear_mem_offset, clear_mem_size, change_address)
2386 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
2387 initialize mem_attrs.
2388 (set_mem_offset, set_mem_size, adjust_address_1)
2389 (adjust_automodify_address_1, offset_address, widen_memory_access):
2390 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
2392 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2393 Alan Hayward <alan.hayward@arm.com>
2394 David Sherwood <david.sherwood@arm.com>
2396 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
2397 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
2398 rather than 0 to represent an unknown size. Assert that the size
2399 is known when the mode isn't BLKmode.
2400 (may_trap_p_1): Use -1 for unknown sizes.
2401 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
2403 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2404 Alan Hayward <alan.hayward@arm.com>
2405 David Sherwood <david.sherwood@arm.com>
2407 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
2408 to poly_int64. Update commentary for positions_needed.large.
2409 (read_info_type): Change offset and width from HOST_WIDE_INT
2411 (set_usage_bits): Likewise.
2412 (canon_address): Return the offset as a poly_int64 rather than
2413 a HOST_WIDE_INT. Use strip_offset_and_add.
2414 (set_all_positions_unneeded, any_positions_needed_p): Use
2415 positions_needed.large to track stores with non-constant widths.
2416 (all_positions_needed_p): Likewise. Take the offset and width
2417 as poly_int64s rather than ints. Assert that rhs is nonnull.
2418 (record_store): Cope with non-constant offsets and widths.
2419 Nullify the rhs of an earlier store if we can't tell which bytes
2421 (find_shift_sequence): Take the access_size and shift as poly_int64s
2423 (get_stored_val): Take the read_offset and read_width as poly_int64s
2424 rather than HOST_WIDE_INTs.
2425 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
2426 non-constant offsets and widths.
2428 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2429 Alan Hayward <alan.hayward@arm.com>
2430 David Sherwood <david.sherwood@arm.com>
2432 * inchash.h (inchash::hash::add_poly_int): New function.
2433 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
2434 Use poly_int64 rather than HOST_WIDE_INT.
2435 (ao_ref::max_size_known_p): New function.
2436 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
2437 rather than HOST_WIDE_INT.
2438 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
2439 to temporaries until its interface is adjusted to match.
2440 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
2441 (aliasing_component_refs_p, decl_refs_may_alias_p)
2442 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
2443 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
2444 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
2446 * alias.c (ao_ref_from_mem): Likewise.
2447 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2448 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
2449 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
2450 (maybe_trim_complex_store, maybe_trim_constructor_store)
2451 (live_bytes_read, dse_classify_store): Likewise.
2452 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
2453 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
2454 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
2455 (vn_reference_lookup_3): Likewise.
2456 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
2458 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2459 Alan Hayward <alan.hayward@arm.com>
2460 David Sherwood <david.sherwood@arm.com>
2462 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
2463 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
2464 instead of ranges_overlap_p.
2466 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2467 Alan Hayward <alan.hayward@arm.com>
2468 David Sherwood <david.sherwood@arm.com>
2470 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
2471 sizes as poly_int64s rather than HOST_WIDE_INTs.
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 * gimple-fold.h (fold_ctor_reference): Take the offset and size
2478 as poly_uint64 rather than unsigned HOST_WIDE_INT.
2479 * gimple-fold.c (fold_ctor_reference): Likewise.
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 * target.def (dwarf_poly_indeterminate_value): New hook.
2486 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
2487 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
2488 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
2489 * doc/tm.texi: Regenerate.
2490 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
2491 offset as a poly_int64.
2492 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
2493 offset as a poly_int64.
2494 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
2495 Take the offset as a poly_int64.
2496 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
2497 (frame_pointer_fb_offset): Change to a poly_int64.
2498 (int_loc_descriptor): Take the offset as a poly_int64. Use
2499 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
2500 (based_loc_descr): Take the offset as a poly_int64.
2501 Use strip_offset_and_add to handle (plus X (const)).
2502 Use new_reg_loc_descr instead of an open-coded version of the
2503 previous implementation.
2504 (mem_loc_descriptor): Handle CONST_POLY_INT.
2505 (compute_frame_pointer_to_fb_displacement): Take the offset as a
2506 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
2508 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2509 Alan Hayward <alan.hayward@arm.com>
2510 David Sherwood <david.sherwood@arm.com>
2512 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
2513 (gen_rtx_REG_offset): Take the offset as a poly_int64.
2514 * inchash.h (inchash::hash::add_poly_hwi): New function.
2515 * gengtype.c (main): Register poly_int64.
2516 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
2517 offset as a poly_int.
2518 (reg_attr_hasher::equal): Use must_eq to compare offsets.
2519 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
2520 offset as a poly_int64.
2521 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
2522 * print-rtl.c (print_poly_int): New function.
2523 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
2525 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
2527 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
2528 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
2529 (vt_get_decl_and_offset): Return the offset as a poly_int64.
2530 Enforce track_offset_p for parts of a PARALLEL.
2531 (vt_add_function_parameter): Use const_offset for the final
2532 offset to track. Use get_tracked_reg_offset for the parts
2535 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2536 Alan Hayward <alan.hayward@arm.com>
2537 David Sherwood <david.sherwood@arm.com>
2539 * target.def (truly_noop_truncation): Take poly_uint64s instead of
2540 unsigned ints. Change default to hook_bool_puint64_puint64_true.
2541 * doc/tm.texi: Regenerate.
2542 * hooks.h (hook_bool_uint_uint_true): Delete.
2543 (hook_bool_puint64_puint64_true): Declare.
2544 * hooks.c (hook_bool_uint_uint_true): Delete.
2545 (hook_bool_puint64_puint64_true): New function.
2546 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
2547 instead of unsigned ints.
2548 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
2549 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
2551 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2552 Alan Hayward <alan.hayward@arm.com>
2553 David Sherwood <david.sherwood@arm.com>
2555 * optabs.h (expand_operand): Add an int_value field.
2556 (create_expand_operand): Add an int_value parameter and use it
2557 to initialize the new expand_operand field.
2558 (create_integer_operand): Replace with a declaration of a function
2559 that accepts poly_int64s. Move the implementation to...
2560 * optabs.c (create_integer_operand): ...here.
2561 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
2562 the mode preserves the value of int_value, instead of calling
2563 const_int_operand on the rtx. Use gen_int_mode to generate
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 * dumpfile.h (dump_dec): Declare.
2571 * dumpfile.c (dump_dec): New function.
2572 * pretty-print.h (pp_wide_integer): Turn into a function and
2573 declare a poly_int version.
2574 * pretty-print.c (pp_wide_integer): New function for poly_ints.
2576 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2577 Alan Hayward <alan.hayward@arm.com>
2578 David Sherwood <david.sherwood@arm.com>
2580 * doc/generic.texi (POLY_INT_CST): Document.
2581 * tree.def (POLY_INT_CST): New tree code.
2582 * treestruct.def (TS_POLY_INT_CST): New tree layout.
2583 * tree-core.h (tree_poly_int_cst): New struct.
2584 (tree_node): Add a poly_int_cst field.
2585 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
2586 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
2587 instead of a wide_int_ref.
2588 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
2590 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2591 instead of an unsigned HOST_WIDE_INT.
2592 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
2593 (ptrdiff_tree_p): Declare.
2594 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
2595 extern inline implementations if the target doesn't use POLY_INT_CST.
2596 (poly_int_tree_p): New function.
2597 (wi::unextended_tree): New class.
2598 (wi::int_traits <unextended_tree>): New override.
2599 (wi::extended_tree): Add a default constructor.
2600 (wi::extended_tree::get_tree): New function.
2601 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
2602 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
2603 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
2604 (wi::tree_to_poly_wide_ref): New typedefs.
2605 (wi::ints_for): Provide overloads for extended_tree and
2607 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
2608 (wi::to_wide): New functions.
2609 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
2610 * tree.c (poly_int_cst_hasher): New struct.
2611 (poly_int_cst_hash_table): New variable.
2612 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
2613 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
2615 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
2616 (init_ttree): Initialize poly_int_cst_hash_table.
2617 (build_int_cst, build_int_cst_type, build_invariant_address): Take
2618 a poly_int64 instead of a HOST_WIDE_INT.
2619 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2620 instead of an unsigned HOST_WIDE_INT.
2621 (wide_int_to_tree): Rename to...
2622 (wide_int_to_tree_1): ...this.
2623 (build_new_poly_int_cst, build_poly_int_cst): New functions.
2624 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
2625 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
2626 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
2627 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
2628 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
2630 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
2631 (streamer_read_tree_body): Likewise.
2632 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
2633 (streamer_write_tree_body): Likewise.
2634 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
2635 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
2636 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
2637 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
2638 * gimple-expr.h (is_gimple_constant): Likewise.
2639 * gimplify.c (maybe_with_size_expr): Likewise.
2640 * print-tree.c (print_node): Likewise.
2641 * tree-data-ref.c (data_ref_compare_tree): Likewise.
2642 * tree-pretty-print.c (dump_generic_node): Likewise.
2643 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2644 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
2645 * tree-vrp.c (compare_values_warnv): Likewise.
2646 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
2647 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
2648 (force_expr_to_var_cost): Likewise.
2649 * tree-ssa-loop.c (for_each_index): Likewise.
2650 * fold-const.h (build_invariant_address, size_int_kind): Take a
2651 poly_int64 instead of a HOST_WIDE_INT.
2652 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
2653 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
2655 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
2656 (int_const_binop_2): New function, split out from...
2657 (int_const_binop_1): ...here. Handle POLY_INT_CST.
2658 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
2659 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
2660 * gimple-ssa-strength-reduction.c (slsr_process_add)
2661 (slsr_process_mul): Check for INTEGER_CSTs before using them
2663 * stor-layout.c (bits_from_bytes): New function.
2664 (bit_from_pos): Use it.
2665 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
2666 by BITS_PER_UNIT to get the TYPE_SIZE.
2667 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
2668 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
2670 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2671 Alan Hayward <alan.hayward@arm.com>
2672 David Sherwood <david.sherwood@arm.com>
2674 * doc/rtl.texi (const_poly_int): Document. Also document the
2675 rtl sharing behavior.
2676 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
2677 * rtl.h (const_poly_int_def): New struct.
2678 (rtx_def::u): Add a cpi field.
2679 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
2680 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
2681 (wi::rtx_to_poly_wide_ref): New typedef
2682 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
2683 (poly_int_rtx_p): New functions.
2684 (trunc_int_for_mode): Declare a poly_int64 version.
2685 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2686 (immed_wide_int_const): Take a poly_wide_int_ref rather than
2688 (strip_offset): Declare.
2689 (strip_offset_and_add): New function.
2690 * rtl.def (CONST_POLY_INT): New rtx code.
2691 * rtl.c (rtx_size): Handle CONST_POLY_INT.
2692 (shared_const_p): Use poly_int_rtx_p.
2693 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
2695 (gen_int_shift_amount): Likewise.
2696 * emit-rtl.c (const_poly_int_hasher): New class.
2697 (const_poly_int_htab): New variable.
2698 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
2699 (const_poly_int_hasher::hash): New function.
2700 (const_poly_int_hasher::equal): Likewise.
2701 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
2702 (immed_wide_int_const): Rename to...
2703 (immed_wide_int_const_1): ...this and make static.
2704 (immed_wide_int_const): New function, taking a poly_wide_int_ref
2705 instead of a wide_int_ref.
2706 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
2707 (gen_lowpart_common): Handle CONST_POLY_INT.
2708 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
2709 * cselib.c (cselib_hash_rtx): Likewise.
2710 * dwarf2out.c (const_ok_for_output_1): Likewise.
2711 * expr.c (convert_modes): Likewise.
2712 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
2713 * rtlhash.c (add_rtx): Likewise.
2714 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
2715 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2716 Handle existing CONST_POLY_INT rtxes.
2717 * expmed.h (expand_shift): Take a poly_int64 instead of a
2719 * expmed.c (expand_shift): Likewise.
2720 * rtlanal.c (strip_offset): New function.
2721 (commutative_operand_precedence): Give CONST_POLY_INT the same
2722 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
2724 * rtl-tests.c (const_poly_int_tests): New struct.
2725 (rtl_tests_c_tests): Use it.
2726 * simplify-rtx.c (simplify_const_unary_operation): Handle
2728 (simplify_const_binary_operation): Likewise.
2729 (simplify_binary_operation_1): Fold additions of symbolic constants
2730 and CONST_POLY_INTs.
2731 (simplify_subreg): Handle extensions and truncations of
2733 (simplify_const_poly_int_tests): New struct.
2734 (simplify_rtx_c_tests): Use it.
2735 * wide-int.h (storage_ref): Add default constructor.
2736 (wide_int_ref_storage): Likewise.
2737 (trailing_wide_ints): Use GTY((user)).
2738 (trailing_wide_ints::operator[]): Add a const version.
2739 (trailing_wide_ints::get_precision): New function.
2740 (trailing_wide_ints::extra_size): Likewise.
2742 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2743 Alan Hayward <alan.hayward@arm.com>
2744 David Sherwood <david.sherwood@arm.com>
2746 * emit-rtl.h (gen_int_shift_amount): Declare.
2747 * emit-rtl.c (gen_int_shift_amount): New function.
2748 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
2750 * calls.c (shift_return_value): Likewise.
2751 * cse.c (fold_rtx): Likewise.
2752 * dse.c (find_shift_sequence): Likewise.
2753 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
2754 (expand_shift, expand_smod_pow2): Likewise.
2755 * lower-subreg.c (shift_cost): Likewise.
2756 * optabs.c (expand_superword_shift, expand_doubleword_mult)
2757 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
2758 (expand_vec_perm_var): Likewise.
2759 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2760 (simplify_binary_operation_1): Likewise.
2761 * combine.c (try_combine, find_split_point, force_int_to_mode)
2762 (simplify_shift_const_1, simplify_shift_const): Likewise.
2763 (change_zero_ext): Likewise. Use simplify_gen_binary.
2765 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2767 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2769 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2771 * doc/invoke.texi (ARM Options): Document accepted extension options
2772 for -march=armv8.3-a.
2774 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2777 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2778 or --with-float=softfp, set the default CPU to arm10e.
2780 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2782 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2783 * config/visium/predicates.md (const_shift_operand): Likewise.
2784 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2785 (visium_legitimize_reload_address): Likewise.
2787 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2790 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2792 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2795 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2797 2017-12-20 Julia Koval <julia.koval@intel.com>
2799 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2800 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2801 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2802 * config.gcc: Include vpclmulqdqintrin.h.
2803 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2804 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2806 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2807 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2808 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2809 * config/i386/i386.c (isa2_opts): Add -mcx16.
2810 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2811 (ix86_option_override_internal): Move mcx16 to flags2.
2812 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2813 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2814 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2815 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2816 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2817 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2818 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2819 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2820 * doc/invoke.texi: Add -mvpclmulqdq.
2822 2017-12-20 Tom de Vries <tom@codesourcery.com>
2825 * config/i386/i386.c (ix86_static_chain): Move
2826 DECL_STATIC_CHAIN test ...
2827 * calls.c (rtx_for_static_chain): ... here. New function.
2828 * calls.h (rtx_for_static_chain): Declare.
2829 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2830 instead of targetm.calls.static_chain.
2831 * df-scan.c (df_get_entry_block_def_set): Same.
2833 2017-12-19 Tom de Vries <tom@codesourcery.com>
2835 PR tree-optimization/83493
2836 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2839 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2841 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2842 inform with hardcoded english plural handling.
2844 2017-12-18 Jeff Law <law@redhat.com>
2846 PR tree-optimization/83477
2847 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2848 a non-virtual PHI, always push a new range.
2850 2017-12-19 Martin Sebor <msebor@redhat.com>
2853 * builtins.c (compute_objsize): Handle non-constant offsets.
2855 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2857 PR tree-optimization/83444
2858 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2859 character load case, if get_stridx on MEM_REF's operand doesn't
2860 look usable, retry with get_addr_stridx.
2862 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2865 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2866 (variable_tracking_main_1): Keep markers even when VTA fails.
2869 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2870 even if there are markers before it.
2871 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2873 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2875 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2876 typical order conditions.
2877 * sel-sched.c (extract_new_fences_from): Likewise.
2878 * config/visium/constraints.md (J, K, L): Likewise.
2879 * config/visium/predicates.md (const_shift_operand): Likewise.
2880 * config/visium/visium.c (visium_legitimize_address,
2881 visium_legitimize_reload_address): Likewise.
2882 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2883 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2884 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2885 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2886 SET_DUMP_DETAIL): Likewise.
2887 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2888 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2889 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2890 avr_set_core_architecture, avr_set_current_function,
2891 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2892 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2893 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2894 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2895 avr_map_decompose, avr_fold_builtin): Likewise.
2896 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2897 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2898 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2899 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2900 * config/m32c/m32c.c (m32c_conditional_register_usage,
2901 m32c_address_cost): Likewise.
2902 * config/m32c/predicates.md (shiftcount_operand,
2903 longshiftcount_operand): Likewise.
2904 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2905 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2906 can_use_cdx_ldstw): Likewise.
2907 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2908 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2910 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2911 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2912 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2913 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2914 * config/vax/vax.c (adjacent_operands_p): Likewise.
2915 * config/ft32/constraints.md (L, b, KA): Likewise.
2916 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2918 * cfgexpand.c (expand_stack_alignment): Likewise.
2919 * gcse.c (insert_expr_in_table): Likewise.
2920 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2921 * cgraphunit.c (cgraph_node::expand): Likewise.
2922 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2923 * emit-rtl.c (add_insn): Likewise.
2924 * input.c (dump_location_info): Likewise.
2925 * passes.c (NEXT_PASS): Likewise.
2926 * read-rtl-function.c (parse_note_insn_name,
2927 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2929 * sched-rgn.c (sched_rgn_init): Likewise.
2930 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2931 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2932 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2934 * explow.c (eliminate_constant_term): Likewise.
2935 * final.c (leaf_renumber_regs_insn): Likewise.
2936 * cfgrtl.c (print_rtl_with_bb): Likewise.
2937 * genhooks.c (emit_init_macros): Likewise.
2938 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2939 * tree-data-ref.c (conflict_fn): Likewise.
2940 * selftest.c (assert_streq): Likewise.
2941 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2942 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2943 fold_binary_loc, multiple_of_p): Likewise.
2944 * reload.c (push_reload, find_equiv_reg): Likewise.
2945 * et-forest.c (et_nca, et_below): Likewise.
2946 * dbxout.c (dbxout_symbol_location): Likewise.
2947 * reorg.c (relax_delay_slots): Likewise.
2948 * dojump.c (do_compare_rtx_and_jump): Likewise.
2949 * gengtype-parse.c (type): Likewise.
2950 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2951 simplify_const_relational_operation): Likewise.
2952 * reload1.c (do_output_reload): Likewise.
2953 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2954 * gengtype.c (type_for_name): Likewise.
2955 * gimple-ssa-sprintf.c (format_directive): Likewise.
2957 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2960 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2961 accessing it. Adjust comment.
2963 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2966 * predict.c (zero_one_minusone): New function.
2967 (apply_return_prediction): Avoid return prediction for functions
2968 returning only -1, 0 and 1 values, unless they only return -1 and 0
2971 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2973 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2974 fall-through warning.
2975 (arc_compute_frame_size): Remove unused variables.
2976 (arc_print_operand): Fix fprintif format.
2977 (arc_can_follow_jump): Clean fall-through warning.
2979 2017-12-19 Marek Polacek <polacek@redhat.com>
2982 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2985 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2987 * config/arc/arc.c (overriderregs): New variable.
2988 (arc_override_options): Track fixed/call saved/call options.
2989 (arc_conditional_register_usage): Check against overrideregs
2990 variable whenever we change register properties.
2992 2017-12-19 Nathan Sidwell <nathan@acm.org>
2994 * opts.c (finish_options): Don't prefix dump_base_name if it
2995 already contains directories.
2997 2017-12-19 Martin Liska <mliska@suse.cz>
2999 PR rtl-optimization/82675
3000 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
3001 more element in sbitmap.
3003 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
3005 * gimplify.c (gimplify_expr): Use error_operand_p.
3007 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
3010 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
3013 2017-12-19 Daniel Cederman <cederman@gaisler.com>
3015 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
3016 the jump is to a label.
3018 2017-12-19 Jakub Jelinek <jakub@redhat.com>
3020 PR tree-optimization/83444
3021 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
3026 * ipa-inline.c (flatten_remove_node_hook): New function.
3027 (ipa_inline): Keep only nodes with flatten attribute at the end of
3028 the array in the order from ipa_reverse_postorder, only walk that
3029 portion of array for flattening, if there is more than one such
3030 node, temporarily register a removal hook and ignore removed nodes.
3032 PR tree-optimization/80631
3033 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
3034 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
3035 IFN_REDUC_MAX or IFN_REDUC_MIN.
3037 2017-12-18 Jeff Law <law@redhat.com>
3039 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
3040 record symbolic equivalences from backedges in the CFG.
3043 2017-11-19 Jeff Law <law@redhat.com>
3045 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
3046 of degenerates resulting from ignoring an edge.
3048 2017-12-18 Martin Sebor <msebor@redhat.com>
3051 PR tree-optimization/78450
3052 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
3053 (handle_builtin_strlen): Call it.
3055 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
3057 PR rtl-optimization/83424
3058 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
3060 2017-12-18 Marek Polacek <polacek@redhat.com>
3063 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
3064 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
3067 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
3069 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
3070 (umaddsidi4, umaddsidi_split): Likewise.
3072 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
3074 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
3077 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3080 * config/s390/s390.c (s390_option_override): Avoid strncpy.
3082 2017-12-18 Richard Biener <rguenther@suse.de>
3084 PR tree-optimization/81877
3085 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
3086 (outermost_indep_loop): Adjust.
3087 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
3088 (can_sm_ref_p): Adjust.
3090 2017-12-18 Richard Biener <rguenther@suse.de>
3093 * tree.c (array_at_struct_end_p): Return true if the underlying
3094 object has space for at least one element in excess of what
3095 the array domain specifies.
3097 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3099 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
3100 example to avoid overfull hbox.
3101 * doc/invoke.texi (Option Summary): Add missing @gol.
3102 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
3105 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3106 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3108 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
3109 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
3110 (AArch64 Options): Document -mverbose-cost-dump.
3111 (ARM Options): Likewise, plus -mflip-thumb.
3113 2017-12-17 Martin Sebor <msebor@redhat.com>
3116 * gimple-ssa-warn-restrict.c
3117 (builtin_memref::offset_out_of_bounds): Correct the handling of
3120 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3122 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
3124 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
3126 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
3128 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
3130 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
3133 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
3135 * doc/invoke.texi: Fix some typos.
3137 2017-12-16 Martin Sebor <msebor@redhat.com>
3139 PR tree-optimization/78918
3140 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
3141 * builtins.c (check_sizes): Rename...
3142 (check_access): ...to this. Rename function arguments for clarity.
3143 (check_memop_sizes): Adjust names.
3144 (expand_builtin_memchr, expand_builtin_memcpy): Same.
3145 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
3146 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
3147 (check_strncat_sizes, expand_builtin_strncat): Same.
3148 (expand_builtin_strncpy, expand_builtin_memset): Same.
3149 (expand_builtin_bzero, expand_builtin_memcmp): Same.
3150 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
3151 (maybe_emit_sprintf_chk_warning): Same.
3152 (expand_builtin_strcpy): Adjust.
3153 (expand_builtin_stpcpy): Same.
3154 (expand_builtin_with_bounds): Detect out-of-bounds accesses
3155 in pointer-checking forms of memcpy, memmove, and mempcpy.
3156 (gcall_to_tree_minimal, max_object_size): Define new functions.
3157 * builtins.h (max_object_size): Declare.
3158 * calls.c (alloc_max_size): Call max_object_size instead of
3159 hardcoding ssizetype limit.
3160 (get_size_range): Handle new argument.
3161 * calls.h (get_size_range): Add a new argument.
3162 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
3163 * doc/invoke.texi (-Wrestrict): Adjust, add example.
3164 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
3166 (gimple_fold_builtin_memory_chk): Same.
3167 (gimple_fold_builtin_stxcpy_chk): New function.
3168 * gimple-ssa-warn-restrict.c: New source.
3169 * gimple-ssa-warn-restrict.h: New header.
3170 * gimple.c (gimple_build_call_from_tree): Propagate location.
3171 * passes.def (pass_warn_restrict): Add new pass.
3172 * tree-pass.h (make_pass_warn_restrict): Declare.
3173 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
3175 (handle_builtin_strcat): Same.
3176 (strlen_optimize_stmt): Rename...
3177 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
3178 stpncpy, strncpy, and their checking forms.
3180 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
3182 PR rtl-optimization/82849
3183 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
3184 and get_max_loop_iterations_int.
3186 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3187 Alan Hayward <alan.hayward@arm.com>
3188 David Sherwood <david.sherwood@arm.com>
3190 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
3191 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
3192 size as a poly_uint64.
3193 (mode_for_vector, mode_for_int_vector): Take the number of vector
3194 elements as a poly_uint64.
3195 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
3196 size as a poly_uint64.
3197 (mode_for_vector, mode_for_int_vector): Take the number of vector
3198 elements as a poly_uint64.
3200 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3201 Alan Hayward <alan.hayward@arm.com>
3202 David Sherwood <david.sherwood@arm.com>
3204 * machmode.h (MACRO_MODE): New macro.
3205 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
3206 * targhooks.c (default_libcall_value, default_secondary_reload)
3207 (default_memory_move_cost, default_register_move_cost)
3208 (default_class_max_nregs): Likewise.
3210 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3211 Alan Hayward <alan.hayward@arm.com>
3212 David Sherwood <david.sherwood@arm.com>
3214 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
3215 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
3217 * genautomata.c (main): Likewise.
3218 * genconditions.c (write_header): Likewise.
3219 * genemit.c (main): Likewise.
3220 * genextract.c (print_header): Likewise.
3221 * genopinit.c (main): Likewise.
3222 * genoutput.c (output_prologue): Likewise.
3223 * genpeep.c (main): Likewise.
3224 * genpreds.c (write_insn_preds_c): Likewise.
3225 * genrecog.c (writer_header): Likewise.
3226 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
3227 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
3228 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
3229 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
3230 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
3231 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
3232 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
3233 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
3234 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
3235 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
3236 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
3237 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
3238 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
3239 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
3240 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
3241 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
3242 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
3243 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
3244 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
3245 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
3246 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
3247 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
3248 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
3249 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
3250 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
3251 * config/darwin.c (IN_TARGET_CODE): Likewise.
3252 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
3253 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
3254 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
3255 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
3256 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
3257 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
3258 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
3259 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
3260 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
3261 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
3262 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
3263 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
3264 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
3265 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
3266 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
3267 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
3268 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
3269 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
3270 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
3271 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
3272 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
3273 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
3274 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
3275 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
3276 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
3277 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
3278 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
3279 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
3280 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
3281 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
3282 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
3283 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
3284 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
3285 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
3286 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
3287 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
3288 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
3289 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
3290 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
3291 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
3292 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
3293 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
3294 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
3295 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
3296 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
3297 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
3298 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
3299 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
3300 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
3301 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
3302 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
3303 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
3304 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
3305 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
3306 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
3307 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
3308 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
3309 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
3310 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
3311 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
3312 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3313 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
3314 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
3315 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
3316 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
3317 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
3318 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
3319 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
3320 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
3321 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
3322 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
3323 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3324 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
3325 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
3326 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
3327 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
3328 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
3329 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
3330 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
3331 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
3332 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
3333 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
3334 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
3335 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
3336 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
3337 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
3338 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
3339 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
3340 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
3341 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
3342 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
3343 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
3344 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
3345 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
3346 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
3347 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
3348 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
3349 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
3350 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
3351 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
3352 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
3353 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
3354 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
3355 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
3356 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
3357 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
3359 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3360 Alan Hayward <alan.hayward@arm.com>
3361 David Sherwood <david.sherwood@arm.com>
3363 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
3366 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3367 Alan Hayward <alan.hayward@arm.com>
3368 David Sherwood <david.sherwood@arm.com>
3370 * doc/generic.texi (VEC_SERIES_EXPR): Document.
3371 * doc/md.texi (vec_series@var{m}): Document.
3372 * tree.def (VEC_SERIES_EXPR): New tree code.
3373 * tree.h (build_vec_series): Declare.
3374 * tree.c (build_vec_series): New function.
3375 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
3376 * tree-pretty-print.c (dump_generic_node): Likewise.
3377 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
3378 * tree-inline.c (estimate_operator_cost): Likewise.
3379 * expr.c (expand_expr_real_2): Likewise.
3380 * optabs-tree.c (optab_for_tree_code): Likewise.
3381 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
3382 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
3383 * expmed.c (make_tree): Handle VEC_SERIES.
3384 * optabs.def (vec_series_optab): New optab.
3385 * optabs.h (expand_vec_series_expr): Declare.
3386 * optabs.c (expand_vec_series_expr): New function.
3387 * tree-vect-generic.c (expand_vector_operations_1): Check that
3388 the operands also have vector type.
3390 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3391 Alan Hayward <alan.hawyard@arm.com>
3392 David Sherwood <david.sherwood@arm.com>
3394 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
3395 (VEC_COND_EXPR): Add missing @tindex.
3396 * doc/md.texi (vec_duplicate@var{m}): Document.
3397 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
3398 * tree.c (build_vector_from_val): Add stubbed-out handling of
3399 variable-length vectors, using VEC_DUPLICATE_EXPR.
3400 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
3401 * cfgexpand.c (expand_debug_expr): Likewise.
3402 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3403 * tree-inline.c (estimate_operator_cost): Likewise.
3404 * tree-pretty-print.c (dump_generic_node): Likewise.
3405 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
3406 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
3407 (test_vec_duplicate_folding): New function.
3408 (fold_const_c_tests): Call it.
3409 * optabs.def (vec_duplicate_optab): New optab.
3410 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
3411 * optabs.h (expand_vector_broadcast): Declare.
3412 * optabs.c (expand_vector_broadcast): Make non-static. Try using
3413 vec_duplicate_optab.
3414 * expr.c (store_constructor): Try using vec_duplicate_optab for
3416 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
3418 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
3421 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
3422 div/mod latencies a bit.
3424 2017-12-15 Jeff Law <law@redhat.com>
3426 PR tree-optimization/36550
3427 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
3428 (mark_threaded_blocks): Rewrite code to avoid block copying when
3429 optimizing for size. Don't pessimize blocks which will be
3430 copied, but all the statements will be dead.
3432 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
3434 PR tree-optimization/81165
3435 * tree-ssa-threadupdate.c (uses_in_bb): New.
3436 (estimate_threading_killed_stmts): New.
3437 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
3438 * tree-ssa-threadedge.c
3439 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
3442 2017-12-15 Jeff Law <law@redhat.com>
3444 PR tree-optimization/83410
3445 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
3446 threads when parallelizing loops.
3448 2017-12-15 Jakub Jelinek <jakub@redhat.com>
3450 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
3452 * config/alpha/alpha.c (vms_attribute_table): Swap
3453 affects_type_identity and handler fields, adjust comments.
3454 * config/mips/mips.c (mips_attribute_table): Likewise.
3455 * config/visium/visium.c (visium_attribute_table): Likewise.
3456 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
3457 * config/microblaze/microblaze.c (microblaze_attribute_table):
3459 * config/spu/spu.c (spu_attribute_table): Likewise.
3460 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
3461 * config/arc/arc.c (arc_attribute_table): Likewise.
3462 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
3463 * config/v850/v850.c (v850_handle_interrupt_attribute,
3464 v850_handle_data_area_attribute): Formatting fixes.
3465 (v850_attribute_table): Swap affects_type_identity and handler
3466 fields, adjust comments.
3467 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
3468 * config/arm/arm.c (arm_attribute_table): Likewise.
3469 * config/avr/avr.c (avr_attribute_table): Likewise.
3470 * config/s390/s390.c (s390_attribute_table): Likewise.
3471 * config/sh/sh.c (sh_attribute_table): Likewise.
3472 * config/i386/i386.c (ix86_handle_cconv_attribute,
3473 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
3474 (ix86_attribute_table): Swap affects_type_identity and handler
3475 fields, adjust comments.
3476 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3477 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
3478 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
3479 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
3480 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
3481 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3482 * config/rx/rx.c (rx_attribute_table): Likewise.
3483 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
3484 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
3485 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
3486 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
3487 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3488 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
3489 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
3490 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
3491 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3492 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3493 * doc/plugins.texi (user_attr): Likewise. Add NULL for
3495 * attribs.c (empty_attribute_table): Swap affects_type_identity and
3497 (register_scoped_attributes, decl_attributes): Formatting fixes.
3499 PR tree-optimization/83269
3500 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
3501 subtraction in arg0's type if type is signed and arg0 is unsigned.
3505 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
3506 convert? on @0 instead of convert. Check type of @1, not @0.
3507 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
3508 convert? on @0 instead of convert. Check type of @1, not @0.
3509 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
3510 only optimize if either both @1 and @2 types are narrower
3511 precision, or both are wider or equal precision, and in the former
3512 case only if both have undefined overflow.
3514 2017-12-15 Richard Biener <rguenther@suse.de>
3517 * internal-fn.def (IFN_NOP): Add.
3518 * internal-fn.c (expand_NOP): Do nothing.
3519 * lto-streamer-in.c (input_function): Instead of removing
3520 sanitizer calls replace them with IFN_NOP calls.
3522 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
3523 Alan Hayward <alan.hayward@arm.com>
3524 David Sherwood <david.sherwood@arm.com>
3526 * dse.c (store_info, read_info_type): Replace begin and end with
3528 (print_range): New function.
3529 (set_all_positions_unneeded, any_positions_needed_p)
3530 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
3532 (record_store): Likewise. Optimize the case in which all positions
3534 (get_stored_val): Replace read_begin and read_end with read_offset
3536 (replace_read): Update call accordingly.
3538 2017-12-15 Bin Cheng <bin.cheng@arm.com>
3540 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
3541 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
3542 (loop_cand::loop_cand): Initialize above members.
3543 (loop_cand::supported_operations): Delete.
3544 (loop_cand::can_interchange_p): Inline above function.
3545 (loop_cand::classify_simple_reduction): Record number of constant
3546 initialized simple reductions.
3547 (should_interchange_loops): New parameters. Check stmt cost of loops
3549 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
3550 Update call to should_interchange_loops.
3551 (should_interchange_loop_nest): Update call to
3552 should_interchange_loops.
3554 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
3557 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
3558 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
3559 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
3561 2017-12-15 Julia Koval <julia.koval@intel.com>
3563 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
3564 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
3566 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
3567 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
3568 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
3570 2017-12-15 Julia Koval <julia.koval@intel.com>
3572 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
3573 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
3574 * config/i386/sse.md (vaesenc_<mode>): New pattern.
3575 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
3576 _mm_aesenc_epi128): New intrinsics.
3578 2017-12-15 Julia Koval <julia.koval@intel.com>
3580 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
3581 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
3583 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
3584 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
3585 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
3587 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3589 * gimple-ssa-strength-reduction.c (analyze_increments):
3590 Distinguish replacement costs for constant strides from those for
3593 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3595 * var-tracking.c (variable_tracking_main_1): Formatting fix.
3597 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3599 * doc/invoke.texi: Document -Wcast-function-type.
3600 * recog.h (stored_funcptr): Change signature.
3601 * tree-dump.c (dump_node): Avoid warning.
3602 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
3604 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
3607 * gimple-fold.c (get_range_strlen): Add the handling of non-member
3610 2017-12-14 David Malcolm <dmalcolm@redhat.com>
3612 PR tree-optimization/83312
3613 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
3614 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
3615 "val" param, and to cope with arbitrary basic blocks.
3616 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
3617 handle NULL_TREE for "val", dropping "bb" param.
3618 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
3619 drop "bb" param. Handle NULL_TREE for "val".
3620 (find_case_label_for_value): Make "switch_stmt" param const.
3621 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
3623 (vrp_prop::check_all_array_refs): Reimplement as...
3624 (check_array_bounds_dom_walker::before_dom_children): ...this new
3625 vfunc. Replace linear search through BB block list, excluding
3626 those with non-executable in-edges via dominator walk.
3628 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3630 * config/arm/arm.opt (mverbose-cost-dump): New option.
3631 * config/arm/arm.c (arm_rtx_costs): Use it.
3633 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
3636 * reload1.c (emit_input_reload_insns): Skip debug markers.
3638 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
3640 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
3644 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
3645 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
3646 * config/sh/sh.c: Skip debug insns besides notes.
3647 * config/sh/sh.md: Likewise.
3648 * config/sh/sh_treg_combine.cc: Likewise.
3649 * config/sh/sync.md: Likewise.
3651 2017-12-14 Tom de Vries <tom@codesourcery.com>
3653 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3654 Add item for weak_undefined.
3656 2017-12-14 Richard Biener <rguenther@suse.de>
3658 PR tree-optimization/67842
3659 * sese.h (bb_in_region): Remove #if 0'ed code.
3661 2017-12-14 Richard Biener <rguenther@suse.de>
3663 PR tree-optimization/83326
3664 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3665 may_be_zero parameter and handle it by not marking the first
3666 peeled copy as not exiting the loop.
3667 (try_peel_loop): Likewise.
3668 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
3669 to handle the case of constant or zero iterations and perform
3670 loop header copying on-the-fly.
3672 2017-12-14 Richard Biener <rguenther@suse.de>
3674 PR tree-optimization/83418
3675 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
3676 Instead of asserting we don't get unfolded comparisons deal with them.
3678 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3681 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
3682 basic blocks. Assert debug bind insns don't appear outside of bbs,
3683 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
3686 PR tree-optimization/83198
3687 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
3688 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
3689 value if arg is a REAL_CST with incompatible type.
3691 2017-12-14 Sudakshina Das <sudi.das@arm.com>
3692 Bin Cheng <bin.cheng@arm.com>
3695 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
3697 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
3700 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3702 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
3703 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
3705 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
3706 Alan Hayward <alan.hayward@arm.com>
3707 David Sherwood <david.sherwood@arm.com>
3709 * poly-int.h: New file.
3710 * poly-int-types.h: Likewise.
3711 * coretypes.h: Include them.
3712 (POLY_INT_CONVERSION): Define.
3713 * target.def (estimated_poly_value): New hook.
3714 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
3715 * doc/tm.texi: Regenerate.
3716 * doc/poly-int.texi: New file.
3717 * doc/gccint.texi: Include it.
3718 * doc/rtl.texi: Describe restrictions on subreg modes.
3719 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
3720 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
3721 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
3722 * targhooks.h (default_estimated_poly_value): Declare.
3723 * targhooks.c (default_estimated_poly_value): New function.
3724 * target.h (estimated_poly_value): Likewise.
3725 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
3726 (wi::unary_traits): Delete.
3727 (wi::binary_traits::signed_shift_result_type): Define for
3728 offset_int << HOST_WIDE_INT, etc.
3729 (generic_wide_int::operator <<=): Define for all types and use
3730 wi::lshift instead of <<.
3731 (wi::hwi_with_prec): Add a default constructor.
3732 (wi::ints_for): New class.
3733 (operator <<): Define for all wide-int types.
3734 (operator /): New function.
3735 (operator %): Likewise.
3736 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
3737 (ASSERT_MAYBE_NE_AT): New macros.
3739 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
3740 Dominik Vogt <vogt@linux.vnet.ibm.com>
3743 * emit-rtl.c (init_emit): Remove ??? comment.
3744 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3745 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
3746 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
3747 alignment of 3 virtual registers to BITS_PER_WORD.
3749 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
3751 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
3753 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
3754 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
3756 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3759 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
3761 2017-12-13 Jakub Jelinek <jakub@redhat.com>
3763 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
3764 with the exception of debug begin stmt markers appear before
3768 * final.c (rest_of_handle_final): Call variable_tracking_main only
3769 if !flag_var_tracking.
3771 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3772 Jakub Jelinek <jakub@redhat.com>
3776 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3777 labels debug stmts that can only appear after labels.
3779 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3781 PR rtl-optimization/82398
3782 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3783 EXPR_USEFULNESS in priority comparison.
3785 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3787 PR rtl-optimization/83393
3788 * combine.c (move_deaths): If reg_stat points to a too new insn in
3789 last_death, do not use it: find the proper insn instead.
3791 2017-12-12 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3794 * config/rs6000/rs6000-string.c (expand_block_compare,
3795 expand_strn_compare): Fix set_mem_size() calls.
3797 2017-12-12 Jeff Law <law@redhat.com>
3799 PR tree-optimization/83298
3800 PR tree-optimization/83362
3801 PR tree-optimization/83383
3802 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3803 push_value_range a public interface. Add new argument to
3804 record_ranges_from_stmt.
3805 * gimple-ssa-evrp-analyze.c
3806 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3807 Update comments. Handle recording temporary equivalences.
3808 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3809 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3810 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3811 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3812 gimple-ssa-evrp-analyze.h.
3813 (record_temporary_equivalences_from_phis): Add new argument. When
3814 the PHI arg is an SSA_NAME, set the result's range to the range
3816 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3817 from statements too.
3818 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3819 Pass it down to children as needed.
3820 (thread_outgoing_edges): Likewise.
3821 (thread_across_edge): Likewise. Push/pop range state as needed.
3822 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3824 2017-12-12 Julia Koval <julia.koval@intel.com>
3826 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3827 (PTA_CANNONLAKE): Remove PTA_CLWB.
3829 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3832 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3833 (vcondv2div2df): Likewise.
3834 (vconduv2dfv2di): Likewise.
3836 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3838 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3839 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3840 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3841 ATTR_MATHFN_FPROUNDING_ERRNO.
3843 2017-12-12 Richard Biener <rguenther@suse.de>
3845 PR tree-optimization/83385
3846 * graphite-scop-detection.c (get_order, order): Remove.
3847 (bb_to_rpo): New global.
3849 (build_scops): Sort pbbs in RPO order.
3851 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3853 * combine.c (simplify_set): Do not transform subregs to zero_extends
3854 if the destination mode is a vector mode.
3856 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3858 PR tree-optimization/83359
3859 * tree-cfg.h (fold_loop_internal_call): Declare.
3860 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3861 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3862 (find_loop_dist_alias): New function.
3863 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3864 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3867 PR tree-optimization/80631
3868 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3869 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3870 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3871 hardcoding zero as the value if COND_EXPR is never true. For
3872 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3873 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3874 hardcoding MAX_EXPR as the reduction operation.
3875 (is_nonwrapping_integer_induction): Allow negative step.
3876 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3877 vect_create_epilog_for_reduction, if no value is suitable, don't
3878 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3880 2017-12-12 Richard Biener <rguenther@suse.de>
3882 PR tree-optimization/81889
3883 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3884 range info from the non-wrapping IV instead of just the range
3887 2017-12-12 Julia Koval <julia.koval@intel.com>
3889 * config.gcc: Add vaesintrin.h.
3890 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3892 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3893 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3895 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3896 * config/i386/immintrin.h: Include vaesintrin.h.
3897 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3898 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3899 _mm_aesdec_epi128): New intrinsics.
3901 2017-12-12 Julia Koval <julia.koval@intel.com>
3903 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3904 OPTION_MASK_ISA_VAES_UNSET): New.
3905 (ix86_handle_option): Handle -mvaes.
3906 * config/i386/cpuid.h: Define bit_VAES.
3907 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3908 * config/i386/i386-c.c (__VAES__): New.
3909 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3910 (ix86_valid_target_attribute_inner_p): Ditto.
3911 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3912 * config/i386/i386.opt: Add -mvaes.
3913 * doc/invoke.texi: Ditto.
3915 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3917 * debug.h (gcc_debug_hooks): Add inline_entry.
3918 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3919 * debug.c (do_nothing_debug_hooks): Likewise.
3920 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3921 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3922 (dwarf2_lineno_debug_hooks): Likewise.
3924 * common.opt (gstatement-frontiers): New, setting
3925 debug_nonbind_markers_p.
3926 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3927 * toplev.c (process_options): Autodetect value for debug statement
3929 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3930 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3932 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3933 markers. Integrate source bind into debug stmt expand loop.
3934 (pass_expand::execute): Check debug marker limit. Avoid deep
3935 TER and expand debug locations for debug bind insns only.
3936 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3938 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3939 * final.c (reemit_insn_block_notes): Take current block from
3940 nonbind markers. Declare note where it's first set.
3941 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3942 begin stmt markers if enabled.
3943 (notice_source_line): Handle nonbind markers. Fail if their
3944 location is unknown or that of builtins.
3945 (rest_of_handle_final): Convert begin stmt markers to notes if
3946 var-tracking didn't run.
3947 (rest_of_clean_state): Skip begin stmt markers.
3948 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3950 * function.c (allocate_struct_function): Set begin_stmt_markers.
3951 * function.h (struct function): Add debug_marker_count counter
3952 and debug_nonbind_markers flag.
3953 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3954 * gimple-low.c (lower_function_body): Adjust
3955 debug_nonbind_markers.
3956 (lower_stmt): Drop or skip gimple debug stmts.
3957 (lower_try_catch): Skip debug stmts.
3958 * gimple.c (gimple_build_debug_begin_stmt): New.
3959 (gimple_copy): Increment debug_marker_count if copying one.
3960 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3961 * gimplify.c (rexpr_location): New.
3962 (rexpr_has_location): New.
3963 (warn_switch_unreachable_r): Handle gimple debug stmts.
3964 (shortcut_cond_r): Call expr_location.
3966 (find_goto_label): New.
3967 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3969 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3971 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3972 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3973 (LANG_HOOKS_INITIALIZER): ... this.
3974 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3975 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3977 (update_ebb_live_info): Skip debug insn markers.
3978 * lra.c (debug_insn_static_data): Rename to...
3979 (debug_bind_static_data): ... this.
3980 (debug_marker_static_data): New.
3981 (lra_set_insn_recog_data): Select one of the above depending
3983 (lra_update_isn_regno_info): Don't assume debug insns have
3985 (push_insns): Skip debug insns.
3986 * lto-streamer-in.c (input_function): Drop debug stmts
3987 depending on active options. Adjust debug_nonbind_markers.
3988 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3989 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3990 begin stmt marker notes.
3991 (print_insn): Likewise.
3992 * recog.c (extract_insn): Recognize rtl for debug markers.
3993 * rtl.def (DEBUG_MARKER): New.
3994 * tree-inline.c: Include params.h.
3995 (remap_gimple_stmt): Handle nonbind markers.
3996 (maybe_move_debug_stmts_to_successors): Likewise.
3997 (copy_debug_stmt): Likewise.
3998 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3999 markers regardless of no side effects.
4000 (tsi_link_before): Don't update container's side effects when adding
4001 a begin stmt marker.
4002 (tsi_link_after): Likewise.
4003 (expr_first): Skip begin stmt markers.
4004 (expr_last): Likewise.
4005 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
4006 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
4007 Disregard nonbind markers.
4008 * tree.c (make_node_stat): Don't set side effects for begin stmt
4010 (build1_stat): Likewise.
4011 * tree.def (DEBUG_BEGIN_STMT): New.
4012 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
4013 * var-tracking.c (delete_debug_insns): Renamed to...
4014 (delete_vta_debug_insns): ... this.
4015 (reemit_marker_as_note): New.
4016 (vt_initialize): Reemit markers.
4017 (delete_vta_debug_insns): Likewise.
4018 (vt_debug_insns_local): Reemit or delete markers.
4019 (variable_tracking_main_1): Likewise.
4020 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
4021 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
4022 (gimple_debug_nonbind_marker_p): New.
4023 (gimple_build_debug_bind): Adjust.
4024 (gimple_build_debug_begin_stmt): New.
4025 * doc/invoke.texi (max-debug-marker-count): New param.
4026 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
4027 (debug_parameter_ref, debug_marker): New.
4028 (NOTE_INSN_BEGIN_STMT): New.
4029 (DEBUG_INSN): Describe begin stmt markers.
4031 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
4032 without debug insns, we wouldn't, but clean up debug insns
4033 after a control flow insn nevertheless.
4035 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
4036 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
4037 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
4038 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
4039 (rtl_tidy_fallthru_edge): Likewise.
4040 (rtl_verify_fallthru): Likewise.
4041 (rtl_verify_bb_layout): Likewise.
4042 (skip_insns_after_block): Likewise.
4043 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
4044 * dwarf2out.c: Include print-rtl.h.
4045 (dwarf2out_next_real_insn): New.
4046 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
4047 Dump debug binds in asm comments.
4048 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
4049 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
4050 callers to use gsi_start_nondebug_bb instead.
4051 (gsi_after_labels): Skip gimple debug stmts.
4052 (gsi_start_nondebug): New.
4053 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
4054 (proper_loop_form_for_interchange): Adjust.
4055 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
4056 * gimple.h (gimple_seq_last_nondebug_stmt): New.
4057 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
4058 (collect_fallthrough_labels): Likewise.
4059 (should_warn_for_implicit_fallthrough): Likewise.
4060 (warn_implicit_fallthrough_r): Likewise.
4061 (expand_FALLTHROUGH_r): Likewise.
4062 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
4063 (graphite_copy_stmts_from_block): Skip nonbind markers.
4064 * haifa-sched.c (sched_extend_bb): Skip debug insns.
4065 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
4066 * jump.c (clean_barriers): Skip debug insns.
4067 * omp-expand.c (expand_parallel_call): Skip debug insns.
4068 (expand_task_call): Likewise.
4069 (remove_exit_barrier): Likewise.
4070 (expand_omp_taskreg): Likewise.
4071 (expand_omp_for_init_counts): Likewise.
4072 (expand_omp_for_generic): Likewise.
4073 (expand_omp_for_static_nochunk): Likewise.
4074 (expand_omp_for_static_chunk): Likewise.
4075 (expand_omp_simd): Likewise.
4076 (expand_omp_taskloop_for_outer): Likewise.
4077 (expand_omp_taskloop_for_inner): Likewise.
4078 (expand_oacc_for): Likewise.
4079 (expand_omp_sections): Likewise.
4080 (expand_omp_single): Likewise.
4081 (expand_omp_synch): Likewise.
4082 (expand_omp_atomic_load): Likewise.
4083 (expand_omp_atomic_store): Likewise.
4084 (expand_omp_atomic_fetch_op): Likewise.
4085 (expand_omp_atomic_pipeline): Likewise.
4086 (expand_omp_atomic_mutex): Likewise.
4087 (expand_omp_target): Likewise.
4088 (grid_expand_omp_for_loop): Likewise.
4089 (grid_expand_target_grid_body): Likewise.
4090 (build_omp_regions_1): Likewise.
4091 * omp-low.c (check_combined_parallel): Skip debug stmts.
4092 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
4093 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
4094 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
4096 * tree-cfg.c (make_blobs_1): Skip debug stmts.
4097 (make_edges): Likewise.
4098 (cleanup_dead_labels): Likewise.
4099 (gimple_can_merge_blocks_p): Likewise.
4100 (stmt_starts_bb_p): Likewise.
4101 (gimple_block_label): Likewise.
4102 (gimple_redirect_edge_and_branch): Likewise.
4103 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
4105 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
4107 * tree-pretty-print (print_declaration): Omit initializer in slim
4109 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
4111 (eliminate_unnecessary_stmts): Stabilize block removal order.
4112 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
4113 * var-tracking.c (get_first_insn): New.
4114 (vt_emit_notes): Call it.
4115 (vt_initialize): Walk any insns before the first BB.
4116 (delete_debug_insns): Likewise.
4118 * gimple.h (enum gimple_debug_subcode): Add
4119 GIMPLE_DEBUG_BEGIN_STMT.
4120 (gimple_debug_begin_stmt_p): New.
4121 (gimple_debug_nonbind_marker_p): New.
4122 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
4123 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
4124 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
4125 * insn-notes.def (BEGIN_STMT): New.
4126 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
4127 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
4128 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
4129 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
4130 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
4131 (INSN_DEBUG_MARKER_KIND): New.
4132 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
4133 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
4134 (INSN_VAR_LOCATION_PTR): New.
4135 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
4137 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
4138 deps only if debug bind insns are enabled.
4139 (pass_expand::execute): Avoid deep TER and expand
4140 debug locations for debug bind insns only.
4141 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
4142 debug stmts special handling down to debug bind stmts.
4143 * combine.c (try_combine): Narrow debug insns special handling
4144 down to debug bind insns.
4145 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
4146 Narrow debug insns preexisting special handling down to debug
4148 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
4149 handling down to debug bind insns.
4150 * function.c (instantiate_virtual_regs): Skip debug markers,
4151 adjust handling of debug binds.
4152 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
4153 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
4154 * haifa-sched.c (schedule_insn): Narrow special handling of debug
4155 insns to debug bind insns.
4156 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
4157 special handling of debug stmts to debug bind stmts.
4158 * ipa-split.c (split_function): Likewise.
4159 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
4160 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
4162 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
4163 * regrename.c (build_def_use): Likewise.
4164 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
4165 (pass_cprop_hardreg): Narrow special casing of debug insns to
4167 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
4168 * reload1.c (reload): Likewise.
4169 * sese.c (sese_insert_phis_for_liveouts): Narrow special
4170 casing of debug stmts to debug bind stmts.
4171 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
4172 * ssa-iterators.h (num_imm_uses): Likewise.
4173 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
4174 debug stmts to debug bind stmts.
4175 * tree-inline.c (tree_function_versioning): Narrow special casing
4176 of debug stmts to debug bind stmts.
4177 * tree-loop-distribution.c (generate_loops_for_partition):
4178 Narrow special casing of debug stmts to debug bind stmts.
4179 * tree-sra.c (analyze_access_subtree): Narrow special casing
4180 of debug stmts to debug bind stmts.
4181 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
4182 stmts to debug bind stmts.
4183 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
4184 casing of debug stmts to debug bind stmts.
4185 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
4186 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
4187 casing of debug stmts to debug bind stmts.
4188 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
4190 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
4191 debug stmts to debug bind stmts.
4192 (gimple_replace_ssa_lhs): Likewise.
4193 (insert_debug_temp_for_var_def): Likewise.
4194 (insert_debug_temps_for_defs): Likewise.
4195 (reset_debug_uses): Likewise.
4196 * tree-ssanames.c (release_ssa_name_fn): Likewise.
4197 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
4198 (adjust_debug_stmts): Likewise.
4199 (adjust_phi_and_debug_stmts): Likewise.
4200 (vect_do_peeling): Likewise.
4201 * tree-vect-loop.c (vect_transform_loop): Likewise.
4202 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
4203 * var-tracking.c (adjust_mems): Narrow special casing of debug
4204 insns to debug bind insns.
4205 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
4206 (compute_bb_dataflow, vt_find_locations): Likewise.
4207 (vt_expand_loc, emit_notes_for_changes): Likewise.
4208 (vt_init_cfa_base): Likewise.
4209 (vt_emit_notes): Likewise.
4210 (vt_initialize): Likewise.
4211 (vt_finalize): Likewise.
4213 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
4214 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
4215 (next_nonnote_nondebug_insn_bb): New.
4216 (prev_nonnote_nondebug_insn_bb): New.
4217 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
4218 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
4219 (prev_nonnote_nondebug_insn_bb): Declare.
4220 (next_nonnote_nondebug_insn_bb): Declare.
4221 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
4222 * cfgrtl.c (get_last_bb_insn): Likewise.
4223 * lra.c (push_insns): Likewise.
4225 2017-12-11 David Malcolm <dmalcolm@redhat.com>
4228 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
4230 * selftest.c (selftest::test_runner::test_runner): New ctor.
4231 (selftest::test_runner::~test_runner): New dtor.
4232 * selftest.h (class selftest::test_runner): New class.
4234 2017-12-11 Carl Love <cel@us.ibm.com>
4236 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
4237 vec_extract_fp32_from_shortl]): Add #defines.
4238 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
4239 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
4240 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
4241 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
4242 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
4243 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
4244 * doc/extend.texi: Add documentation for the added builtins.
4246 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
4248 PR rtl-optimization/80693
4249 PR rtl-optimization/81019
4250 PR rtl-optimization/81020
4251 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
4252 are not mentioned in i3. Place the REG_UNUSED note on i2,
4253 possibly modified to REG_DEAD, if it did not originate in i3.
4255 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4257 * recog.c (store_data_bypass_p_1): New function.
4258 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
4259 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
4261 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
4263 PR rtl-optimization/83361
4264 * ifcvt.c (if_convert): Call fixup_partitions.
4266 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
4268 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4269 early folding of splat_u{8,16,32}.
4271 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4273 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
4274 output_operand_lossage first argument with capital letter.
4275 (aarch64_override_options): Don't start error and sorry first argument
4276 with capital letter.
4278 2017-12-11 Andi Kleen <ak@linux.intel.com>
4280 PR gcov-profile/83355
4281 * auto-profile.c (string_table::get_index_by_decl): Don't
4282 recurse when abstract origin points to itself.
4284 2017-12-11 Bin Cheng <bin.cheng@arm.com>
4286 PR tree-optimization/83320
4287 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
4288 (prune_datarefs_not_in_loop): Ditto.
4290 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
4292 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
4294 2017-12-10 Jakub Jelinek <jakub@redhat.com>
4296 PR tree-optimization/83337
4297 * gimple-loop-interchange.cc (compute_access_stride): Handle
4298 bitfield DRs properly.
4300 2017-12-09 Jakub Jelinek <jakub@redhat.com>
4302 PR tree-optimization/83338
4303 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
4304 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
4305 vector of unsigned integers to vector of signed integers.
4307 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
4309 PR rtl-optimization/83317
4310 * lra-constraints.c (process_address_1): Add insn code check.
4312 2017-12-08 Michael Matz <matz@suse.de>
4314 Fix PR tree-optimization/83323
4315 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
4316 head-controlled loops and loop BBs.
4317 * common.opt (funroll-and-jam): Remove, instead ...
4318 (floop-unroll-and-jam): ... reuse this option.
4319 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
4320 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
4321 (-floop-unroll-and-jam): ... this option.
4323 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4325 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
4326 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
4327 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
4329 2017-12-08 Richard Biener <rguenther@suse.de>
4332 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
4333 handle accesses outside of zero-sized vars.
4335 2017-12-08 Martin Jambor <mjambor@suse.cz>
4337 PR tree-optimization/83141
4338 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
4339 test for MEM_REFs implicitely changing types with padding. Remove
4341 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
4343 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4345 * config/arc/arc.c (arc_attribute_table): Add exclusions to
4347 * config/avr/avr.c (avr_attribute_table): Likewise.
4348 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
4349 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
4350 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
4351 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
4352 of struct attribute_spec.
4353 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
4355 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
4358 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
4360 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
4363 * profile-count.c (profile_count::from_gcov_type): Move from
4364 profile-count.h; handle overflow.
4365 * profile-count.h (profile_count::from_gcov_type): Move offline.
4367 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
4369 PR rtl-optimization/83304
4370 * combine.c (move_deaths): If we do not know where a register died,
4373 2017-12-08 Richard Biener <rguenther@suse.de>
4375 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
4376 Provide -fopt-info-loop feedback when we interchange in a nest.
4378 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4380 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
4381 for armv6 ARM CPU IDs.
4383 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4385 * common/config/arm/arm-common.c: Include <algorithm>.
4386 (INCLUDE_VECTOR): Define.
4387 (compare_opt_names): New function.
4388 (arm_rewrite_selected_arch): Only strip out extensions that can be
4389 expressed through -mfpu. Sort the remaining extensions
4392 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4394 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
4395 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
4396 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
4397 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
4399 2017-06-08 Tristan Gingold <gindold@adacore.com>
4402 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
4403 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
4405 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4407 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
4408 of struct attribute_spec.
4410 2017-12-08 Julia Koval <julia.koval@intel.com>
4412 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
4413 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
4414 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
4415 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
4416 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
4417 _mm_maskz_dpwssds_epi32): Ditto.
4419 2017-12-08 Richard Biener <rguenther@suse.de>
4421 PR tree-optimization/81303
4422 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
4423 conditions try to create a comparison vector type matching
4424 the data vector type.
4425 (vectorizable_condition): Adjust.
4426 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
4427 Leave invariant conditions alone in case we can vectorize those.
4429 2017-12-08 Julia Koval <julia.koval@intel.com>
4431 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
4432 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
4433 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
4434 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
4435 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
4437 2017-12-08 Julia Koval <julia.koval@intel.com>
4439 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
4440 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
4441 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
4442 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
4443 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
4444 _mm_maskz_dpbusds_epi32): New intrinsics.
4446 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4448 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
4450 (ld<bh>io_signed): Likewise.
4451 (st<bhw>io): Likewise.
4452 * config/nios2/predicates.md (ldstio_memory_operand): Allow
4453 SMALL_INT12 constant integer operand.
4455 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4457 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
4459 (nios2_large_constant_allowed): ...this. Adjust uses.
4460 (nios2_plus_symbolic_constant_p): Rename to...
4461 (nios2_plus_large_constant_p): ...this. Adjust uses.
4462 (nios2_legitimate_address_p): Correct CONST_INT handling.
4463 (nios2_symbolic_memory_operand_p): Rename to...
4464 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
4465 (nios2_large_constant_p): Check for large constant integers too.
4466 (nios2_split_large_constant): Handle constant integers.
4467 (nios2_split_symbolic_memory_operand): Rename to...
4468 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
4469 (nios2_legitimize_constant_address): Handle constant integers.
4470 (r0rel_constant_p): Handle small constant integers.
4471 (nios2_print_operand_address): Handle r0-relative integer addresses.
4472 * config/nios2/nios2-protos.h: Adjust for renamed functions.
4473 * config/nios2/nios2.md: Adjust for renamed functions.
4475 2017-12-07 Andrew Waterman <andrew@sifive.com>
4477 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
4478 (TARGET_HAVE_SRODATA_SECTION): New define.
4479 (riscv_select_section): New function.
4481 2017-12-08 Joseph Myers <joseph@codesourcery.com>
4482 Alexander Monakov <amonakov@ispras.ru>
4483 Jakub Jelinek <jakub@redhat.com>
4486 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
4488 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4490 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
4491 a bool success value. Don't call output_operand_lossage here.
4492 (aarch64_print_ldpstp_address): Return a bool success value.
4493 (aarch64_print_operand_address): Call output_addr_const if
4494 aarch64_print_address_internal fails.
4495 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
4496 'y'; call output_operand_lossage instead. Call output_operand_lossage
4497 if aarch64_print_ldpstp_address fails.
4499 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4501 * tree-vector-builder.h
4502 (tree_vector_builder::binary_encoded_nelts): Declare.
4503 * tree-vector-builder.c
4504 (tree_vector_builder::binary_encoded_nelts): New function.
4505 * fold-const.c (negate_expr_p): Likewise.
4506 (operand_equal_p, fold_checksum_tree): Likewise.
4507 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
4508 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
4509 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
4510 (uniform_vector_p): Likewise.
4511 * varasm.c (const_hash_1, compare_constant): Likewise.
4512 * tree-ssa-ccp.c: Include tree-vector-builder.h.
4513 (valid_lattice_transition): Operate directly on the VECTOR_CST
4515 * ipa-icf.c: Include tree-vector-builder.h.
4516 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
4517 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
4519 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4521 * tree.c (build_vector): Delete.
4522 * tree.h (build_vector): Make static and move into the self-testing
4525 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4527 * vector-builder.h (vector_builder::derived): New const overload.
4528 (vector_builder::elt): New function.
4529 * tree-vector-builder.h (tree_vector_builder::type): New function.
4530 (tree_vector_builder::apply_step): Declare.
4531 * tree-vector-builder.c (tree_vector_builder::apply_step): New
4533 * gimple-fold.h (tree_vector_builder): Declare.
4534 (gimple_build_vector): Take a tree_vector_builder instead of a
4535 type and vector of elements.
4536 * gimple-fold.c (gimple_build_vector): Likewise.
4537 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
4539 (get_initial_defs_for_reduction): Likewise.
4540 (vectorizable_induction): Likewise.
4542 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4544 * tree-vector-builder.h
4545 (tree_vector_builder::new_binary_operation): Declare.
4546 * tree-vector-builder.c
4547 (tree_vector_builder::new_binary_operation): New function.
4548 * fold-const.c (fold_relational_const): Use it.
4549 (const_binop): Likewise. Check that both input vectors have
4550 the same number of elements, thus excluding things like WIDEN_SUM.
4551 Check whether it is possible to operate directly on the encodings
4554 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4556 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
4557 new_unary_operation, operating only on the encoded elements.
4558 (const_unop): Likewise.
4559 (exact_inverse): Likewise.
4560 (distributes_over_addition_p): New function.
4561 (const_binop): Use tree_vector_builder and new_unary_operation
4562 for combinations of VECTOR_CST and INTEGER_CST. Operate only
4563 on the encoded elements unless the encoding is strided and the
4564 operation does not distribute over addition.
4565 (fold_convert_const): Use tree_vector_builder and
4566 new_unary_operation. Operate only on the encoded elements
4567 for truncating integer conversions, or for non-stepped encodings.
4569 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4571 * config/sparc/sparc.c: Include tree-vector-builder.h.
4572 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
4573 * expmed.c: Include tree-vector-builder.h.
4574 (make_tree): Use tree_vector_builder instead of build_vector.
4575 * fold-const.c: Include tree-vector-builder.h.
4576 (const_binop): Use tree_vector_builder instead of build_vector.
4577 (const_unop): Likewise.
4578 (native_interpret_vector): Likewise.
4579 (fold_vec_perm): Likewise.
4580 (fold_ternary_loc): Likewise.
4581 * gimple-fold.c: Include tree-vector-builder.h.
4582 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
4584 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
4585 (simplify_vector_constructor): Use tree_vector_builder instead
4587 * tree-vect-generic.c: Include tree-vector-builder.h.
4588 (add_rshift): Use tree_vector_builder instead of build_vector.
4589 (expand_vector_divmod): Likewise.
4590 (optimize_vector_constructor): Likewise.
4591 * tree-vect-loop.c: Include tree-vector-builder.h.
4592 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
4593 of build_vector. Explicitly use a stepped encoding for
4595 * tree-vect-slp.c: Include tree-vector-builder.h.
4596 (vect_get_constant_vectors): Use tree_vector_builder instead
4598 (vect_transform_slp_perm_load): Likewise.
4599 (vect_schedule_slp_instance): Likewise.
4600 * tree-vect-stmts.c: Include tree-vector-builder.h.
4601 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
4602 (vect_gen_perm_mask_any): Likewise.
4603 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
4604 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
4606 (build_vector_from_val): Likewise. Explicitly use a duplicate
4609 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
4611 * doc/generic.texi (VECTOR_CST): Describe new representation of
4613 * vector-builder.h: New file.
4614 * tree-vector-builder.h: Likewise.
4615 * tree-vector-builder.c: Likewise.
4616 * Makefile.in (OBJS): Add tree-vector-builder.o.
4617 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
4618 * tree-core.h (tree_base): Add a vector_cst field to the u union.
4619 (tree_vector): Change the number of elements to
4620 vector_cst_encoded_nelts.
4621 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
4622 (VECTOR_CST_ELTS): Delete.
4623 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
4624 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
4625 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
4626 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
4627 (VECTOR_CST_ENCODED_ELT): Likewise.
4628 (vector_cst_encoded_nelts): New function.
4629 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4630 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4631 (vector_cst_int_elt, vector_cst_elt): Declare.
4632 * tree.c: Include tree-vector-builder.h.
4633 (tree_code_size): Abort if passed VECTOR_CST.
4634 (tree_size): Update for new VECTOR_CST layout.
4635 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4636 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4637 (build_vector): Use tree_vector_builder.
4638 (vector_cst_int_elt, vector_cst_elt): New functions.
4639 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
4640 encoded elements and then create the vector in the canonical form.
4641 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
4642 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
4643 (tree_c_tests): Call test_vector_cst_patterns.
4644 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
4646 (hash_tree): Likewise.
4647 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
4648 (streamer_write_tree_header): Likewise.
4649 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
4650 (streamer_alloc_tree): Likewise. Update call to make_vector.
4651 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
4653 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4655 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
4656 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
4657 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
4659 2017-12-07 Bin Cheng <bin.cheng@arm.com>
4660 Richard Biener <rguenther@suse.de>
4662 PR tree-optimization/81303
4663 * Makefile.in (gimple-loop-interchange.o): New object file.
4664 * common.opt (floop-interchange): Reuse the option from graphite.
4665 * doc/invoke.texi (-floop-interchange): Ditto. New document for
4666 -floop-interchange and mention it for -O3.
4667 * opts.c (default_options_table): Enable -floop-interchange at -O3.
4668 * gimple-loop-interchange.cc: New file.
4669 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
4670 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
4671 * passes.def (pass_linterchange): New pass.
4672 * timevar.def (TV_LINTERCHANGE): New time var.
4673 * tree-pass.h (make_pass_linterchange): New declaration.
4674 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
4675 interchange. Record IV before/after increment in new parameters.
4676 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
4677 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
4679 (check_reduction_path): ...New function here.
4680 * tree-vectorizer.h (check_reduction_path): New declaration.
4682 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
4685 PR rtl-optimization/80818
4686 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
4687 always early clobbered.
4688 * lra-lives.c (process_bb_lives): Check input hard regs for early
4689 clobbered non-operand hard reg.
4691 2017-12-07 Jakub Jelinek <jakub@redhat.com>
4694 * tree-cfg.c (verify_gimple_assign_binary): Don't require
4695 types_compatible_p, just that TYPE_MODE is the same.
4697 2017-12-07 Martin Sebor <msebor@redhat.com>
4700 * attribs.c (empty_attribute_table): Initialize new member of
4701 struct attribute_spec.
4702 (decl_attributes): Add argument. Handle mutually exclusive
4703 combinations of attributes.
4704 (selftests::test_attribute_exclusions): New function.
4705 (selftests::attribute_c_tests): Ditto.
4706 * attribs.h (decl_attributes): Add default argument.
4707 * selftest.h (attribute_c_tests): Declare.
4708 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
4709 * tree-core.h (attribute_spec::exclusions, exclude): New type and
4711 * doc/extend.texi (Common Function Attributes): Update const and pure.
4712 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
4713 of struct attribute_spec.
4714 * config/arc/arc.c (arc_attribute_table): Same.
4715 * config/arm/arm.c (arm_attribute_table): Same.
4716 * config/avr/avr.c ( avr_attribute_table): Same.
4717 * config/bfin/bfin.c (bfin_attribute_table): Same.
4718 * config/cr16/cr16.c (cr16_attribute_table): Same.
4719 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
4720 * config/h8300/h8300.c (h8300_attribute_table): Same.
4721 * config/i386/i386.c (ix86_attribute_table): Same.
4722 * config/ia64/ia64.c (ia64_attribute_table): Same.
4723 * config/m32c/m32c.c (m32c_attribute_table): Same.
4724 * config/m32r/m32r.c (m32r_attribute_table): Same.
4725 * config/m68k/m68k.c (m68k_attribute_table): Same.
4726 * config/mcore/mcore.c (mcore_attribute_table): Same.
4727 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
4728 * config/mips/mips.c (mips_attribute_table): Same.
4729 * config/msp430/msp430.c (msp430_attribute_table): Same.
4730 * config/nds32/nds32.c (nds32_attribute_table): Same.
4731 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
4732 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
4733 * config/rl78/rl78.c (rl78__attribute_table): Same.
4734 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
4735 * onfig/rx/rx.c (rx_attribute_table): Same.
4736 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
4737 * config/sh/sh.c (sh_attribute_table): Same.
4738 * config/sparc/sparc.c (sparc_attribute_table): Same.
4739 * config/spu/spu.c (spu_attribute_table): Same.
4740 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
4741 * config/v850/v850.c (v850_attribute_table): Same.
4742 * config/visium/visium.c (visium_attribute_table): Same.
4744 2017-12-07 Tamar Christina <tamar.christina@arm.com>
4747 * config/arm/arm.c (INCLUDE_STRING): Define.
4748 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
4749 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
4752 2017-12-07 Michael Matz <matz@suse.de>
4754 Add unroll and jam pass
4756 * gimple-loop-jam.c: New file.
4757 * Makefile.in (OBJS): Add gimple-loop-jam.o.
4758 * common.opt (funroll-and-jam): New option.
4759 * opts.c (default_options_table): Add unroll-and-jam at -O3.
4760 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
4761 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
4762 * passes.def: Add pass_loop_jam.
4763 * timevar.def (TV_LOOP_JAM): Add.
4764 * tree-pass.h (make_pass_loop_jam): Declare.
4765 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
4766 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
4767 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
4768 to flow_loop_tree_node_add.
4769 (duplicate_subloops, copy_loops_to): Append to sibling list.
4770 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
4771 * doc/invoke.texi (-funroll-and-jam): Document new option.
4772 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
4774 2017-12-07 Richard Biener <rguenther@suse.de>
4776 PR tree-optimization/83296
4777 PR tree-optimization/67769
4778 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4779 flow sensitive info in an unrelated BB.
4780 (value_replacement): Use reset_flow_sensitive_info.
4781 (minmax_replacement): Reset flow sensitive info on the def
4782 we move. Do not reset flow sensitive info in the whole BB
4783 we move the stmt to.
4784 (abs_replacement): Likewise.
4786 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4789 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4790 rs6000_cpu to the given -mcpu=, or to the default processor.
4792 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4794 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4795 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4796 instead of rs6000_cpu_attr.
4797 (rs6000_adjust_cost): Ditto.
4798 (is_microcoded_insn): Ditto.
4799 (rs6000_adjust_priority): Ditto.
4800 (rs6000_issue_rate): Ditto.
4801 (rs6000_use_sched_lookahead): Ditto.
4802 (rs6000_use_sched_lookahead_guard): Ditto.
4803 (rs6000_sched_reorder): Ditto.
4804 (force_new_group): Ditto.
4805 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4806 (group_ending_nop): Ditto.
4808 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4810 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4811 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4812 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4813 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4815 (rs6000_reassociation_width): Ditto.
4816 (rs6000_emit_epilogue): Ditto.
4817 (rs6000_adjust_cost): Ditto.
4818 (is_microcoded_insn): Ditto.
4819 (is_cracked_insn): Ditto.
4820 (rs6000_adjust_priority): Ditto.
4821 (rs6000_sched_reorder): Ditto.
4822 (rs6000_sched_reorder2): Ditto.
4823 (insn_must_be_first_in_group): Ditto.
4824 (insn_must_be_last_in_group): Ditto.
4825 (rs6000_register_move_cost): Ditto.
4826 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4829 2017-12-07 Julia Koval <julia.koval@intel.com>
4831 * config.gcc: Add vaesintrin.h.
4832 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4834 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4835 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4837 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4838 * config/i386/immintrin.h: Include vaesintrin.h.
4839 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4840 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4841 _mm_aesdec_epi128): New intrinsics.
4843 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4845 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4846 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4847 c-family/c-spellcheck.cc.
4848 (best_macro_match::best_macro_match): Likewise.
4850 (struct edit_distance_traits<cpp_hashnode *>): Move to
4851 c-family/c-spellcheck.h.
4852 (class best_macro_match): Likewise.
4854 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4856 PR tree-optimization/83293
4857 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4858 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4859 might insert into empty bb.
4862 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4863 simplify for plus with :c added, and pointer_plus without that.
4864 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4865 with undefined overflow and the conversion is not widening,
4866 perform negation in utype and only convert to type afterwards.
4867 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4868 simplify for plus with :c added, and pointer_plus without that.
4869 If type is integral with undefined overflow and the conversion is
4870 not widening, perform minus in utype and only convert to type
4871 afterwards. Move the last pointer_diff_expr simplify into the
4874 2017-12-06 Martin Sebor <msebor@redhat.com>
4876 PR tree-optimization/82646
4877 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4878 strncpy, not maxlen.
4880 2017-12-06 Martin Sebor <msebor@redhat.com>
4882 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4885 PR tree-optimization/83075
4886 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4887 strncat/strncpy don't change length of source string.
4889 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4892 2017-11-29 Martin Aberg <maberg@gaisler.com>
4894 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4895 to prevent b2bst errata sequence.
4896 (sqrtdf2_fix): Likewise.
4898 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4900 PR tree-optimization/81945
4901 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4902 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4903 to dest_cfun has orig_loop_num set, either remap it to the new
4904 loop number if the loop got moved too, or clear it.
4906 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4908 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4910 (thunderx2t99_nothing): New insn reservation.
4911 (thunderx2t99_mrs): New insn reservation.
4912 (thunderx2t99_multiple): New insn reservation.
4913 (thunderx2t99_alu_basi): Add bfx to reservation.
4914 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4916 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4919 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4921 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4923 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4924 tree-ssa-pre.c::remove_dead_inserted_code.
4925 * tree-ssa-dce.h: New file.
4926 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4927 (remove_dead_inserted_code): Move and rename to function
4928 tree-ssa-dce.c::simple_dce_from_worklist.
4929 (pass_pre::execute): Update use.
4931 2017-12-05 Richard Biener <rguenther@suse.de>
4933 PR tree-optimization/83277
4934 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4935 to code-gen liveout vars.
4937 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4939 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4940 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4941 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4944 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4946 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4948 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4949 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4952 2017-12-05 Richard Biener <rguenther@suse.de>
4954 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4955 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4956 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4957 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4959 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4961 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4963 2017-12-05 Martin Liska <mliska@suse.cz>
4964 Jakub Jelinek <jakub@redhat.com>
4966 * doc/invoke.texi: Document the options.
4967 * flag-types.h (enum sanitize_code): Add
4968 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4969 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4970 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4971 * opts.c: Define new sanitizer options.
4972 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4973 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4975 2017-12-05 Julia Koval <julia.koval@intel.com>
4977 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4978 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4979 (ix86_handle_option): Handle -mavx512vnni.
4980 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4981 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4982 * config/i386/i386-c (__AVX512VNNI__): New.
4983 * config/i386/i386.c (ix86_target_string): Handle new option.
4984 (ix86_valid_target_attribute_inner_p): Handle new option.
4985 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4986 * config/i386/i386.opt (mavx512vnni): New option.
4988 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4991 * athlon.md: Disable for generic.
4992 * haswell.md: Enable for generic.
4993 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4994 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4996 (ix86_adjust_cost): Move generic to haswell path.
4998 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
5000 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
5001 instead of MEM_P in a couple more places. Fix formatting issues.
5003 2017-12-04 Jim Wilson <jimw@sifive.com>
5005 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
5006 instead of GP_REG_LAST-1.
5007 (riscv_adjust_libcall_cfi_prologue): Likewise.
5008 (riscv_adjust_libcall_cri_epilogue): Likewise.
5009 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
5012 2017-12-04 Luis Machado <luis.machado@linaro.org>
5014 * ipa-pure-const.c (check_decl): Add missing newline.
5015 (state_from_flags): Likewise.
5017 2017-12-04 Jeff Law <law@redhat.com>
5019 PR tree-optimizatin/78496
5020 * gimple-ssa-evrp-analyze.h
5021 (evrp_range_analyzer::get_vr_values): Simplify.
5022 * gimple-ssa-evrp-analyze.c: Corresponding changes.
5023 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
5024 and gimple-ssa-evrp-analyze.h.
5025 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
5026 (simplify_stmt_for_jump_threading): Copy a blob of code from
5027 tree-vrp.c to use ranges to simplify statements.
5028 (dom_opt_dom_walker::before_dom_children): Call
5029 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
5030 (dom_opt_dom_walker::after_dom_children): Similarly for
5031 evrp_range_analyzer::leave.
5032 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
5035 * gimple-ssa-evrp-analyze.c
5036 (evrp_range_analyzer::extract_range_from_stmt): Always use
5037 vr_values::update_value_range so preexisting range info is
5038 medged with new range info, even if the new range is VR_VARYING.
5040 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
5042 * combine.c: Adjust comment.
5043 (use_crosses_set_p): Delete.
5044 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
5045 (try_combine): Ditto.
5047 2017-12-04 Richard Biener <rguenther@suse.de>
5049 PR tree-optimization/83255
5050 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
5051 Re-add zero-iteration check.
5053 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
5055 PR rtl-optimization/83245
5056 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
5057 hard registers as earlyclobber, also if not in an asm.
5059 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
5063 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5066 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
5067 rs6000_cpu based on cpu_index, not tune_index.
5069 2017-12-04 Richard Biener <rguenther@suse.de>
5071 PR tree-optimization/83238
5072 * graphite-scop-detection.c (scop_detection::merge_sese): Make
5073 code match comment, rejecting invalid SESE regions.
5075 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
5077 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
5078 require base operand is a REG_POINTER prior to reload on targets
5079 with non-equivalent space registers.
5081 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
5083 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
5084 (update_specialized_profile): Fix updating of counts.
5085 (perhaps_add_new_callers): Likewise.
5087 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
5090 * x86-tune.def: Remove obsolette FIXMEs.
5091 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
5092 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
5093 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
5095 (X86_TUNE_PAD_RETURNS): Disable for generic.
5097 2017-12-02 Jakub Jelinek <jakub@redhat.com>
5099 PR tree-optimization/83170
5100 PR tree-optimization/83241
5101 * gimple-ssa-store-merging.c
5102 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
5103 gimple_vuse (ins_stmt) in case it has changed.
5104 (imm_store_chain_info::output_merged_store): Likewise.
5106 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
5110 * tree-cfg.c (pass_warn_function_return::execute): Handle
5111 __builtin_ubsan_handle_missing_return like __builtin_unreachable
5112 with BUILTINS_LOCATION.
5116 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
5117 is BLKmode for vector field with vector raw mode, use TYPE_MODE
5118 instead of DECL_MODE.
5120 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
5121 last argument to rtx pointer.
5122 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
5123 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
5124 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
5125 depending on the chosen ISAs.
5126 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
5127 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
5128 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
5130 * config/i386/sse.md (mov<mode>_internal): Likewise.
5131 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
5133 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5135 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
5137 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5139 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
5140 parameters from prototype.
5141 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
5142 parameters. Don't print a comment.
5143 (emit_fusion_gpr_load): Adjust.
5144 (emit_fusion_load_store): Adjust.
5145 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
5146 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
5147 comment on the second line.
5149 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5152 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
5153 rs6000_cpu based on cpu_index, not tune_index.
5155 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5157 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
5158 which_alternative instead of which_alternative + 1.
5159 (output_asm_insn): Print an extra tab if the template is short.
5161 2017-12-01 Jim Wilson <jimw@sifive.com>
5163 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
5165 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
5166 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
5168 * doc/tm.texi: Regenerate.
5170 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
5173 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
5174 whether we can allocate pseudos before trying to fix an address.
5175 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
5176 memory address is indexed or indirect.
5177 (floatuns_<mode>si2_hw2): Likewise.
5179 2017-12-01 Jason Merrill <jason@redhat.com>
5181 * Makefile.in (TAGS): Add c-family/*.cc.
5183 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
5185 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
5186 (call_value_insn): Likewise.
5187 (sibcall_insn): Likewise.
5188 (sibcall_value_insn): Likewise.
5189 (movsi_aarch64): Likewise.
5190 (movdi_aarch64): Likewise.
5191 (add_losym_): Likewise.
5192 (ldr_got_small_): Likewise.
5193 (ldr_got_small_sidi): Likewise.
5194 (ldr_got_small_28k_): Likewise.
5195 (ldr_got_small_28k_sidi): Likewise.
5196 * config/aarch64/aarch64.c (aarch64_print_address_internal):
5197 Move output_addr_const to symbolic case. Add error check.
5199 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5201 * config/s390/predicates.md (plus16_Q_operand): New predicate.
5202 * config/s390/s390.md: Disable MVC merging peephole if it would
5203 disable operand forwarding.
5204 (new peephole2): Split MVCs if it would turn them into up to 2
5207 2017-12-01 Richard Biener <rguenther@suse.de>
5209 PR tree-optimization/83232
5210 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
5211 detection of same access. Instead of breaking the group here
5212 do not consider the duplicate. Add comment explaining real fix.
5214 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
5216 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
5218 2017-12-01 Sudakshina Das <sudi.das@arm.com>
5220 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
5222 2017-12-01 Jakub Jelinek <jakub@redhat.com>
5224 * function.h (struct function): Remove cilk_frame_decl,
5225 is_cilk_function and calls_cilk_spawn fields.
5226 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
5228 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
5229 cilk_elemental field.
5230 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
5231 * target.def: Adjust comment.
5232 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
5233 Don't test cilk_elemental.
5235 PR tree-optimization/83233
5236 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
5237 bswap_stat name for the struct.
5240 * tree.h (SWITCH_BREAK_LABEL_P): Define.
5241 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
5242 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
5243 SWITCH_BREAK_LABEL_P set on the label.
5244 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
5245 added for default case if it was missing and not all cases covered.
5246 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
5247 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
5249 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
5250 to avoid -Wimplicit-fallthrough warning.
5251 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
5252 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
5254 PR tree-optimization/83221
5255 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
5257 (init_reassoc): Formatting fix.
5260 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
5261 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
5265 * tree-cfg.c: Include asan.h.
5266 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
5268 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5270 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
5272 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5274 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
5275 vpcmpeqd instruction.
5277 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5279 * config/i386/i386.c (standard_sse_constant_opcode): Fix
5280 registers type for 128bit mode.
5282 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5284 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
5286 * gengtype-state.c (read_state_pair): Likewise.
5287 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
5288 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
5289 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
5290 * ifcvt.c (noce_try_store_flag_constants): Likewise.
5291 * tree-ssa-ccp.c (ccp_finalize): Likewise.
5292 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
5293 * builtins.c (fold_builtin_3): Likewise.
5294 * graphite-scop-detection.c
5295 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
5296 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
5299 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
5302 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
5303 parameter. Return true if ignore_decl_rtl_set_p is true and other
5304 conditions are satisfied.
5305 * asan.h (asan_protect_global): Add new parameter.
5306 * varasm.c (categorize_decl_for_section): Pass true as second parameter
5307 to asan_protect_global calls.
5309 2017-11-30 Jim Wilson <jimw@sifive.com>
5311 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
5312 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
5313 -mstrict-align, add info on default value. Delete redundant lines for
5314 -mabi. Add missing -mexplicit-relocs docs.
5316 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5318 * config/arc/arc.md (trap): New pattern.
5320 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5322 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
5323 instruction to end into a delay slot.
5324 * config/arc/arc.md (cond_delay_insn): Check if the instruction
5325 can be placed into a delay slot against reg_note.
5327 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5329 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
5330 labels number of usages.
5332 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5334 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
5336 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
5338 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5341 * internal-fn.c (expand_mul_overflow): Optimize unsigned
5342 multiplication by power of 2 constant into two shifts + comparison.
5344 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
5347 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
5349 2017-11-30 Richard Biener <rguenther@suse.de>
5351 PR tree-optimization/83202
5352 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
5353 allow_peel argument and guard peeling.
5354 (canonicalize_loop_induction_variables): Likewise.
5355 (canonicalize_induction_variables): Pass false.
5356 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
5357 peeling from cunrolli.
5359 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5361 * combine.c (try_combine): Print a message to dump file whenever
5362 I0, I1, or I2 cannot be combined into I3.
5364 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5366 PR rtl-optimization/83156
5367 PR rtl-optimization/82621
5368 * combine.c (try_combine): Don't split an I2 if one of the dests is
5369 set again before I3. Allow unused dests.
5371 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5373 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
5375 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
5377 PR rtl-optimization/80818
5378 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
5379 recursively. Use insn code for clobber.
5380 (lra_set_insn_recog_data): Pass the new arg to
5381 collect_non_operand_hard_regs.
5382 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
5384 (lra_update_insn_regno_info): Pass insn to
5385 add_regs_to_insn_regno_info.
5387 2017-11-29 Jim Wilson <jimw@sifive.com>
5388 Andrew Waterman <andrew@sifive.com>
5390 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
5391 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
5392 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
5393 use SINGLE_SHIFT_COST.
5394 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
5395 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
5397 2017-11-29 Julia Koval <julia.koval@intel.com>
5399 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
5400 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
5401 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
5402 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
5403 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
5404 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
5405 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
5406 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
5407 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
5408 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
5409 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
5410 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
5411 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
5412 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
5413 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
5414 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
5415 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
5416 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
5417 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
5418 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
5419 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
5420 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
5421 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
5422 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
5423 __builtin_ia32_vpshldv_v2di_mask,
5424 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
5425 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
5426 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
5428 2017-11-29 Julia Koval <julia.koval@intel.com>
5430 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
5431 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
5432 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
5433 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
5434 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
5435 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
5436 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
5437 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
5438 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
5439 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
5440 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
5441 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
5442 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
5443 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
5444 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
5445 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
5446 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
5447 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
5448 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5449 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
5450 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
5452 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5454 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
5455 movsi_pic_gotdata_op instruction as a load for the UT699 errata
5458 2017-11-29 Martin Aberg <maberg@gaisler.com>
5460 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
5461 to prevent b2bst errata sequence.
5462 (sqrtdf2_fix): Likewise.
5464 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5466 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
5467 (div_sqrt_insn_p): New function.
5468 (sparc_do_work_around_errata): Insert NOP instructions to
5469 prevent sequences that could trigger the TN-0013 errata for
5470 certain LEON3 processors.
5471 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
5472 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
5473 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
5474 (in_branch_delay): Prevent div and sqrt in delay slot if
5476 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
5478 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5480 * config/sparc/sparc.c (atomic_insn_p): New function.
5481 (sparc_do_work_around_errata): Insert NOP instructions to
5482 prevent sequences that could trigger the TN-0010 errata for
5484 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
5485 instruction referable in atomic_insns_p.
5487 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5489 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
5490 (atomic_compare_and_swap_leon3_1): Likewise.
5493 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5495 * config/sparc/sparc.c (fpop_insn_p): New function.
5496 (sparc_do_work_around_errata): Insert NOP instructions to
5497 prevent sequences that could trigger the TN-0012 errata for
5499 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
5500 * config/sparc/sparc.md (fix_gr712rc): New attribute.
5501 (in_branch_annul_delay): Prevent floating-point instructions
5502 in delay slot of annulled integer branch.
5504 2017-11-29 Richard Biener <rguenther@suse.de>
5506 PR tree-optimization/83202
5507 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
5509 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
5510 nodes vectorized to the same stmts multiple times.
5511 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
5512 (vect_analyze_slp_instance): Adjust.
5513 (scalar_stmts_to_slp_tree_map_t): New typedef.
5514 (vect_schedule_slp_instance): Add a map recording the SLP node
5515 representing the vectorized stmts for a set of scalar stmts.
5516 Avoid code-generating redundancies.
5517 (vect_schedule_slp): Allocate map and pass it down.
5519 2017-11-29 Nathan Sidwell <nathan@acm.org>
5522 * tree.c (build_complex_type): Fix canonicalization. Only fill in
5525 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
5527 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
5528 cases for printing LDP/STP memory addresses.
5529 (aarch64_print_address_internal): Renamed from
5530 aarch64_print_operand_address, added parameter, add Pmode check.
5531 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
5532 (aarch64_print_operand_address): Indirect to
5533 aarch64_print_address_internal.
5534 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
5535 'y' operand output specifier.
5537 2017-11-29 Jakub Jelinek <jakub@redhat.com>
5540 * tree.c (build_simple_mem_ref_loc): Handle
5541 get_addr_base_and_unit_offset returning a MEM_REF.
5544 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
5547 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
5550 2017-11-28 David Malcolm <dmalcolm@redhat.com>
5552 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
5554 (layout::move_to_column): Likewise.
5556 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5559 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
5560 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
5562 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5563 Martin Jambor <mjambor@suse.cz>
5566 * tree.h (expr_type_first_operand_type_p): Declare
5567 * tree.c (expr_type_first_operand_type_p): New function.
5568 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
5569 (ipa_value_from_jfunc): Adjust declaration.
5570 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
5571 Use it as result type for arithmetics, unless it is NULL in which case
5572 be more conservative.
5573 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
5574 ipa_get_jf_pass_through_result.
5575 (propagate_vals_across_pass_through): Likewise.
5576 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
5577 is to propagate_vals_across_pass_through.
5578 (propagate_constants_across_call): Pass PARM_TYPE to
5579 propagate_scalar_across_jump_function.
5580 (find_more_scalar_values_for_callers_subset): Pass parameter type to
5581 ipa_value_from_jfunc.
5582 (cgraph_edge_brings_all_scalars_for_node): Likewise.
5583 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
5584 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
5585 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
5586 target_type, pass it to ipa_value_from_jfunc.
5587 (update_indirect_edges_after_inlining): Pass parameter type to
5588 try_make_edge_direct_simple_call.
5590 2017-11-28 Jeff Law <law@redhat.com>
5592 * gimple-ssa-evrp-analyze.c
5593 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
5594 refine ranges if scev_initialized_p returns true.
5595 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
5597 2017-11-28 Julia Koval <julia.koval@intel.com>
5599 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
5600 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
5601 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
5602 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
5603 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
5604 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
5605 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
5606 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
5607 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
5608 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
5609 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
5610 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
5611 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
5612 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
5613 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
5614 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
5615 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
5616 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
5617 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
5618 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
5619 __builtin_ia32_vpshrd_v2di_mask): New builtins.
5620 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
5622 2017-11-28 Julia Koval <julia.koval@intel.com>
5624 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
5625 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
5626 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
5627 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
5628 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
5629 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
5630 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
5631 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
5632 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
5633 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
5634 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
5635 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
5636 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
5637 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
5638 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
5639 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
5640 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
5641 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
5642 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
5643 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
5644 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
5645 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
5646 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
5647 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
5648 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
5649 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
5650 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
5651 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
5652 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
5653 __builtin_ia32_vpshld_v2di_mask): New builtins.
5654 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5655 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
5657 2017-11-28 Richard Biener <rguenther@suse.de>
5659 PR tree-optimization/80776
5660 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
5662 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
5664 (evrp_range_analyzer::record_ranges_from_incoming_edges):
5665 If the incoming edge is an effective fallthru because the other
5666 edge only reaches a __builtin_unreachable () then record ranges
5667 derived from the controlling condition in SSA info.
5668 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
5669 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5671 2017-11-28 Olivier Hainque <hainque@adacore.com>
5673 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
5674 on mingw build hosts.
5676 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5679 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
5681 2017-11-28 Richard Biener <rguenther@suse.de>
5684 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
5685 copies generated from memcpy use a character array as reference
5688 2017-11-28 Julia Koval <julia.koval@intel.com>
5689 Sebastian Peryt <sebastian.peryt@intel.com>
5691 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
5692 c-family/c-cilkplus.o, c-family/array-notation-common.o,
5693 cilk-common.o, cilk.h, cilk-common.c): Remove.
5695 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
5696 * builtins.c (is_builtin_name): Remove cilkplus condition.
5697 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
5698 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
5699 cilk-builtins.def, cilkplus.def): Remove.
5700 * cif-code.def (CILK_SPAWN): Remove.
5701 * cilk-builtins.def: Delete.
5702 * cilk-common.c: Ditto.
5704 * cilkplus.def: Ditto.
5705 * config/darwin.h (fcilkplus): Delete.
5706 * cppbuiltin.c: Ditto.
5707 * doc/extend.texi: Remove cilkplus doc.
5708 * doc/generic.texi: Ditto.
5709 * doc/invoke.texi: Ditto.
5710 * doc/passes.texi: Ditto.
5711 * gcc.c (fcilkplus): Remove.
5712 * gengtype.c (cilk.h): Remove.
5713 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
5715 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
5717 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
5719 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
5720 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
5721 cilkplus conditions.
5722 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
5723 inline_read_section): Ditto.
5724 * ipa-inline-analysis.c (cilk.h): Remove.
5725 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
5726 * lto-wrapper.c (merge_and_complain, append_compiler_options,
5727 append_linker_options): Remove condition for fcilkplus.
5728 * lto/lto-lang.c (cilk.h): Remove.
5729 (lto_init): Remove condition for fcilkplus.
5730 * omp-expand.c (expand_cilk_for_call): Delete.
5731 (expand_omp_taskreg, expand_omp_for_static_chunk,
5732 expand_omp_for): Remove cilkplus
5734 (expand_cilk_for): Delete.
5735 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
5736 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
5737 execute_lower_omp, diagnose_sb_0): Ditto.
5738 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
5739 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5740 * tree-nested.c: Ditto.
5741 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
5742 (dump_generic_node): Ditto.
5743 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5744 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
5746 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
5748 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5750 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
5751 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
5752 for vector mode and !TARGET_SIMD.
5754 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5756 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
5758 * tree.h (SWITCH_LABELS): Remove.
5759 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
5760 assert SWITCH_BODY is non-NULL.
5761 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
5763 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
5765 PR tree-optimization/80788
5766 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
5767 has TREE_OVERFLOW set, call drop_tree_overflow.
5769 2017-11-28 Richard Biener <rguenther@suse.de>
5771 PR tree-optimization/83158
5772 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
5774 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5777 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5778 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5780 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5782 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5783 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5785 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5788 * varasm.c (compare_constant): Take the mode of the constants into
5789 account when comparing floating point constants.
5791 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5793 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5794 from explicit instantiation of debug_helper.
5795 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5797 2017-11-27 Richard Biener <rguenther@suse.de>
5799 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5802 2017-11-27 Richard Biener <rguenther@suse.de>
5804 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5805 (type_hash_canon): Also clear int_cst_hash_table entry for
5807 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5809 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5811 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5812 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5813 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5815 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5817 * hash-map.h (gt_cleare_cache): Avoid UB.
5819 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5821 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5822 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5823 (decide_unroll_runtime_iterations): Likewise.
5824 (decide_unroll_stupid): Likewise.
5826 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5829 * config/i386/i386.md: Add a loop with incssp.
5831 2017-11-27 Martin Jambor <mjambor@suse.cz>
5833 PR tree-optimization/81248
5834 * tree-sra.c (splice_param_accesses): Remove size check.
5835 (decide_one_param_reduction): Fix size check.
5836 * gimple-pretty-print.c (dump_profile): Silence warning.
5837 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5839 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5842 * dbxout.c (lastlineno): New variable.
5843 (dbx_debug_hooks): Use dbxout_switch_text_section as
5844 switch_text_section debug hook.
5845 (dbxout_function_end): Switch to current_function_section
5846 rather than function_section. If crtl->has_bb_partition,
5847 output just one N_FUN, depending on in_cold_section_p.
5848 (dbxout_source_line): Remember last lineno in lastlineno.
5849 (dbxout_switch_text_section): New function.
5850 (dbxout_function_decl): Adjust dbxout_block caller.
5851 (dbx_block_with_cold_children): New function.
5852 (dbxout_block): Return true if any LBRAC/RBRAC have been
5853 emitted. Use dbx_block_with_cold_children at depth == 0
5854 in second partition. Add PARENT_BLOCKNUM argument, pass
5855 it optionally adjusted to children. Output LBRAC/RBRAC
5856 around recursive call only if the block is in the current
5857 partition, if not and anything was output, emit empty
5859 * final.c (final_scan_insn): Compute cold_function_name
5860 before calling switch_text_section debug hook. Call
5861 that hook even if dwarf2out_do_frame if not emitting
5865 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5866 TREE_READONLY decls.
5868 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5870 PR rtl-optimization/82488
5871 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5873 2017-11-26 Julia Koval <julia.koval@intel.com>
5875 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5877 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5880 2017-11-26 Julia Koval <julia.koval@intel.com>
5882 * config/i386/driver-i386.c (host_detect_local_cpu):
5883 Detect skylake-avx512.
5885 2017-11-26 Julia Koval <julia.koval@intel.com>
5887 * config.gcc: Add -march=cannonlake.
5888 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5889 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5890 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5891 (PTA_CANNONLAKE): New.
5892 (processor_target_table): Add cannonlake.
5893 (ix86_option_override_internal): Ditto.
5894 (fold_builtin_cpu): Ditto.
5895 (get_builtin_code_for_version): Handle cannonlake.
5896 (M_INTEL_COREI7_CANNONLAKE): New.
5897 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5898 * doc/invoke.texi: Add -march=cannonlake.
5900 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5902 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5903 (try_init_one_plugin): Alternative implementation for MinGW.
5904 * Makefile.in (plugin_implib): New.
5905 (gengtype-lex.c): Fix broken AIX workaround.
5906 * configure: Regenerate.
5907 * doc/plugins.texi: Document support for MinGW.
5909 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5911 PR rtl-optimization/81553
5912 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5913 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5914 is a shift where C1 has different mode than the whole shift, use C1's
5915 mode for MULT rather than the shift's mode.
5918 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5919 builtins not enabled in the currently selected ISA.
5921 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5923 PR tree-optimization/71026
5924 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5925 (insert_reciprocals): Change to insert reciprocals before a division
5926 by a square and to insert the square of a reciprocal.
5927 (execute_cse_reciprocals_1): Change to consider division by a square.
5928 (register_division_in): Add importance parameter.
5930 2017-11-24 Richard Biener <rguenther@suse.de>
5932 PR tree-optimization/82402
5933 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5934 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5936 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5938 * match.pd (0-ptr): New transformation.
5940 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5943 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5944 when skipping checking.
5946 2017-11-24 Richard Biener <rguenther@suse.de>
5948 PR tree-optimization/83128
5949 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5950 (vn_reference_lookup_3): Likewise.
5952 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5955 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5956 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5959 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5960 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5961 check for the latter with gimple_call_builtin_p. Do not handle
5962 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5964 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5966 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5969 2017-11-23 Julia Koval <julia.koval@intel.com>
5971 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5972 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5973 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5974 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5975 _mm512_maskz_expandloadu_epi16): New intrinsics.
5976 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5977 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5978 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5979 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5980 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5981 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5982 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5983 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5984 _mm256_maskz_expandloadu_epi8): New intrinsics.
5985 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5986 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5987 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5988 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5989 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5990 * config/i386/sse.md (VI248_VLBW): New iterator.
5991 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5993 2017-11-23 Julia Koval <julia.koval@intel.com>
5995 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5996 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5997 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5998 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5999 _mm512_mask_compressstoreu_epi16): New.
6000 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
6001 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
6002 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
6003 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
6004 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
6005 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
6006 _mm256_mask_compressstoreu_epi8): New.
6007 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
6008 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
6009 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
6010 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
6011 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
6012 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
6013 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
6014 __builtin_ia32_compresshi128_mask,
6015 __builtin_ia32_compressstoreuqi512_mask,
6016 __builtin_ia32_compressstoreuhi512_mask,
6017 __builtin_ia32_compressstoreuqi256_mask,
6018 __builtin_ia32_compressstoreuqi128_mask,
6019 __builtin_ia32_compressstoreuhi256_mask,
6020 __builtin_ia32_compressstoreuhi128_mask): New builtins.
6021 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
6023 (ix86_expand_special_args_builtin): Handle new types.
6024 (s4fma_expand): Handle new builtin array.
6025 * config/i386/immintrin.h: Include new headers.
6026 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
6027 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
6029 2017-11-23 Jakub Jelinek <jakub@redhat.com>
6031 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
6032 fixes. Declare temp and g variables at the top in order to avoid
6033 {} in most of the cases.
6035 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
6037 * match.pd (ptr-0): New transformation.
6039 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
6041 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
6042 (struct addr_mode_cost_table): New.
6043 (struct tune_params): Add field addr_mode_costs.
6044 * config/arm/arm.c (generic_addr_mode_costs): New.
6045 (arm_slowmul_tune): Initialise addr_mode_costs field.
6046 (arm_fastmul_tune): Likewise.
6047 (arm_strongarm_tune): Likewise.
6048 (arm_xscale_tune): Likewise.
6049 (arm_9e_tune): Likewise.
6050 (arm_marvell_pj4_tune): Likewise.
6051 (arm_v6t2_tune): Likewise.
6052 (arm_cortex_tune): Likewise.
6053 (arm_cortex_a8_tune): Likewise.
6054 (arm_cortex_a7_tune): Likewise.
6055 (arm_cortex_a15_tune): Likewise.
6056 (arm_cortex_a35_tune): Likewise.
6057 (arm_cortex_a53_tune): Likewise.
6058 (arm_cortex_a57_tune): Likewise.
6059 (arm_exynosm1_tune): Likewise.
6060 (arm_xgene1_tune): Likewise.
6061 (arm_cortex_a5_tune): Likewise.
6062 (arm_cortex_a9_tune): Likewise.
6063 (arm_cortex_a12_tune): Likewise.
6064 (arm_cortex_a73_tune): Likewise.
6065 (arm_v7m_tune): Likewise.
6066 (arm_cortex_m7_tune): Likewise.
6067 (arm_v6m_tune): Likewise.
6068 (arm_fa726te_tune): Likewise.
6069 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
6071 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
6073 * config/arm/arm.c (arm_mem_costs): New function.
6074 (arm_rtx_costs_internal): Use arm_mem_costs.
6076 2017-11-23 Mark Wielaard <mark@klomp.org>
6078 * dwarf2out.c (init_sections_and_labels): Use generation to create
6079 unique ranges_section_label and ranges_base_label. Return generation.
6080 (output_rnglists): Add generation argument. Use generation to create
6081 unique ranges labels.
6082 (dwarf2out_finish): Get generation from init_sections_and_labels
6083 and pass generation to output_rnglists.
6085 2017-11-23 Mike Stump <mikestump@comcast.net>
6086 Eric Botcazou <ebotcazou@adacore.com>
6088 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
6089 * cfgloop.h (struct loop): Add unroll field.
6090 * function.h (struct function): Add has_unroll bitfield.
6091 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
6092 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
6093 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
6094 (pass_rtl_unroll_loops::gate): Likewise.
6095 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
6096 for which loop->unroll==1.
6097 (decide_unroll_constant_iterations): Use note for consistency and
6098 take loop->unroll into account. Return early if loop->unroll is set.
6099 Fix thinko in existing test.
6100 (decide_unroll_runtime_iterations): Use note for consistency and
6101 take loop->unroll into account.
6102 (decide_unroll_stupid): Likewise.
6103 * lto-streamer-in.c (input_cfg): Read loop->unroll.
6104 * lto-streamer-out.c (output_cfg): Write loop->unroll.
6105 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
6107 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
6108 (print_loop): Print loop->unroll if set.
6109 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
6110 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
6111 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
6113 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
6114 loop->unroll is set and smaller than the trip count. Otherwise bypass
6115 entirely the heuristics if loop->unroll is set. Remove dead note.
6116 Fix off-by-one bug in other note.
6117 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
6118 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
6120 (tree_unroll_loops_completely): Make static.
6121 (pass_complete_unroll::execute): Use correct type for variable.
6122 (pass_complete_unrolli::execute): Fix formatting.
6123 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
6125 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
6127 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
6128 enable when TARGET_PREFER_AVX128 is set.
6130 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6132 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
6134 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
6135 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
6137 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6139 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
6140 frequencies. Fix estimation of aggregate parameters.
6142 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6144 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
6147 2017-11-23 Tom de Vries <tom@codesourcery.com>
6149 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
6151 2017-11-23 Tom de Vries <tom@codesourcery.com>
6153 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
6156 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
6159 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
6160 sibcall_value_pcrel_fdpic): Use local variable instead of
6162 (calli_tbr_rel): Add missing operand 2.
6163 (call_valuei_tbr_rel): Add missing operand 3.
6165 2017-11-23 Jakub Jelinek <jakub@redhat.com>
6168 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
6169 bitpos/bitsize covering the whole destination, use store_expr only if
6170 the complex mode is the same. Otherwise, use expand_normal and if
6171 it returns CONCAT, subreg each part separately instead of trying to
6172 subreg the whole result.
6174 2017-11-23 Richard Biener <rguenther@suse.de>
6176 PR tree-optimization/23094
6177 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
6178 come here from walking over backedges in the first iteration.
6179 (vn_reference_lookup_3): Skip clobbers that store the same value.
6181 2017-11-23 Richard Biener <rguenther@suse.de>
6183 PR tree-optimization/81403
6184 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
6185 a block we need a leader relative to.
6186 (phi_translate_1): For nary processing require a leader from
6187 get_representative_for given we run expression simplification
6188 using match-and-simplify. Remove previous fix.
6190 2017-11-22 Jeff Law <law@redhat.com>
6192 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
6193 Use new method allocate_value_range rather than accessing the
6194 vrp_value_range_pool data member directly.
6195 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
6196 to use extract_range_from_stmt method to avoid need for
6197 extract_range_from_assignment method.
6198 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
6199 method rather than setting values_propgated data member directly.
6200 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
6201 and values propagated data members and extract_range_from_assignment
6202 method. Reorder private data members to conform to standards.
6203 Add new methods set_lattice_propagation_complete and
6204 allocate_value_range.
6206 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
6208 PR rtl-optimization/83030
6209 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
6210 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
6211 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
6212 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
6213 CROSSING_JUMP_P flag is already set before setting it.
6214 * cfgrtl.c (fixup_partition_crossing): Likewise.
6215 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
6218 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6220 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
6221 Handle the case where both arguments are using gen_const_vec_series.
6223 2017-11-22 David Malcolm <dmalcolm@redhat.com>
6226 * pretty-print.c (pp_format): Move quoting implementation to
6227 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
6228 to pass address of "quote" local.
6229 (pp_begin_quote): New function.
6230 (pp_end_quote): New function.
6231 * pretty-print.h (printer_fn): Convert penultimate param from bool
6233 (pp_begin_quote): New decl.
6234 (pp_end_quote): New decl.
6235 * tree-diagnostic.c (default_tree_printer): Convert penultimate
6236 param from bool to bool *.
6237 * tree-diagnostic.h (default_tree_printer): Likewise.
6239 2017-11-22 Jeff Law <law@redhat.com>
6241 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
6242 Initialize vr_values.
6243 (evrp_range_analyzer::try_find_new_range): Call methods attached to
6244 vr_values via vr_values class instance rather than delegators.
6245 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
6246 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
6247 (evrp_range_analyzer::push_value_range): Likewise.
6248 (evrp_range_analyzer::pop_value_range): Likewise.
6249 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
6250 most delegators. Those remaining are exposed as public interfaces.
6251 Make vr_values a pointer and private.
6252 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
6254 (evrp_range_analyzer::get_vr_value): New method.
6255 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
6256 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
6257 (class evrp_dom_walker): Attach evrp_folder class, initialize
6258 it in the ctor. Remove temporary delegators.
6259 (evrp_dom_walker::before_dom_children): Call methods in attached
6260 evrp_range_analyzer class via class instance pointer. Use
6261 free value_range_constant_singleton to remove need for
6262 op_with_constant_singleton_value delegator method. Do not
6263 create a vrp_prop class instance for every call! Narrow
6264 scope of a couple variables.
6265 (evrp_dom_walker::cleanup): Call methods in attached
6266 evrp_range_analyzer class via class instance pointer.
6267 * vr-values.h (class vr_values): Privatize many methods and
6270 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6272 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
6273 * cfgexpand.c (expand_debug_expr): Remove handling for them.
6274 * expr.c (expand_expr_real_2): Likewise.
6275 * fold-const.c (const_unop): Likewise.
6276 * optabs-tree.c (optab_for_tree_code): Likewise.
6277 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
6278 * tree-inline.c (estimate_operator_cost): Likewise.
6279 * tree-pretty-print.c (dump_generic_node): Likewise.
6280 (op_code_prio): Likewise.
6281 (op_symbol_code): Likewise.
6282 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
6283 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
6284 * internal-fn.c (direct_internal_fn_optab): New function.
6285 (direct_internal_fn_array, direct_internal_fn_supported_p
6286 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
6287 * fold-const-call.c (fold_const_reduction): New function.
6288 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
6290 * tree-vect-loop.c: Include internal-fn.h.
6291 (reduction_code_for_scalar_code): Rename to...
6292 (reduction_fn_for_scalar_code): ...this and return an internal
6294 (vect_model_reduction_cost): Take an internal_fn rather than
6296 (vect_create_epilog_for_reduction): Likewise. Build calls rather
6298 (vectorizable_reduction): Use internal functions rather than tree
6299 codes for the reduction operation. Update calls to the functions
6301 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
6302 Use calls to internal functions rather than REDUC tree codes.
6303 * config/aarch64/aarch64-simd.md: Update comment accordingly.
6305 2017-11-22 Olivier Hainque <hainque@adacore.com>
6307 * config/vxworks.c (vxworks_override_options): Pick default
6308 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
6309 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
6310 DWARF_GNAT_ENCODINGS_DEFAULT.
6311 * config/vxworksae.h: Likewise.
6313 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
6315 PR tree-optimization/83104
6316 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
6319 2017-11-22 Marek Polacek <polacek@redhat.com>
6320 H.J. Lu <hongjiu.lu@intel.com>
6321 Jason Merrill <jason@redhat.com>
6326 * calls.c (initialize_argument_information): Call
6327 warn_parameter_passing_abi target hook.
6328 (store_one_arg): Use 0 for empty record size. Don't push 0 size
6329 argument onto stack.
6330 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
6331 * common.opt: Update -fabi-version description.
6332 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
6333 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
6335 (ix86_is_empty_record): New function.
6336 (ix86_warn_parameter_passing_abi): New function.
6337 (TARGET_EMPTY_RECORD_P): Redefine.
6338 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
6339 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
6340 * doc/tm.texi: Regenerated.
6341 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
6342 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
6343 * dwarf2out.c (get_ultimate_context): Move to tree.c.
6344 * explow.c (hard_function_value): Call arg_int_size_in_bytes
6345 instead of int_size_in_bytes.
6346 * expr.c (copy_blkmode_to_reg): Likewise.
6347 * function.c (aggregate_value_p): Return 0 for empty types.
6348 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
6350 (locate_and_pad_parm): Call arg size_in_bytes instead
6352 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
6353 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
6354 * target.def (empty_record_p, warn_parameter_passing_abi): New target
6356 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
6357 (std_gimplify_va_arg_expr): Skip empty records. Call
6358 arg_size_in_bytes instead size_in_bytes.
6359 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
6360 * tree-core.h (tree_type_common): Add empty_flag.
6361 (tree_decl_common): Update comments.
6362 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
6364 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6365 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
6367 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6368 * tree.c (default_is_empty_type): New function.
6369 (default_is_empty_record): New function.
6370 (arg_int_size_in_bytes): New function.
6371 (arg_size_in_bytes): New function.
6372 (get_ultimate_context): New function.
6373 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
6374 TRANSLATION_UNIT_WARN_EMPTY_P.
6375 (default_is_empty_record, arg_int_size_in_bytes,
6376 arg_size_in_bytes, get_ultimate_context): Declare.
6378 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6380 * config/arm/arm.c (cmse_clear_registers): New function.
6381 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
6382 code by call to cmse_clear_registers.
6383 (cmse_nonsecure_entry_clear_before_return): Likewise.
6385 2017-11-22 Tamar Christina <tamar.christina@arm.com>
6387 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
6388 (vdot_s32, vdotq_s32): New.
6389 (vdot_lane_u32, vdotq_lane_u32): New.
6390 (vdot_lane_s32, vdotq_lane_s32): New.
6393 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6396 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
6397 for unsigned values with fewer HWIs than the precision.
6398 (test_overflow): New function.
6399 (wide_int_cc_tests): Call it.
6401 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6403 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
6404 has the same mode class as Pmode.
6406 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6408 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
6409 padding_bits_to_clear_ptr.
6410 (cmse_nonsecure_entry_clear_before_return): Likewise.
6412 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6414 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
6415 auto_sbitap instead of integer bitfield to control register needing
6418 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6420 PR tree-optimization/83044
6421 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
6422 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
6423 that need the upper bound. Subtract offset from
6424 get_addr_base_and_unit_offset only if positive and subtract it
6425 before division by eltsize rather than after it.
6428 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
6429 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
6430 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
6434 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
6435 expand_binop with *mul_widen_optab, make sure at least one of the
6436 operands doesn't have VOIDmode.
6439 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
6441 PR rtl-optimization/82044
6442 PR tree-optimization/82042
6443 * dse.c (record_store): Check for overflow.
6444 (check_mem_read_rtx): Properly check for overflow if width == -1, call
6445 add_wild_read instead of clear_rhs_from_active_local_stores on
6446 overflow and log it into dump_file.
6448 2017-11-22 Richard Biener <rguenther@suse.de>
6450 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
6451 fake edges to exit when looking for a place to insert.
6452 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
6454 (insert_into_preds_of_block): Commit edge insertion immediately,
6455 assert that doesn't require new BBs.
6456 (fini_pre): Release expressions.
6457 (pass_pre::execute): Shuffle things around a bit, if the fn
6458 is too large do not compute AVAIL either as this is really the
6461 2017-11-22 Richard Biener <rguenther@suse.de>
6463 PR tree-optimization/83089
6464 * tree-if-conv.c (pass_if_conversion::execute): If anything
6465 changed reset SCEV and free the number of iteration estimates.
6467 2017-11-21 Martin Sebor <msebor@redhat.com>
6469 PR tree-optimization/82945
6470 * calls.h (warn_nonstring_bound): Remove unused function.
6472 2017-11-21 Martin Sebor <msebor@redhat.com>
6474 PR tree-optimization/82945
6475 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
6476 * calls.h (maybe_warn_nonstring_arg): Declare new function.
6477 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
6479 (initialize_argument_information): Call maybe_warn_nonstring_arg.
6480 * calls.h (get_attr_nonstring_decl): Declare new function.
6481 * doc/extend.texi (attribute nonstring): Update.
6482 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
6483 get_attr_nonstring_decl and handle it.
6484 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
6485 detection of nul-termination.
6486 (strlen_to_stridx): Change to a pointer.
6487 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
6488 (pass_strlen::execute): Same.
6490 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
6492 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
6493 for the new option -mprefer-vector-width=[none|128|256|512].
6494 * config/i386/i386.c (ix86_target_string): remove old style options
6495 -mprefer-avx256 and make -mprefer-avx128 as alias.
6496 (ix86_option_override_internal): Apply defaults for the
6497 -mprefer-vector-width=[128|256] option.
6498 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
6499 Implement macros to work with -mprefer-vector-width=.
6500 * config/i386/i386.opt: Implemented option
6501 -mprefer-vector-width=[none|128|256|512].
6502 * doc/invoke.texi: Documentation for
6503 -mprefer-vector-width=[none|128|256|512].
6505 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
6507 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
6509 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
6511 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
6513 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
6515 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
6517 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
6518 * expr.c (expand_expr_real_2): Likewise.
6519 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
6520 fold_binary_loc): Likewise.
6521 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
6522 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
6523 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
6524 MINUS_EXPR transformations.
6525 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
6526 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
6527 * tree-inline.c (estimate_operator_cost): Likewise.
6528 * tree-pretty-print.c (dump_generic_node, op_code_prio,
6529 op_symbol_code): Likewise.
6530 * tree-vect-stmts.c (vectorizable_operation): Likewise.
6531 * vr-values.c (extract_range_from_binary_expr): Likewise.
6532 * varasm.c (initializer_constant_valid_p_1): Likewise.
6533 * tree.def: New tree code POINTER_DIFF_EXPR.
6535 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
6537 * config/i386/i386.md (*bswap<mode>2_movbe): Add
6538 integer suffix to movbe mnemonic.
6539 (*bswaphi2_movbe): Ditto.
6540 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
6542 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6545 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
6546 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
6547 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
6549 2017-11-21 Martin Liska <mliska@suse.cz>
6551 * tree-inline.c (expand_call_inline): Remove not needed
6554 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
6555 Jakub Jelinek <jakub@redhat.com>
6558 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
6559 Remove static keyword from f variable.
6561 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6563 PR tree-optimization/83086
6564 * gimple-ssa-store-merging.c
6565 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
6566 rather than n.base_addr.
6568 2017-11-21 Martin Liska <mliska@suse.cz>
6570 PR rtl-optimization/82044
6571 PR tree-optimization/82042
6572 * dse.c (check_mem_read_rtx): Check for overflow.
6574 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6576 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
6579 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6581 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
6582 and its corresponding call arg location note.
6584 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6585 Andrew Burgess <andrew.burgess@embecosm.com>
6587 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
6589 (arc_return_slot_offset): Likewise.
6590 (arc_eh_return_address_location): New declaration.
6591 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
6592 (MUST_SAVE_REGISTER): Add exception handler case.
6593 (MUST_SAVE_RETURN_ADDR): Likewise.
6594 (arc_frame_pointer_required): Likewise.
6595 (arc_frame_pointer_needed): New function.
6596 (arc_compute_frame_size): Changed.
6597 (arc_expand_prologue): Likewise.
6598 (arc_expand_epilogue): Likewise.
6599 (arc_initial_elimination_offset): Likewise.
6600 (arc_return_slot_offset): Delete.
6601 (arc_eh_return_address_location): New function.
6602 (arc_builtin_setjmp_frame_value): Likewise.
6603 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
6604 (EH_RETURN_STACKADJ_RTX): Define.
6605 (EH_RETURN_HANDLER_RTX): Likewise.
6606 * config/arc/arc.md (eh_return): Delete.
6608 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6610 * print-tree.h (debug_vec_tree): Remove prototype.
6611 * gdbinit.in (pvt): Remove macro.
6613 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6615 PR tree-optimization/83047
6616 * gimple-ssa-store-merging.c
6617 (imm_store_chain_info::output_merged_store): If the loads with the
6618 same vuse are in different basic blocks, for load_gsi pick a load
6619 location that is dominated by the other loads.
6622 * config/i386/i386.c (ix86_memmodel_check): Start
6623 -Winvalid-memory-model diagnostics with lowercase letter.
6626 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
6627 set in_cold_section_p to the partition containing loc_list->first.
6628 When seeing loc_list->last_before_switch node, update secname and
6629 perform range_across_switch second partition handling only after that.
6632 * run-rtl-passes.c: Include debug.h.
6633 (run_rtl_passes): Call debug_hooks->assembly_start.
6634 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
6638 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
6639 OPTAB_DIRECT in calls to expand_simple_binop.
6641 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6644 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
6646 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6649 * spellcheck.h (best_match::blithely_get_best_candidate): New
6652 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6654 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
6655 with lower case letter.
6657 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
6659 * config/i386/i386.md (bswaphi2): New expander.
6660 (*bswaphi2_movbe): New insn pattern.
6661 (bswaphi -> rorhi pepehole2): New peephole pattern.
6663 2017-11-20 Jeff Law <law@redhat.com>
6665 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
6666 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
6667 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
6668 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
6669 gimple-ssa-evrp-analyze.h.
6671 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
6672 set BB_VISITED here.
6673 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
6675 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6677 * tree-predcom.c: Add general comment on Store-Store chains.
6678 (split_data_refs_to_components): Postpone clearing eliminate_store_p
6680 (get_chain_last_ref_at): Rename into...
6681 (get_chain_last_write_at): ...this.
6682 (get_chain_last_write_before_load): New function.
6683 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
6684 CT_STORE_STORE when write reference is added.
6685 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
6686 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
6687 (initialize_root_vars_store_elim_1): Ditto.
6688 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
6689 definition is created.
6690 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
6691 chain by replacing it with dominant stored value.
6693 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6695 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
6697 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
6699 * vr-values.c (extract_range_from_binary_expr): Use a full range
6702 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
6704 * config/arm/arm.md (R4_REGNUM): Define constant.
6705 (nonsecure_call_internal): Remove r4 clobber.
6706 (nonsecure_call_value_internal): Likewise.
6707 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
6708 clobber and resequence match_operands.
6709 (nonsecure_call_value_reg_thumb1_v5): Likewise.
6710 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
6711 (nonsecure_call_value_reg_thumb2): Likewise.
6713 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6715 PR tree-optimization/78821
6716 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
6717 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
6718 to the address of the base rather than the base itself.
6719 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
6720 (find_bswap_or_nop_finalize): New function.
6721 (find_bswap_or_nop): Use it.
6722 (bswap_replace): Return a tree rather than bool, change first
6723 argument from gimple * to gimple_stmt_iterator, allow inserting
6724 into an empty sequence, allow ins_stmt to be NULL - then emit
6725 all stmts into gsi. Fix up MEM_REF address gimplification.
6726 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
6727 (struct store_immediate_info): Add N and INS_STMT non-static
6729 (store_immediate_info::store_immediate_info): Initialize them
6730 from newly added ctor args.
6731 (merged_store_group::apply_stores): Formatting fixes. Sort by
6733 (stmts_may_clobber_ref_p): For stores call also
6734 refs_anti_dependent_p.
6735 (gather_bswap_load_refs): New function.
6736 (imm_store_chain_info::try_coalesce_bswap): New method.
6737 (imm_store_chain_info::coalesce_immediate_stores): Use it.
6738 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
6739 (imm_store_chain_info::output_merged_store): Fail if number of
6740 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
6741 and NOP_EXPR rhs_code.
6742 (pass_store_merging::process_store): Compute n and ins_stmt, if
6743 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
6744 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
6746 (pass_store_merging::execute): Calculate dominators.
6748 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
6749 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
6750 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
6751 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6752 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
6753 class pass_optimize_bswap, bswap_replace,
6754 pass_optimize_bswap::execute): Moved to ...
6755 * gimple-ssa-store-merging.c: ... this file.
6756 Include optabs-tree.h.
6757 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
6758 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6759 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
6760 anonymous namespace, remove static keywords.
6761 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
6762 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
6764 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
6767 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6769 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
6771 * vec.h (debug_helper): New function.
6772 (DEFINE_DEBUG_VEC): New macro.
6773 * hash-set.h (debug_helper): New function.
6774 (DEFINE_DEBUG_HASH_SET): New macro.
6775 * cfg.c (debug_slim (edge)): New function.
6776 Call DEFINE_DEBUG_VEC for edges.
6777 Call DEFINE_DEBUG_HASH_SET for edges.
6778 * cfghooks.c (debug_slim (basic_block)): New function.
6779 Call DEFINE_DEBUG_VEC for basic blocks.
6780 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6781 * print-tree.c (debug_slim): New function to handle trees.
6782 Call DEFINE_DEBUG_VEC for trees.
6783 Call DEFINE_DEBUG_HASH_SET for trees.
6784 (debug (vec<tree, va_gc>) &): Remove.
6785 (debug (<vec<tree, va_gc>) *): Remove.
6786 * print-rtl.c (debug_slim): New function to handle const_rtx.
6787 Call DEFINE_DEBUG_VEC for rtx_def.
6788 Call DEFINE_DEBUG_VEC for rtx_insn.
6789 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6790 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6791 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6792 (debug (vec<rtx_insn *> *ptr): Remove.
6793 (debug_insn_vector): Remove.
6794 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6796 2017-11-20 Tom de Vries <tom@codesourcery.com>
6798 PR rtl-optimization/82020
6799 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6800 IF_THEN_ELSE condition.
6802 2017-11-19 Jeff Law <law@redhat.com>
6804 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6805 of degenerates resulting from ignoring an edge.
6807 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6810 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6812 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6815 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6816 for uninitialied counts.
6818 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6821 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6823 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6826 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6828 2017-11-19 Tom de Vries <tom@codesourcery.com>
6830 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6833 2017-11-19 Tom de Vries <tom@codesourcery.com>
6835 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6838 2017-11-19 Tom de Vries <tom@codesourcery.com>
6840 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6843 2017-11-19 Tom de Vries <tom@codesourcery.com>
6845 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6847 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6849 2017-11-19 Tom de Vries <tom@codesourcery.com>
6851 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6853 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6854 semicolon after MCORE_EXPORT_NAME call.
6856 2017-11-19 Tom de Vries <tom@codesourcery.com>
6859 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6860 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6862 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6864 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6865 (cgraph_edge::clone): Cleanup updating of profile.
6866 * ipa-cp.c (update_profiling_info): Likewise.
6867 * ipa-inline-transform.c (inline_transform): Likewise.
6868 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6869 * ipa-split.c (execute_split_functions): Do not split when function
6871 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6872 * profile-count.c (profile_count::dump): Add global0.
6873 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6875 (profile_count::to_sreal_scale): Likewise.
6876 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6877 (profile_count::combine_with_ipa_count): New function.
6878 * profile-count.h (profile_guessed_global0adjusted): New.
6879 (profile_count::adjusted_zero): New.
6880 (profile_count::global0adjusted): New.
6881 (profile_count::combine_with_ipa_count): New.
6882 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6883 correct profile of return block of split functions.
6884 (copy_cfg_body): Remove unused profile_count.
6885 (copy_body): Likewise.
6886 (expand_call_inline): Update.
6887 (tree_function_versioning): Update.
6889 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6891 * hash-set.h (hash_set::empty): New.
6892 * tree-ssa-threadbackward.h: Delete.
6893 * tree-ssa-threadbackward.c (class thread_jumps): New.
6894 Move max_threaded_paths into class.
6895 (fsm_find_thread_path): Remove arguments that are now in class.
6896 (profitable_jump_thread_path): Rename to...
6897 (thread_jumps::profitable_jump_thread_path): ...this.
6898 (convert_and_register_jump_thread_path): Rename to...
6899 (thread_jumps::convert_and_register_current_path): ...this.
6900 (check_subpath_and_update_thread_path): Rename to...
6901 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6902 (register_jump_thread_path_if_profitable): Rename to...
6903 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6904 (handle_phi): Rename to...
6905 (thread_jumps::handle_phi): ...this.
6906 (handle_assignment): Rename to...
6907 (thread_jumps::handle_assignment): ...this.
6908 (fsm_find_control_statement_thread_paths): Rename to...
6909 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6910 (find_jump_threads_backwards): Rename to...
6911 (thread_jumps::find_jump_threads_backwards): ...this.
6912 Initialize path local data.
6913 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6914 from within thread_jumps class.
6915 (pass_early_thread_jumps::execute): Same.
6917 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6919 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6921 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6923 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6924 to canonical location.
6926 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6929 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6931 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6933 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6935 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6936 rcrt1.o%s/grcrt1.o%s for -static-pie.
6938 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6940 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6941 ix86_shift_rotate_cost): Break out from ...
6942 (ix86_rtx_costs): ... here.
6943 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6946 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6948 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6949 when entry block was promoted unlikely.
6950 (estimate_bb_frequencies): Increase frequency scale.
6951 * profile-count.h (profile_count): Export precision info.
6953 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6955 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6956 disturbin profile of entry block.
6958 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6960 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6963 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6965 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6966 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6968 2017-11-17 Jeff Law <law@redhat.com>
6970 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6971 from evrp_dom_walker class. Various methods moved into new class.
6972 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6973 (evrp_range_analyzer::enter): New method.
6974 (evrp_range_analyzer::leave): New method.
6975 (evrp_dom_walker): Remove delegators no longer needed by this class.
6976 Replace vr_values data member with evrp_range_analyzer
6978 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6979 method extracted from evrp_dom_walker::before_dom_children.
6980 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6981 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6983 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6984 Add private copy constructor and move assignment operators.
6985 Privatize methods and class data where trivially possible.
6986 (evrp_dom_walker::cleanup): New function, extracted from
6987 execute_early_vrp. Simplify access to class data.
6989 * vr-values.h (get_output_for_vrp): Prototype.
6990 * vr-values.c (get_output_for_vrp): New function extracted from
6991 vrp_visit_assignment_or_call and extract_range_from_stmt.
6992 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6994 2017-11-17 Luis Machado <luis.machado@linaro.org>
6996 * config/aarch64/aarch64.c
6997 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6998 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6999 tune_params::AUTOPREFETCHER_WEAK.
7001 2017-11-17 Tamar Christina <tamar.christina@arm.com>
7004 * config/arm/arm.c (arm_valid_target_attribute_rec):
7005 Parse "arch=" and "+<ext>".
7006 (arm_valid_target_attribute_tree): Re-init global options.
7007 (arm_option_override): Make non-static.
7008 (arm_options_perform_arch_sanity_checks): Make errors fatal.
7009 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
7010 (__ARM_FEATURE_CRC32): Support undef.
7011 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
7012 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
7014 2017-11-17 David Malcolm <dmalcolm@redhat.com>
7016 * gdbinit.in (break-on-diagnostic): New command.
7018 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7020 * config/i386/i386.c (ix86_expand_epilogue): Change simple
7021 return to indirect jump for EH return if control-flow
7022 protection is enabled. Change explicit 'false' argument in
7023 pro_epilogue_adjust_stack with a value of
7025 * config/i386/i386.md (simple_return_indirect_internal):
7026 Remove SImode restriction to support 64-bit.
7028 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
7030 * combine.c (added_notes_insn): New.
7031 (try_combine): Handle added_notes_insn like added_links_insn.
7032 Rewrite return value code.
7033 (distribute_notes): Set added_notes_insn to the earliest insn we added
7036 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
7038 PR rtl-optimization/82621
7039 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
7040 dest of one of those SETs is unused.
7042 2017-11-17 Richard Biener <rguenther@suse.de>
7045 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
7046 * tree-pretty-print.c (dump_generic_node): Handle
7047 annot_expr_parallel_kind.
7048 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
7049 * gimplify.c (gimple_boolify): Likewise.
7051 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7053 * config.gcc (extra_headers): Add cet.h for x86 targets.
7054 * config/i386/cet.h: New file.
7055 * doc/install.texi: Add --enable-cet/--disable-cet.
7057 2017-11-17 Richard Biener <rguenther@suse.de>
7059 PR tree-optimization/83017
7060 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
7061 (gen_parallel_loop): Properly count iterations.
7062 (parallelize_loops): Handle loop->can_be_parallel independent
7063 of flag_loop_parallelize_all. Make static profitability test match
7065 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
7066 * invoke.texi (parloops-min-per-thread): Document.
7068 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
7070 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
7071 upstreaming review comments.
7073 2017-11-17 Tamar Christina <tamar.christina@arm.com>
7075 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
7076 with fast unaligned access.
7077 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
7079 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
7081 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
7082 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
7084 2017-11-17 Richard Biener <rguenther@suse.de>
7086 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
7087 folding of references.
7089 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
7092 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
7093 of replacing call to strncmp with corresponding call to strcmp when
7096 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
7098 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
7099 option prefer-avx256 for skylake-avx512 configuration.
7100 * config/i386/i386.c (ix86_option_override_internal): Ditto.
7101 (get_builtin_code_for_version): Ditto.
7103 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
7104 Monk Chiang <sh.chiang04@gmail.com>
7106 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
7107 (FIXED_REGISTERS): Reserve more register numbers.
7108 (CALL_USED_REGISTERS): Likewise.
7109 (REG_ALLOC_ORDER): Likewise.
7110 (REG_CLASS_CONTENTS): Likewise.
7111 (REGISTER_NAMES): Likewise.
7113 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
7114 Kito Cheng <kito.cheng@gmail.com>
7116 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
7118 * config/nds32/iterators.md: Add vector mode iterators and attributes.
7120 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
7122 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
7123 parameter list for vec_splats.
7125 2017-11-16 Joseph Myers <joseph@codesourcery.com>
7127 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
7129 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
7131 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
7134 * doc/install.texi: Document incompatibility of Perl >=5.6.26
7135 with the required version of automake 1.11.6.
7137 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
7139 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
7141 (DU_C2_3_power9): Correct reservation combinations.
7142 (FP_DIV_power9, VEC_DIV_power9): New.
7143 (power9-alu): Split out rotate/shift...
7144 (power9-rot): ...to here, correct dispatch resource.
7145 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
7147 (power9-fp): Correct latency.
7148 (power9-sdiv): Add div/sqrt resource.
7149 (power9-ddiv): Correct latency, add div/sqrt resource.
7150 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
7151 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
7153 (power9-qpdiv, power9-qpmul): Adjust resource usage.
7155 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
7157 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
7158 switch statement mapping KF built-ins to TF built-ins if we don't
7159 have the proper ISA 3.0 assembler support.
7161 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7163 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
7164 (lower_emutls_function_body): Do not compute it.
7166 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7168 * ipa-split.c (split_bb_info): Turn time to sreal.
7169 (split_point): Likewise.
7170 (dump_split_point): Likewise.
7171 (fine_split_points): Likewise.
7172 (execute_split_functions): Only zero split_bbs; turn time to sreals.
7174 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7176 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
7179 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7181 * predict.c (combine_predictions_for_bb): Preserve zero predicted
7183 (expensive_function_p): Remove useless assert.
7184 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
7186 2017-11-16 Martin Sebor <msebor@redhat.com>
7188 PR tree-optimization/82588
7189 PR tree-optimization/82583
7190 * tree-vrp.c (check_array_ref): Handle flexible array members,
7191 string literals, and inner indices.
7192 (search_for_addr_array): Add detail to diagnostics.
7194 2017-11-16 Nathan Sidwell <nathan@acm.org>
7198 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
7199 (SET_DECL_ASSEMBLER_NAME): Forward to
7200 overwrite_decl_assembler_name.
7201 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
7202 (overwrite_decl_assembler_name): Declare.
7203 * tree.c (overwrite_decl_assembler_name): New.
7204 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
7205 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
7206 (LANG_HOOKS_INITIALIZER): Add it.
7207 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
7208 * langhooks.c (lhd_set_decl_assembler_name): Use
7209 SET_DECL_ASSEMBLER_NAME.
7210 (lhd_overwrite_decl_assembler_name): Default implementation.
7212 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7213 Jackson Woodruff <jackson.woodruff@arm.com>
7215 PR tree-optimization/71026
7216 * match.pd: Canonicalize constant multiplies in division.
7218 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7220 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
7221 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
7222 Remove OPT_fomit_frame_pointer entry.
7223 * common/config/arc/arc-common.c: Likewise.
7224 * common/config/arm/arm-common.c: Likewise.
7225 * common/config/avr/avr-common.c: Likewise.
7226 * common/config/c6x/c6x-common.c: Likewise.
7227 * common/config/cr16/cr16-common.c: Likewise.
7228 * common/config/cris/cris-common.c: Likewise.
7229 * common/config/epiphany/epiphany-common.c: Likewise.
7230 * common/config/fr30/fr30-common.c: Likewise.
7231 * common/config/frv/frv-common.c: Likewise.
7232 * common/config/ia64/ia64-common.c: Likewise.
7233 * common/config/iq2000/iq2000-common.c: Likewise.
7234 * common/config/lm32/lm32-common.c: Likewise.
7235 * common/config/m32r/m32r-common.c: Likewise.
7236 * common/config/mcore/mcore-common.c: Likewise.
7237 * common/config/microblaze/microblaze-common.c: Likewise.
7238 * common/config/mips/mips-common.c: Likewise.
7239 * common/config/mmix/mmix-common.c: Likewise.
7240 * common/config/mn10300/mn10300-common.c: Likewise.
7241 * common/config/nios2/nios2-common.c: Likewise.
7242 * common/config/pa/pa-common.c: Likewise.
7243 * common/config/pdp11/pdp11-common.c: Likewise.
7244 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
7245 * common/config/riscv/riscv-common.c: Likewise.
7246 * common/config/rs6000/rs6000-common.c: Likewise.
7247 * common/config/rx/rx-common.c: Likewise.
7248 * common/config/s390/s390-common.c: Likewise.
7249 * common/config/sh/sh-common.c: Likewise.
7250 * common/config/sparc/sparc-common.c: Likewise.
7251 * common/config/tilegx/tilegx-common.c: Likewise.
7252 * common/config/tilepro/tilepro-common.c: Likewise.
7253 * common/config/v850/v850-common.c: Likewise.
7254 * common/config/visium/visium-common.c: Likewise.
7255 * common/config/xstormy16/xstormy16-common.c: Likewise.
7256 * common/config/xtensa/xtensa-common.c: Likewise.
7257 * invoke.texi (-fomit-frame-pointer): Update documentation.
7259 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7261 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
7263 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7265 * cfg.c (scale_bbs_frequencies_int,
7266 cale_bbs_frequencies_gcov_type): Remove.
7267 * cfg.h (scale_bbs_frequencies_int,
7268 cale_bbs_frequencies_gcov_type): Remove.
7270 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7272 * tree-ssa-loop-manip.c
7273 (scale_dominated_blocks_in_loop): Update to profile counts.
7274 (tree_transform_and_unroll_loop): Likewise.
7276 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7278 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
7279 scale_bbs_frequencies_int.
7281 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7283 * final.c (compute_alignments): Use counts rather than frequencies.
7285 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7287 * cfgloopanal.c: Include sreal.h
7288 (average_num_loop_insns): Use counts and sreal for accounting.
7290 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7292 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
7295 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7296 Kito Cheng <kito.cheng@gmail.com>
7298 * config/nds32/constraints.md: Provide more constraints.
7299 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
7300 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
7301 support constraints usage.
7303 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7305 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
7307 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7308 Kito Cheng <kito.cheng@gmail.com>
7310 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
7311 * config/nds32/nds32.opt: Refine the layout.
7312 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
7313 TARGET_EXT_STRING): Support new options.
7314 * config/nds32/nds32.h: Likewise.
7315 * config/nds32/nds32.md: Likewise.
7316 * config/nds32/nds32-predicates.c: Likewise.
7317 * config/nds32/constraints.md: Likewise.
7318 * common/config/nds32/nds32-common.c: Likewise.
7320 2017-11-16 Julia Koval <julia.koval@intel.com>
7323 * config/i386/gfniintrin.h: Add sse check.
7324 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
7326 2017-11-16 Julia Koval <julia.koval@intel.com>
7328 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
7329 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
7330 (ix86_handle_option): Handle -mavx512vbmi2.
7331 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
7332 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
7333 * config/i386/i386-c.c (__AVX512VBMI2__): New.
7334 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
7335 (ix86_valid_target_attribute_inner_p): Ditto.
7336 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
7337 * config/i386/i386.opt (mavx512vbmi2): New option.
7338 * doc/invoke.texi: Add new option.
7340 2017-11-16 Julia Koval <julia.koval@intel.com>
7342 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
7343 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
7344 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
7345 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
7346 _mm512_gf2p8mul_epi8): New intrinsics.
7347 * config/i386/i386-builtin-types.def
7348 (V64QI_FTYPE_V64QI_V64QI): New type.
7349 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
7350 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
7351 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
7352 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
7353 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
7354 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
7356 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
7358 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
7359 explicitly as a stream of bytes.
7361 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7363 * config/rs6000/altivec.h (vec_xst_be): New #define.
7364 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
7365 and externalize from *altivec_vperm_<mode>_internal.
7366 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
7368 (XL_BE_V8HI): Likewise.
7369 (XL_BE_V4SI): Likewise.
7370 (XL_BE_V4SI): Likewise.
7371 (XL_BE_V2DI): Likewise.
7372 (XL_BE_V4SF): Likewise.
7373 (XL_BE_V2DF): Likewise.
7374 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
7375 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
7376 all array entries with these keys: VSX_BUILTIN_VEC_XL,
7377 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
7378 VSX_BUILTIN_VEC_XST_BE.
7379 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
7380 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
7382 (altivec_init_builtins): Replace conditional calls to def_builtin
7383 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
7384 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
7385 with unconditional calls. Remove calls to def_builtin for
7386 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
7387 __builtin_vec_xst_be.
7388 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
7389 to define_expand, and add alternate RTL generation for P8.
7390 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
7391 vsx_ld_elemrev_v8hi.
7392 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
7393 add alternate RTL generation for P8.
7394 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
7395 vsx_ld_elemrev_v16qi.
7396 (vsx_st_elemrev_v8hi): Convert define_insn
7397 to define_expand, and add alternate RTL generation for P8.
7398 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
7399 vsx_st_elemrev_v8hi.
7400 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
7401 add alternate RTL generation for P8.
7402 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
7403 vsx_st_elemrev_v16qi.
7405 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
7408 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
7409 TARGET_AVX512ER check.
7410 (ix86_option_override_internal): Set MASK_VZEROUPPER if
7411 neither -mzeroupper nor -mno-zeroupper is used and
7412 TARGET_EMIT_VZEROUPPER is set.
7413 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
7414 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
7416 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
7418 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7419 folding of vector compares.
7420 (fold_build_vec_cmp): New helper function.
7421 (fold_compare_helper): New helper function.
7422 (builtin_function_type): Add compare builtins to the list of functions
7423 having unsigned arguments. Cosmetic updates to comment indentation.
7424 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
7425 the not+eq combination.
7427 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7429 PR tree-optimization/82726
7430 PR tree-optimization/70754
7431 * tree-predcom.c (order_drefs_by_pos): New function.
7432 (combine_chains): Move code setting has_max_use_after to...
7433 (try_combine_chains): ...here. New parameter. Sort combined chains
7434 according to position information.
7435 (tree_predictive_commoning_loop): Update call to above function.
7436 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
7438 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7440 PR tree-optimization/82726
7442 2017-01-23 Bin Cheng <bin.cheng@arm.com>
7444 PR tree-optimization/70754
7445 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
7446 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
7447 combined stmt before it if not NULL.
7448 (combine_chains): Process refs reversely and compute dominance point
7452 2017-02-23 Bin Cheng <bin.cheng@arm.com>
7454 PR tree-optimization/79663
7455 * tree-predcom.c (combine_chains): Process refs in reverse order
7456 only for ZERO length chains, and add explaining comment.
7458 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7460 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
7461 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
7462 * doc/invoke.texi (ARM Options): Add armv8.3-a.
7464 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7466 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
7468 2017-11-15 Martin Liska <mliska@suse.cz>
7470 * tree-cfg.c (pass_warn_function_return::execute):
7471 Compare warn_return_type for greater than zero.
7473 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
7477 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
7478 to return true on Xeon and not on Xeon Phi.
7479 (ix86_check_avx256_register): Changed to ...
7480 (ix86_check_avx_upper_register): ... this. Add extra check for
7481 VALID_AVX512F_REG_OR_XI_MODE.
7482 (ix86_avx_u128_mode_needed): Changed
7483 ix86_check_avx256_register to ix86_check_avx_upper_register.
7484 (ix86_check_avx256_stores): Changed to ...
7485 (ix86_check_avx_upper_stores): ... this. Changed
7486 ix86_check_avx256_register to ix86_check_avx_upper_register.
7487 (ix86_avx_u128_mode_after): Changed
7488 avx_reg256_found to avx_upper_reg_found. Changed
7489 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
7490 (ix86_avx_u128_mode_entry): Changed
7491 ix86_check_avx256_register to ix86_check_avx_upper_register.
7492 (ix86_avx_u128_mode_exit): Ditto.
7493 * config/i386/i386.h: (host_detect_local_cpu): New define.
7495 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7497 * config/arm/xgene1.md (xgene1): Split into automatons
7498 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
7499 (xgene1_f_load): Adjust reservations and/or types.
7500 (xgene1_f_store): Likewise.
7501 (xgene1_load_pair): Likewise.
7502 (xgene1_store_pair): Likewise.
7503 (xgene1_fp_load1): Likewise.
7504 (xgene1_load1): Likewise.
7505 (xgene1_store1): Likewise.
7506 (xgene1_move): Likewise.
7507 (xgene1_alu): Likewise.
7508 (xgene1_simd): Likewise.
7509 (xgene1_bfm): Likewise.
7510 (xgene1_neon_load1): Likewise.
7511 (xgene1_neon_store1): Likewise.
7512 (xgene1_neon_logic): Likewise.
7513 (xgene1_neon_st1): Likewise.
7514 (xgene1_neon_ld1r): Likewise.
7515 (xgene1_alu_cond): Added.
7516 (xgene1_shift_reg): Likwise.
7517 (xgene1_bfx): Likewise.
7518 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
7520 2017-11-15 Jakub Jelinek <jakub@redhat.com>
7523 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
7525 (can_widen_mult_without_libcall): New function.
7526 (expand_mul_overflow): If only checking unsigned mul overflow,
7527 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
7528 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
7529 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
7530 (expand_DIVMOD): Formatting fix.
7531 * expmed.h (expand_mult): Add NO_LIBCALL argument.
7532 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
7533 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
7535 PR tree-optimization/82977
7536 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
7537 constructed temporary to strlen_to_stridx.put.
7539 2017-11-15 Martin Liska <mliska@suse.cz>
7541 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
7542 * configure: Regenerate.
7544 2017-11-15 Martin Liska <mliska@suse.cz>
7547 * config/sh/sh-mem.cc: Use proper probability for
7550 2017-11-14 Jeff Law <law@redhat.com>
7552 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
7553 the red zone for stack_clash_protection_final_dynamic_probe targets
7554 when the total dynamic stack size is zero bytes.
7556 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
7557 blocks is post order.
7559 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
7561 * dumpfile.h (TDF_COMPARE_DEBUG): New.
7562 * final.c (rest_of_clean_state): Set it for the
7563 -fcompare-debug dump.
7564 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
7565 class when TDF_COMPARE_DEBUG is set.
7567 * dwarf2out.c (gen_producer_string): Discard
7570 2017-11-15 Joseph Myers <joseph@codesourcery.com>
7573 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
7574 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
7575 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
7576 (__TGMATH_REAL_2_3): Remove macros.
7577 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
7578 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
7580 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
7582 (remquo): Define using __TGMATH_REAL_3.
7584 2017-11-14 Jeff Law <law@redhat.com>
7586 * vr-values.c: New file with contents extracted from tree-vrp.c.
7587 * Makefile.in (OBJS): Add vr-values.o
7588 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
7589 (set_value_range, set_and_canonicalize_value_range): Likewise.
7590 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
7591 (value_range_constant_singleton, symbolic_range_p): Likewise.
7592 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
7593 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
7594 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
7595 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
7596 (find_case_label_range, find_case_label_index): Likewise.
7597 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
7598 (range_int_cst_singleton_p, value_inside_range): Likewise.
7599 (get_single_symbol): Likewise.
7600 (switch_update): Move structure definition here.
7601 (to_remove_edges, to_update_switch_stmts): Provide externs.
7602 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
7603 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
7604 (vrp_val_is_min, set_value_range): Likewise.
7605 (set_and_canonicalize_value_range, copy_value_range): Likewise.
7606 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
7607 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
7608 (range_is_nonnull, range_int_cst_p): Likewwise.
7609 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
7610 (get_single_symbol, operand_less_p): Likewise
7611 (compare_values_warnv, compare_values): Likewise.
7612 (value_inside_range, value_range_constant_singleton): Likewise.
7613 (zero_nonzero_bitgs_from_vr): Likewise.
7614 (extract_range_from_binary_expr_1): Likewise.
7615 (overflow_comparison_p): Likewise.
7616 (to_remove_edges, to_update_switch_stmts): Likewise.
7617 (find_case_label-index, find_case_label_range): Likewise.
7618 (switch_update, set_value_range_to_nonnegative): Remove.
7619 (set_value_range_to_truthvalue): Likewise.
7620 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
7621 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
7622 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
7623 (find_case_label_ranges, test_for_singularity): Likewise.
7624 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
7625 (x_vr_values): Move to its remaining use site.
7627 2017-11-10 Jeff Law <law@redhat.com>
7629 * vr-values.h (VR_INITIALIZER): Move #define here.
7630 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
7631 * Makefile.in (OBJS): Add tree-evrp.o
7632 * tree-vrp.h (assert_info): Move structure definition here.
7633 (set_value_range_to_varying): Prototype.
7634 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7635 (infer_value_range, register_edge_assert_for): Likewise.
7636 (stmt_interesting_for_vrp): Likewise.
7637 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
7638 (set_value_range_to_varying): No longer static.
7639 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7640 (infer_value_range, register_edge_assert_for): Likewise.
7642 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7644 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
7645 generate the XXBRD instruction.
7647 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
7648 (rs6000_builtin_type_compatible): Treat _Float128 and long double
7649 as being compatible if -mabi=ieeelongdouble.
7650 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
7651 to setup float128 built-ins with hardware support.
7652 (BU_FLOAT128_HW_2): Likewise.
7653 (BU_FLOAT128_HW_3): Likewise.
7654 (BU_FLOAT128_HW_VSX_1): Likewise.
7655 (BU_FLOAT128_HW_VSX_2): Likewise.
7656 (scalar_extract_expq): Change float128 built-in functions to
7657 accommodate having both KFmode and TFmode functions. Use the
7658 KFmode variant as the default.
7659 (scalar_extract_sigq): Likewise.
7660 (scalar_test_neg_qp): Likewise.
7661 (scalar_insert_exp_q): Likewise.
7662 (scalar_insert_exp_qp): Likewise.
7663 (scalar_test_data_class_qp): Likewise.
7664 (sqrtf128_round_to_odd): Delete processing the round to odd
7665 built-in functions as special built-in functions, and define them
7666 as float128 built-ins. Use the KFmode variant as the default.
7667 (truncf128_round_to_odd): Likewise.
7668 (addf128_round_to_odd): Likewise.
7669 (subf128_round_to_odd): Likewise.
7670 (mulf128_round_to_odd): Likewise.
7671 (divf128_round_to_odd): Likewise.
7672 (fmaf128_round_to_odd): Likewise.
7673 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7674 support for KFmode and TFmode xststdcqp calls.
7675 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
7676 point, switch the built-in handlers for the get/set float128
7677 exponent, get float128 mantissa, float128 test built-ins, and the
7678 float128 round to odd built-in functions. Eliminate creating the
7679 float128 round to odd built-in functions as special built-ins.
7680 (rs6000_init_builtins): Eliminate special creation of the float128
7681 round to odd built-in functions.
7682 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
7683 function insns to support both TFmode and KFmode variants.
7684 (xsxsigqp_<mode>): Likewise.
7685 (xsiexpqpf_<mode>): Likewise.
7686 (xsiexpqp_<mode>): Likewise.
7687 (xststdcqp_<mode>): Likewise.
7688 (xststdcnegqp_<mode>): Likewise.
7689 (xststdcqp_<mode>): Likewise.
7691 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7693 * tree-ssa-threadupdate.c (compute_path_counts): Remove
7694 unused path_in_freq_ptr parameter.
7695 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
7697 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7699 * ipa-inline.c (edge_badness): Dump sreal frequency.
7700 (compute_inlined_call_time): Match natural implementaiton ...
7701 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
7702 forgotten division by CGRAPH_FREQ_BASE.
7704 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7706 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
7707 Solaris 11. Update comment.
7708 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
7710 * config/sol2.h (STARTFILE_SPEC): Likewise.
7711 * configure: Regenerate.
7713 2017-11-14 Carl Love <cel@us.ibm.com>
7715 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
7716 le_ and be_ prefixes to swap* variables. Remove
7717 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
7720 2017-11-14 Jason Merrill <jason@redhat.com>
7722 Support GTY((cache)) on hash_map.
7723 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
7724 (ggc_cache_remove): Override it instead of ggc_mx.
7725 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
7726 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
7727 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
7728 (simple_cache_map_traits): Override maybe_mx.
7729 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
7730 (hash_map): Friend gt_cleare_cache.
7731 (gt_cleare_cache): New.
7732 * tree.h (tree_cache_traits): New hash_map traits class.
7733 (tree_cache_map): New typedef.
7735 2017-11-14 Richard Biener <rguenther@suse.de>
7737 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
7738 paramter and handling.
7739 (cleanup_control_flow_bb): Likewise.
7740 (cleanup_control_flow_pre): New helper performing a DFS walk
7741 to call cleanup_control_flow_bb in PRE order.
7742 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
7743 via cleanup_control_flow_pre.
7745 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
7747 * config/aarch64/aarch64-simd.md
7748 (aarch64_simd_bsl<mode>_internal): Remove DImode.
7749 (*aarch64_simd_bsl<mode>_alt): Likewise.
7750 (aarch64_simd_bsldi_internal): New.
7751 (aarch64_simd_bsldi_alt): Likewise.
7753 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7755 * tracer.c (better_p): Do not compare frequencies.
7756 * reg-stack.c (better_edge): Likewise.
7757 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
7760 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7762 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
7763 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
7764 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
7765 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7766 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
7767 * predict.c (maybe_hot_frequency_p): Inline to ...
7768 (maybe_hot_count_p): ... here; rewrite to counts.
7769 (counts_to_freqs): Rename to ...
7770 (update_max_bb_count): ... this one.
7771 (expensive_function_p): Use counts.
7772 (estimate_bb_frequencies): Update.
7773 (rebuild_frequencies): Update.
7774 * predict.h (counts_to_freqs): Rename to ...
7775 (update_max_bb_count): ... this one.
7776 * profile.c (compute_branch_probabilities): Add debug info
7777 * tree-inline.c (expand_call_inline): Update debug info.
7778 (optimize_inline_calls): Use update_max_bb_count..
7779 (tree_function_versioning): Use update_max_bb_count..
7780 * value-prof.c (gimple_value_profile_transformations):
7781 Do not use update_max_bb_count.
7783 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7785 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7786 always use frequencies.
7788 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7790 * bb-reorder.c: Remove frequencies from comments.
7791 (better_edge_p): Use profile counts.
7792 (find_traces): Dump profile counts.
7793 (rotate_loop): Use profile counts.
7794 (find_traces_1_round): Likewise.
7795 (connect_better_edge_p): Use counts instead of probabilities for
7797 (copy_bb_p): Drop early check for non-0 frequency.
7798 (sanitize_hot_paths): Update comments.
7800 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7802 * ipa-split.c (struct split_point): Add count.
7803 (consider_split): Do not compute incoming frequency; compute incoming
7804 count and store it to split_point.
7805 (split_function): Set count of the call to split part correctly.
7807 2017-11-13 Carl Love <cel@us.ibm.com>
7809 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7811 2017-11-13 Tom Tromey <tom@tromey.com>
7813 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7815 2017-11-13 Carl Love <cel@us.ibm.com>
7817 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7818 Add support for builtins:
7819 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7820 vector {un,}signed {char,int,short},
7821 vector {un,}signed {char,int,short}) arguments.
7822 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7823 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7824 Add BU_P9V_AV_2 expansions for the builtins.
7825 * config/rs6000/altivec.h (vec_first_match_index,
7826 vec_first_mismatch_index, vec_first_match_or_eos_index,
7827 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7828 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7829 new extern declaration.
7830 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7831 * config/rs6000/vsx.md (first_match_index_<mode>,
7832 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7833 first_mismatch_or_eos_index_<mode>): Add define expand.
7834 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7835 * doc/extend.texi: Update the built-in documenation file for the new
7838 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7840 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7841 into the min/max operations for _Float<N> and _Float<N>X types.
7843 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7846 * dwarf2out.c (do_eh_frame): New static variable.
7847 (dwarf2out_begin_prologue): Set it.
7848 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7850 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7852 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7854 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7856 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7859 * debug.h (dwarf2out_do_eh_frame): Declare.
7860 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7861 (dwarf2out_do_frame): Use it.
7862 (dwarf2out_do_cfi_asm): Likewise.
7863 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7864 (dwarf2out_assembly_start): Likewise.
7865 (dwarf2out_begin_prologue): Fix comment.
7866 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7867 if the target needs either debug or unwind DWARF2 info.
7868 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7869 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7870 -ftrapping-math, -ftrapv and -fwrapv.
7872 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7874 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7875 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7876 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7877 (estimate_edge_size_and_time): Likewise.
7878 (ipa_merge_fn_summary_after_inlining): Likewise.
7879 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7880 (compute_uninlined_call_time): Use sreal_frequency.
7881 (compute_inlined_call_time): Likewise.
7882 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7883 * profile-count.c: Include sreal.h.
7884 (profile_count::to_sreal_scale): New.
7885 * profile-count.h: Forward declare sreal.
7886 (profile_count::to_sreal_scale): Declare.
7888 2017-11-13 Nathan Sidwell <nathan@acm.org>
7890 * diagnostic.c (maybe_line_and_column): New.
7891 (diagnostic_get_location_text): Use it.
7892 (diagnostic_report_current_module): Likewise.
7893 (test_diagnostic_get_location_text): Add tests.
7895 2017-11-13 Luis Machado <luis.machado@linaro.org>
7897 * doc/md.texi (Specifying processor pipeline description): Fix
7898 incorrect latency for the div instruction example.
7900 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7902 PR tree-optimization/78821
7903 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7904 that bit_not_p is the same.
7905 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7906 (split_group): Count precisely bit_not_p bits in each statement.
7907 (invert_op): New function.
7908 (imm_store_chain_info::output_merged_store): Use invert_op to
7909 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7910 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7912 2017-11-13 Martin Liska <mliska@suse.cz>
7914 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7915 (struct source_info): Likewise.
7916 (add_branch_counts): Likewise.
7917 (add_line_counts): Likewise.
7918 (function_summary): Likewise.
7919 (output_intermediate_line): Likewise.
7920 (generate_results): Likewise.
7922 2017-11-13 Martin Liska <mliska@suse.cz>
7924 * gcov.c (struct block_info): Remove typedef for block_t.
7925 (struct line_info): Likewise.
7926 (line_info::has_block): Likewise.
7927 (EXIT_BLOCK): Likewise.
7928 (unblock): Likewise.
7929 (circuit): Likewise.
7930 (get_cycles_count): Likewise.
7931 (process_file): Likewise.
7932 (read_graph_file): Likewise.
7933 (solve_flow_graph): Likewise.
7934 (find_exception_blocks): Likewise.
7935 (add_line_counts): Likewise.
7936 (accumulate_line_info): Likewise.
7937 (output_line_details): Likewise.
7939 2017-11-13 Martin Liska <mliska@suse.cz>
7941 * gcov.c (struct arc_info): Remove typedef for arc_t.
7942 (struct line_info): Likewise.
7943 (add_branch_counts): Likewise.
7944 (output_branch_count): Likewise.
7945 (function_info::~function_info): Likewise.
7946 (circuit): Likewise.
7947 (output_intermediate_line): Likewise.
7948 (read_graph_file): Likewise.
7949 (solve_flow_graph): Likewise.
7950 (find_exception_blocks): Likewise.
7951 (add_line_counts): Likewise.
7952 (accumulate_line_info): Likewise.
7953 (output_line_details): Likewise.
7954 (output_function_details): Likewise.
7956 2017-11-13 Martin Liska <mliska@suse.cz>
7958 * gcov.c (struct function_info): Remove typedef for function_t.
7959 (struct source_info): Likewise.
7960 (source_info::get_functions_at_location): Likewise.
7961 (solve_flow_graph): Likewise.
7962 (find_exception_blocks): Likewise.
7963 (add_line_counts): Likewise.
7964 (output_intermediate_file): Likewise.
7965 (process_file): Likewise.
7966 (generate_results): Likewise.
7967 (release_structures): Likewise.
7968 (read_graph_file): Likewise.
7969 (read_count_file): Likewise.
7970 (accumulate_line_counts): Likewise.
7971 (output_lines): Likewise.
7973 2017-11-13 Martin Liska <mliska@suse.cz>
7975 * gcov.c (function_info::function_info): Remove num_counts
7976 and add vector<gcov_type>.
7977 (function_info::~function_info): Use the vector.
7978 (process_file): Likewise.
7979 (read_graph_file): Likewise.
7980 (read_count_file): Likewise.
7981 (solve_flow_graph): Likewise.
7983 2017-11-13 Martin Liska <mliska@suse.cz>
7985 * gcov.c (function_info::is_artificial): New function.
7986 (process_file): Erase all artificial early.
7987 (generate_results): Skip as all artificial are already
7990 2017-11-13 Martin Liska <mliska@suse.cz>
7992 * gcov.c (read_graph_file): Store to global vector of functions.
7993 (read_count_file): Iterate the vector.
7994 (process_file): Likewise.
7995 (generate_results): Likewise.
7996 (release_structures): Likewise.
7998 2017-11-13 Jakub Jelinek <jakub@redhat.com>
8000 PR tree-optimization/82954
8001 * gimple-ssa-store-merging.c
8002 (imm_store_chain_info::coalesce_immediate_stores): If
8003 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
8005 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
8007 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
8008 Upddate call to ENDIAN_LANE_N.
8009 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
8010 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
8011 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
8012 and use aarch64_endian_lane_rtx.
8013 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
8015 2017-11-12 Tom de Vries <tom@codesourcery.com>
8017 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
8019 2017-11-12 Tom de Vries <tom@codesourcery.com>
8021 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
8023 2017-11-12 Tom de Vries <tom@codesourcery.com>
8025 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
8027 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
8028 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
8029 * defaults.h (ASM_OUTPUT_LABELREF): Same.
8031 2017-11-11 Martin Sebor <msebor@redhat.com>
8034 * doc/extend.texi (attribute nonstring): Remove spurious argument.
8037 * prefic.c (translate_name): Replace strncpy with memcpy to
8038 avoid -Wstringop-truncation.
8040 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8042 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
8044 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8046 * predict.c (maybe_hot_frequency_p): Do not use cfun.
8048 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8050 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
8053 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
8055 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
8056 enable generating XXBRH if the value is in a vector register.
8057 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
8058 value is in a vector register.
8059 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
8060 register to register bswap64's instead of doing the GPR sequence
8061 used on previous machines.
8062 (bswapdi2_xxbrd): New insn.
8063 (bswapdi2_reg): Disallow on ISA 3.0.
8064 (register to register bswap64 splitter): Do not split the insn on
8065 ISA 3.0 systems that use XXBRD.
8067 2017-11-10 Martin Sebor <msebor@redhat.com>
8070 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
8072 (find_subframework_file): Same.
8074 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8076 * auto-profile.c (afdo_indirect_call): Drop frequency.
8077 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
8078 (cgraph_node::create_edge): Drop frequency argument.
8079 (cgraph_node::create_indirect_edge): Drop frequency argument.
8080 (cgraph_edge::make_speculative): Drop frequency arguments.
8081 (cgraph_edge::resolve_speculation): Do not update frequencies
8082 (cgraph_edge::dump_edge_flags): Do not dump frequency.
8083 (cgraph_node::dump): Check consistency in IPA mode.
8084 (cgraph_edge::maybe_hot_p): Use IPA counter.
8085 (cgraph_edge::verify_count_and_frequency): Rename to ...
8086 (cgraph_edge::verify_count): ... this one; drop frequency checking.
8087 (cgraph_node::verify_node): Update.
8088 * cgraph.h (struct cgraph_edge): Drop frequency.
8089 (cgraph_edge::frequency): New function.
8090 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
8092 (cgraph_edge::rebuild_edges): Likewise.
8093 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
8094 (duplicate_thunk_for_node): Do not pass frequency.
8095 (cgraph_node::create_clone): Scale only counts.
8096 (cgraph_node::create_virtual_clone): Do not pass frequency.
8097 (cgraph_node::create_edge_including_clones): Do not pass frequency.
8098 (cgraph_node::create_version_clone): Do not pass frequency.
8099 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
8100 (cgraph_node::expand_thunk): Do not pass frequency.
8101 (cgraph_node::create_wrapper): Do not pass frequency.
8102 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
8104 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
8105 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
8106 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
8107 (gather_caller_stats): Use frequency function.
8108 (ipcp_cloning_candidate_p): Use frequency function.
8109 (ipcp_propagate_stage): Use frequency function.
8110 (get_info_about_necessary_edges): Use frequency function.
8111 (update_profiling_info): Update only IPA profile.
8112 (update_specialized_profile): Use frequency functoin.
8113 (perhaps_add_new_callers): Update only IPA profile.
8114 * ipa-devirt.c (ipa_devirt): Use IPA profile.
8115 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
8116 (dump_ipa_call_summary): Use frequency function.
8117 (estimate_edge_size_and_time): Use frequency function.
8118 (ipa_merge_fn_summary_after_inlining): Use frequency function.
8119 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
8120 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
8121 (update_noncloned_counts): ... ths one; scale counts only.
8122 (clone_inlined_nodes): Do not scale frequency.
8123 (inline_call): Do not pass frequency.
8124 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
8125 (compute_inlined_call_time): Use IPA profile.
8126 (want_inline_small_function_p): Use IPA profile.
8127 (want_inline_self_recursive_call_p): Use IPA profile.
8128 (edge_badness): Use IPA profile.
8129 (lookup_recursive_calls): Use IPA profile.
8130 (recursive_inlining): Do not pass frequency.
8131 (resolve_noninline_speculation): Do not update frequency.
8132 (inline_small_functions): Collect max of IPA profile.
8133 (dump_overall_stats): Dump IPA porfile.
8134 (dump_inline_stats): Dump IPA porfile.
8135 (ipa_inline): Collect IPA stats.
8136 * ipa-inline.h (clone_inlined_nodes): Update prototype.
8137 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
8138 (ipa_propagate_frequency): Use frequency function.
8139 (ipa_profile): Cleanup.
8140 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
8141 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
8142 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
8143 (input_node): Do not stream frequency.
8144 (input_edge): Do not stream frequency.
8145 (merge_profile_summaries): Scale only IPA profiles.
8146 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
8147 * predict.c (drop_profile): Do not recompute frequency.
8148 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
8149 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
8150 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
8151 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
8152 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
8153 * tree-inline.c (copy_bb): Do not scale frequency.
8154 (expand_call_inline): Do not scale frequency.
8155 (tree_function_versioning): Do not scale frequency.
8156 * ubsan.c (ubsan_create_edge): Do not pass frequency.
8158 2017-11-10 Julia Koval <julia.koval@intel.com>
8160 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
8161 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
8162 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
8163 (_mm256_maskz_gf2p8affine_epi64_epi8)
8164 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
8165 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
8166 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
8167 (__builtin_ia32_vgf2p8affineqb_v32qi)
8168 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
8169 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
8171 2017-11-10 Tamar Christina <tamar.christina@arm.com>
8175 (arm_option_override): Refactor.
8176 (arm_option_reconfigure_globals): New.
8177 (arm_options_perform_arch_sanity_checks): New.
8178 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
8180 (arm_options_perform_arch_sanity_checks): Likewise
8182 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
8184 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
8185 (power9-qpmul): New.
8186 * rs6000/rs6000.md ("type" attr): Add qmul.
8187 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
8188 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
8189 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
8191 2017-11-10 Martin Sebor <msebor@redhat.com>
8194 * builtins.c (compute_objsize): Handle arrays that
8195 compute_builtin_object_size likes to fail for. Make extern.
8196 * builtins.h (compute_objsize): Declare.
8197 (check_strncpy_sizes): New function.
8198 (expand_builtin_strncpy): Call check_strncpy_sizes.
8199 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
8200 -Wstringop-truncation.
8201 (gimple_fold_builtin_strncat): Same.
8202 * gimple.c (gimple_build_call_from_tree): Set call location.
8203 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
8204 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
8205 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
8206 (handle_builtin_strlen): Use strlen_to_stridx.
8207 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
8209 Use strlen_to_stridx.
8210 (pass_strlen::execute): Release strlen_to_stridx.
8211 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
8212 (-Wstringop-truncation): Document new option.
8214 2017-11-10 Martin Liska <mliska@suse.cz>
8216 PR gcov-profile/82702
8217 * gcov.c (main): Handle intermediate files in a different
8219 (get_gcov_intermediate_filename): New function.
8220 (output_gcov_file): Remove support of intermediate files.
8221 (generate_results): Allocate intermediate file.
8222 (release_structures): Clean-up properly fn_end.
8223 (output_intermediate_file): Start iterating with line 1.
8225 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8227 PR tree-optimization/82929
8228 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8229 ops_swapped_p non-static data member.
8230 (store_immediate_info::store_immediate_info): Clear it.
8231 (imm_store_chain_info::coalesce_immediate_stores): If swapping
8232 ops set ops_swapped_p.
8233 (count_multiple_uses): Handle ops_swapped_p.
8235 2017-11-10 Martin Liska <mliska@suse.cz>
8237 * coverage.c (coverage_init): Stream information about
8238 support of has_unexecuted_blocks.
8239 * doc/gcov.texi: Document that.
8240 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
8241 * gcov.c (read_graph_file): Likewise.
8242 (output_line_beginning): Fix a small issue with
8245 2017-11-10 Bin Cheng <bin.cheng@arm.com>
8247 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
8248 reference of trivial component.
8250 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8253 * gimple-ssa-store-merging.c
8254 (pass_store_merging::terminate_all_aliasing_chains): For
8255 gimple_store_p stmts also call refs_output_dependent_p.
8257 PR rtl-optimization/82913
8258 * compare-elim.c (try_merge_compare): Punt if def_insn is not
8261 2017-11-09 Jeff Law <law@redhat.com>
8263 * vr-values.h: New file with vr_values class.
8264 * tree-vrp.c: Include vr-values.h
8265 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
8266 data objects into the vr_values class.
8267 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
8268 (get_value_range): Make it a member function within vr_values class.
8269 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
8270 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
8271 (op_with_constant_singleton_value_range): Likewise.
8272 (extract_range_for_var_from_comparison_expr): Likewise.
8273 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
8274 (extract_range_from_binary_expr): Likewise.
8275 (extract_range_from_unary_expr): Likewise.
8276 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
8277 (check_for_binary_op_overflow, extract_range_basic): Likewise.
8278 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
8279 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
8280 (compare_name_with_value, compare_names): Likewise.
8281 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
8282 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
8283 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
8284 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
8285 (extract_range_from_phi_node): Likewise.
8286 (simplify_truth_ops_using_ranges): Likewise.
8287 (simplify_div_or_mod_using_ranges): Likewise.
8288 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
8289 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
8290 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
8291 (simplify_float_conversion_using_ranges): Likewise.
8292 (simplify_internal_call_using_ranges): Likewise.
8293 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
8294 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
8295 poitner via x_vr_values for calls into gimple folder.
8296 (vrp_initialize_lattice): Make this the vr_values ctor.
8297 (vrp_free_lattice): Make this the vr_values dtor.
8298 (set_vr_value): New function.
8299 (class vrp_prop): Add vr_values data member. Add various member
8300 functions as well as member functions that delegate to vr_values.
8301 (check_array_ref): Make a member function within vrp_prop class.
8302 (search_for_addr_array, vrp_initialize): Likewise.
8303 (vrp_finalize): Likewise. Revamp to avoid direct access to
8304 vr_value, values_propagated, etc.
8305 (check_array_bounds): Extract vrp_prop class instance pointer from
8306 walk info structure. Use it to call member functions.
8307 (check_all_array_refs): Make a member function within vrp_prop class.
8308 Smuggle class instance pointer via walk info structure.
8309 (x_vr_values): New local static.
8310 (vrp_valueize): Use x_vr_values to get class instance.
8311 (vr_valueize_1): Likewise.
8312 (class vrp_folder): Add vr_values data member. Add various member
8313 functions as well as member functions that delegate to vr_values.
8314 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
8315 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
8316 class instance from vr_values. Use it to call member functions.
8317 (vrp_dom_walker): Add vr_values data member.
8318 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
8319 instance via x_vr_values.
8320 (identify_jump_threads): Accept vr_values as argument. Store
8321 it into the walker structure.
8322 (evrp_dom_walker): Add vr_values class data member. Add various
8324 (evrp_dom_walker::try_find_new_range): Use vr_values data
8325 member to access the memory allocator.
8326 (evrp_dom_walker::before_dom_children): Store vr_values class
8327 instance into the vrp_folder class.
8328 (evrp_dom_walker::push_value_range): Rework to avoid direct
8329 access to num_vr_values and vr_value.
8330 (evrp_dom_walker::pop_value_range): Likewise.
8331 (execute_early_vrp): Remove call to vrp_initialize_lattice.
8332 Use vr_values to get to dump_all_value_ranges member function.
8333 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
8334 and simplify_cond_using_ranges_2 via vrp_prop class instance.
8335 Pass vr_values class instance down to identify_jump_threads.
8336 Remove call to vrp_free_lattice.
8337 (debug_all_value_ranges): Remove.
8339 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
8340 (vrp_folder): Likewise.
8342 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
8343 Get it from the existing bitmap instead.
8344 (vrp_intersect_ranges_1): Likewise.
8346 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8348 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8350 (store_immediate_info::store_immediate_info): Add bitnotp argument,
8351 set bit_not_p to it.
8352 (imm_store_chain_info::coalesce_immediate_stores): Break group
8353 if bit_not_p is different.
8354 (count_multiple_uses, split_group,
8355 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
8356 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
8357 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
8358 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
8359 to store_immediate_info ctor.
8361 2017-11-09 Jim Wilson <jimw@sifive.com>
8363 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
8364 (scan_prog_file): Likewise.
8366 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8368 * bb-reorder.c (max_entry_frequency): Remove.
8369 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
8370 connect_traces, push_to_next_round_p): Remove prototypes.
8371 (find_traces_1_round): Use counts only.
8372 (push_to_next_round_p): Likewise.
8373 (find_traces): Likewise.
8374 (rotate_loop): Likewise.
8375 (find_traces_1_round): Likewise.
8376 (connect_traces): Likewise.
8377 (edge_order): Likewise.
8379 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
8381 * config/arm/arm.c (output_return_instruction): Add comments to
8382 indicate requirement for cmse_nonsecure_entry return to account
8383 for the size of clearing instruction output here.
8384 (thumb_exit): Likewise.
8385 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
8386 return in hardfloat mode.
8388 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
8390 * config/rs6000/rs6000.c (machine_function): Add a bool,
8391 "toc_is_wrapped_separately".
8392 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
8393 if it wasn't explicitly set or unset, we are optimizing for speed, and
8394 doing separate shrink-wrapping.
8395 (rs6000_get_separate_components): Enable the TOC component if
8396 saving the TOC register in the prologue.
8397 (rs6000_components_for_bb): Handle the TOC component.
8398 (rs6000_emit_prologue_components): Store the TOC register where needed.
8399 (rs6000_set_handled_components): Mark TOC as handled, if handled.
8400 (rs6000_emit_prologue): Don't save the TOC if that is already done.
8402 2017-11-09 Martin Jambor <mjambor@suse.cz>
8404 * ipa-param-manipulation.c: New file.
8405 * ipa-param-manipulation.h: Likewise.
8406 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
8407 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
8408 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
8409 (ipa_parm_adjustment): Likewise.
8410 (ipa_parm_adjustment_vec): Likewise.
8411 (ipa_get_vector_of_formal_parms): Moved declaration to
8412 ipa-param-manipulation.h.
8413 (ipa_get_vector_of_formal_parm_types): Likewise.
8414 (ipa_modify_formal_parameters): Likewise.
8415 (ipa_modify_call_arguments): Likewise.
8416 (ipa_combine_adjustments): Likewise.
8417 (ipa_dump_param_adjustments): Likewise.
8418 (ipa_modify_expr): Likewise.
8419 (ipa_get_adjustment_candidate): Likewise.
8420 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
8421 ipa-param-manipulation.c.
8422 (ipa_get_vector_of_formal_parm_types): Likewise.
8423 (ipa_modify_formal_parameters): Likewise.
8424 (ipa_modify_call_arguments): Likewise.
8425 (ipa_modify_expr): Likewise.
8426 (get_ssa_base_param): Likewise.
8427 (ipa_get_adjustment_candidate): Likewise.
8428 (index_in_adjustments_multiple_times_p): Likewise.
8429 (ipa_combine_adjustments): Likewise.
8430 (ipa_dump_param_adjustments): Likewise.
8431 * tree-sra.c: Also include ipa-param-manipulation.h
8432 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
8435 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8436 Alan Hayward <alan.hayward@arm.com>
8437 David Sherwood <david.sherwood@arm.com>
8439 * doc/sourcebuild.texi (vect_masked_store): Document.
8441 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8442 Alan Hayward <alan.hayward@arm.com>
8443 David Sherwood <david.sherwood@arm.com>
8445 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
8447 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8448 Alan Hayward <alan.hayward@arm.com>
8449 David Sherwood <david.sherwood@arm.com>
8451 * doc/sourcebuild.texi (vect_variable_length): Document.
8453 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8454 Alan Hayward <alan.hayward@arm.com>
8455 David Sherwood <david.sherwood@arm.com>
8457 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
8459 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8460 Alan Hayward <alan.hayward@arm.com>
8461 David Sherwood <david.sherwood@arm.com>
8463 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
8465 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8466 Alan Hayward <alan.hayward@arm.com>
8467 David Sherwood <david.sherwood@arm.com>
8469 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
8470 previously undocumented selectors.
8471 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
8473 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8475 * doc/rtl.texi (const_vector): Say that elements can be
8476 const_wide_ints too.
8477 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
8478 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
8479 (gen_vec_duplicate): Use it instead of CONSTANT_P.
8480 * optabs.c (expand_vector_broadcast): Likewise.
8482 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8483 Alan Hayward <alan.hayward@arm.com>
8484 David Sherwood <david.sherwood@arm.com>
8486 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
8487 scaled index even if the unscaled address was invalid.
8488 Don't increase the complexity of using a scale in that case.
8490 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8491 Alan Hayward <alan.hayward@arm.com>
8492 David Sherwood <david.sherwood@arm.com>
8494 * doc/rtl.texi: Rewrite the subreg rules so that they partition
8495 the inner register into REGMODE_NATURAL_SIZE bytes rather than
8496 UNITS_PER_WORD bytes.
8497 * emit-rtl.c (validate_subreg): Divide subregs into blocks
8498 based on REGMODE_NATURAL_SIZE of the inner mode.
8499 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
8500 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
8501 * expmed.c (lowpart_bit_field_p): Divide the value up into
8502 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
8503 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
8504 whether something is likely to occupy more than one register.
8506 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8509 * ipa-inline-transform.c (update_noncloned_frequencies): Use
8510 profile_count::adjust_for_ipa_scaling.
8511 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
8512 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
8514 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
8516 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8518 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
8519 (split_group): Add total_orig and total_new arguments, estimate the
8520 number of statements related to the store group without store merging
8521 and with store merging.
8522 (imm_store_chain_info::output_merged_store): Adjust split_group
8523 callers, punt if estimated number of statements with store merging
8524 is not smaller than estimated number of statements without it.
8526 (handled_load): Remove has_single_use checks.
8527 (pass_store_merging::process_store): Likewise.
8529 2017-11-09 Richard Biener <rguenther@suse.de>
8531 PR tree-optimization/82902
8532 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
8534 2017-11-09 Martin Liska <mliska@suse.cz>
8537 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
8540 2017-11-09 Martin Liska <mliska@suse.cz>
8542 PR tree-optimization/82669
8543 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
8545 2017-11-09 Martin Liska <mliska@suse.cz>
8547 PR gcov-profile/48463
8548 * coverage.c (coverage_begin_function): Output also end locus
8549 of a function and information whether the function is
8551 * gcov-dump.c (tag_function): Parse and print the information.
8552 * gcov.c (INCLUDE_MAP): Add include.
8553 (INCLUDE_SET): Likewise.
8554 (struct line_info): Move earlier in the source file because
8555 of vector<line_info> in function_info structure.
8556 (line_info::line_info): Likewise.
8557 (line_info::has_block): Likewise.
8558 (struct source_info): Add new member index.
8559 (source_info::get_functions_at_location): New function.
8560 (function_info::group_line_p): New function.
8561 (output_intermediate_line): New function.
8562 (output_intermediate_file): Use the mentioned function.
8563 (struct function_start): New.
8564 (struct function_start_pair_hash): Likewise.
8565 (process_file): Add code that identifies group functions.
8566 Assign lines either to global or function scope.
8567 (generate_results): Skip artificial functions.
8568 (find_source): Assign index for each source file.
8569 (read_graph_file): Read new flag artificial and end_line.
8570 (add_line_counts): Assign it either to global of function scope.
8571 (accumulate_line_counts): Isolate core of the function to
8572 accumulate_line_info and call it for both function and global
8574 (accumulate_line_info): New function.
8575 (output_line_beginning): Fix GNU coding style.
8576 (print_source_line): New function.
8577 (output_line_details): Likewise.
8578 (output_function_details): Likewise.
8579 (output_lines): Iterate both source (global) scope and function
8581 (struct function_line_start_cmp): New class.
8582 * doc/gcov.texi: Reflect changes in documentation.
8584 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8587 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
8588 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
8589 and similarly for not instead of neg.
8591 2017-11-08 Andi Kleen <ak@linux.intel.com>
8593 * config/i386/i386.opt: Add -mforce-indirect-call.
8594 * config/i386/predicates.md: Check for flag_force_indirect_call.
8595 * doc/invoke.texi: Document -mforce-indirect-call
8597 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
8599 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
8601 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
8602 config/riscv/riscv.c (predict.h): New include.
8603 (riscv_slow_unaligned_access_p): No longer static.
8604 (riscv_block_move_straight): Add require.
8605 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8607 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8610 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
8611 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
8612 nonimmediate_operand predicate for operand 1 instead of
8615 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8617 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
8619 * config/aarch64/constraints.md (Uml): New constraint.
8620 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
8623 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8625 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
8626 of two vec_duplicates into a vec_concat.
8628 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8630 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8631 Simplify vec_merge of vec_duplicate and vec_concat.
8632 * config/aarch64/constraints.md (Utq): New constraint.
8633 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
8636 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8638 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8639 Simplify vec_merge of vec_duplicate and const_vector.
8640 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
8642 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
8643 mode iterator. Update predicate on operand 1 to
8644 handle non-const_vec constants. Delete constraints.
8645 (*aarch64_combinez_be<mode>): Likewise for operand 2.
8647 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8649 PR tree-optimization/78821
8650 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
8652 (store_operand_info::store_operand_info): Initialize it to false.
8653 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
8654 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
8655 store in the group, and if chain_info is non-NULL, to ignore altogether
8657 (compatible_load_p): Fail if bit_not_p does not match.
8658 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
8659 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
8660 (pass_store_merging::process_store): Adjust
8661 terminate_all_aliasing_chains calls to pass NULL in all current spots,
8662 call terminate_all_aliasing_chains newly when adding a store into
8663 a chain with non-NULL chain_info.
8665 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8667 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
8669 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8671 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8673 (aarch64_layout_frame): Initialise emit_frame_chain.
8674 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
8675 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
8677 2017-11-08 Martin Liska <mliska@suse.cz>
8679 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
8680 of gimple_call_internal_p.
8682 2017-11-07 Tom de Vries <tom@codesourcery.com>
8684 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
8686 2017-11-07 Tom de Vries <tom@codesourcery.com>
8688 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
8691 2017-11-08 Martin Liska <mliska@suse.cz>
8694 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
8696 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8698 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
8700 2017-11-07 Jakub Jelinek <jakub@redhat.com>
8703 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
8704 (*cmp<mode>_ccz_1): New insn with $k alternative.
8707 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
8708 RTX_COMM_COMPARE as commutative as well.
8709 (ix86_binary_operator_ok): Formatting fix.
8710 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
8711 *<code><mode>3<mask_name><round_saeonly_name>,
8712 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
8713 *<s>mul<mode>3_highpart<mask_name>,
8714 *vec_widen_umult_even_v16si<mask_name>,
8715 *vec_widen_umult_even_v8si<mask_name>,
8716 *vec_widen_umult_even_v4si<mask_name>,
8717 *vec_widen_smult_even_v16si<mask_name>,
8718 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
8719 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
8720 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
8721 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
8722 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
8723 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
8724 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
8725 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
8726 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
8727 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
8728 ix86_binary_operator_ok. Formatting fixes.
8729 (*<plusminus_insn><mode>3<mask_name><round_name>,
8730 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
8733 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
8735 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
8736 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
8737 eq and ne if TARGET_ISEL.
8738 (cmp): New code_iterator.
8739 (UNS, UNSU_, UNSIK): New code_attrs.
8740 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
8741 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
8743 ("eq<mode>3"): ... this.
8744 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
8746 ("ne<mode>3"): ... this.
8748 2017-11-07 Julia Koval <julia.koval@intel.com>
8751 * common/config/i386/i386-common.c
8752 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
8753 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
8754 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
8755 * config/i386/i386.opt: Ditto.
8756 * config/i386/i386.c (ix86_target_string): Ditto.
8757 (ix86_option_override_internal): Ditto.
8758 (ix86_init_mpx_builtins): Move MPX to args2.
8759 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
8760 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8761 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8762 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8763 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8764 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8765 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
8767 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
8770 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
8771 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
8772 (zero-extendsidi peephole2): Remove peephole.
8774 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8777 * stor-layout.c: Include attribs.h.
8778 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8779 explicit lookup of "aligned" attribute.
8781 2017-11-07 Andrew Waterman <andrew@sifive.com>
8783 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8784 (riscv_expand_block_move): Likewise.
8785 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8787 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8788 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8789 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8790 (riscv_adjust_block_mem): Likewise.
8791 (riscv_block_move_loop): Likewise.
8792 (riscv_expand_block_move): Likewise.
8793 * config/riscv/riscv.md (movmemsi): New pattern.
8795 2017-11-07 Michael Clark <michaeljclark@mac.com>
8797 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8798 (MUSL_DYNAMIC_LINKER): Likewise.
8800 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8802 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8803 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8806 2017-11-07 Richard Biener <rguenther@suse.de>
8808 * match.pd: Fix build.
8810 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8811 Jackson Woodruff <jackson.woodruff@arm.com>
8813 PR tree-optimization/71026
8814 * match.pd: Canonicalize negate in division.
8816 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8819 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8821 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8823 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8824 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8826 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8828 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8829 non-scalar integral types.
8830 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8831 (-(A-B), -(~A)): New transformations.
8833 2017-11-07 Tom de Vries <tom@codesourcery.com>
8835 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8836 semicolon after "do {} while (0)".
8837 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8838 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8839 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8840 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8841 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8843 2017-11-07 Tom de Vries <tom@codesourcery.com>
8845 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8846 after "do {} while (0)".
8847 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8848 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8849 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8850 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8851 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8853 2017-11-07 Tom de Vries <tom@codesourcery.com>
8856 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8858 (arm_rtx_costs_internal): Add missing semicolon after
8859 HANDLE_NARROW_SHIFT_ARITH call.
8861 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8863 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8864 disable isel if it was not set explicitly.
8866 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8868 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8869 (add_type_attribute) likewise.
8871 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8873 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8874 of struct ix86_frame.
8875 (ix86_initial_elimination_offset): Likewise.
8876 (ix86_expand_split_stack_prologue): Likewise.
8878 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8880 * tree-vrp.h (enum value_range_type): Update stale comment.
8882 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8883 Alan Hayward <alan.hayward@arm.com>
8884 David Sherwood <david.sherwood@arm.com>
8886 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8887 (aarch64_expand_vec_perm_const): Take the number of units too.
8888 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8889 (aarch64_expand_vec_perm_const): Likewise.
8890 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8891 (vec_perm<mode>): Update accordingly.
8893 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8894 Alan Hayward <alan.hayward@arm.com>
8895 David Sherwood <david.sherwood@arm.com>
8897 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8898 Take the number of units too.
8899 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8900 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8901 but check for a vector mode before rather than after the call.
8902 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8903 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8904 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8905 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8906 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8907 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8908 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8909 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8910 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8911 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8912 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8913 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8914 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8915 (aarch64_sqdmull2_n<mode>): Update accordingly.
8917 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8918 Alan Hayward <alan.hayward@arm.com>
8919 David Sherwood <david.sherwood@arm.com>
8921 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8922 the number of units too.
8923 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8924 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8925 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8926 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8927 (vec_store_lanesxi<mode>): Update accordingly.
8929 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8930 Alan Hayward <alan.hayward@arm.com>
8931 David Sherwood <david.sherwood@arm.com>
8933 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8934 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8935 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8936 of units rather than the mode.
8937 * config/aarch64/iterators.md (nunits): New mode attribute.
8938 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8939 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8940 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8941 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8942 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8943 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8944 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8945 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8946 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8947 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8948 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8949 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8950 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8951 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8952 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8953 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8954 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8955 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8956 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8957 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8958 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8959 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8960 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8961 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8962 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8963 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8964 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8965 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8966 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8967 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8968 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8969 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8970 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8971 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8972 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8973 (aarch64_simd_vec_setv2di): Likewise.
8975 2017-11-06 Carl Love <cel@us.ibm.com>
8977 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8979 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8980 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8982 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8984 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8985 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8986 Add power 8 macro expansions.
8987 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8988 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8989 power 8 instructions. (VSX_XXBR): Add iterator.
8991 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8993 * config/arm/arm.md (predicable_short_it): Change default to "no",
8994 improve documentation, remove uses that are identical to the default.
8995 (enabled_for_depr_it): Rename to enabled_for_short_it.
8996 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8997 Remove default uses.
8998 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8999 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
9000 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
9001 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
9003 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
9006 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
9007 float128 helper macros, which are no longer used after deleting
9008 the old 'q' built-in functions, and moving the round to odd
9009 built-in functions to being special built-in functions.
9010 (BU_FLOAT128_2): Likewise.
9011 (BU_FLOAT128_1_HW): Likewise.
9012 (BU_FLOAT128_2_HW): Likewise.
9013 (BU_FLOAT128_3_HW): Likewise.
9014 (FABSQ): Delete old 'q' built-in functions.
9015 (COPYSIGNQ): Likewise.
9016 (SQRTF128_ODD): Move round to odd built-in functions to be
9017 special built-in functions, so that we can handle
9018 -mabi=ieeelongdouble.
9019 (TRUNCF128_ODD): Likewise.
9020 (ADDF128_ODD): Likewise.
9021 (SUBF128_ODD): Likewise.
9022 (MULF128_ODD): Likewise.
9023 (DIVF128_ODD): Likewise.
9024 (FMAF128_ODD): Likewise.
9025 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
9026 built-in names to 'f128'.
9027 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
9028 old 'q' built-in functions, as the machine independent code for
9029 'f128' built-in functions handles this.
9030 (rs6000_expand_builtin): Add expansion for float128 round to odd
9031 functions, keying off on -mabi=ieeelongdouble of whether to use
9032 the KFmode or TFmode variant.
9033 (rs6000_init_builtins): Initialize the _Float128 round to odd
9035 * doc/extend.texi (PowerPC Built-in Functions): Document the old
9036 _Float128 'q' built-in functions are now mapped into the new
9037 'f128' built-in functions.
9039 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
9041 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
9042 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
9044 2017-11-06 David Malcolm <dmalcolm@redhat.com>
9047 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
9048 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
9049 * toplev.c: Include "ipa-fnsummary.h".
9050 (toplev::finalize): Call ipa_fnsummary_c_finalize.
9052 2017-11-06 Jakub Jelinek <jakub@redhat.com>
9054 PR tree-optimization/82838
9055 * gimple-ssa-store-merging.c
9056 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
9057 on a separate gimple_seq which is then appended to seq.
9059 2017-11-06 Jeff Law <law@redhat.com>
9062 * config/i386/i386.c (PROBE_INTERVAL): Remove.
9063 (get_probe_interval): New functions.
9064 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
9065 (ix86_adjust_stack_and_probe): Likewise.
9066 (output_adjust_stack_and_probe): Likewise.
9067 (ix86_emit_probe_stack_range): Likewise.
9068 (ix86_expand_prologue): Likewise.
9070 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
9072 PR tree-optimization/82816
9073 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
9074 if the modes of the two types are the same.
9075 (convert_plusminus_to_widen): Likewise.
9077 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9079 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
9081 (usadv16qi): New define_expand.
9082 (usadv8hi): New define_expand.
9084 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
9087 * ipa-inline-transform.c (update_noncloned_frequencies): Always
9089 (inline_transform): Likewise.
9090 * predict.c (counts_to_freqs): Remove useless conditional.
9091 * profile-count.h (profile_count::apply_scale): Move sanity check.
9092 * tree-inline.c (copy_bb): Always scale.
9093 (copy_cfg_body): Likewise.
9095 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
9098 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
9100 (*cmp_ite0): Add enabled_for_depr_it attribute.
9101 (*cmp_ite1): Likewise.
9103 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
9105 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
9108 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
9110 * tree-vrp.c (vrp_int_const_binop): Return true on success and
9111 return the value by pointer.
9112 (extract_range_from_multiplicative_op_1): Update accordingly.
9113 Return as soon as an operation fails.
9115 2017-11-05 Tom de Vries <tom@codesourcery.com>
9118 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
9119 (DEF_SANITIZER_BUILTIN): ... here.
9120 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
9121 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
9123 2017-11-05 Tom de Vries <tom@codesourcery.com>
9126 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
9128 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
9129 ASM_OUTPUT_BEFORE_CASE_LABEL call.
9130 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
9132 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9133 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9134 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9136 2017-11-05 Tom de Vries <tom@codesourcery.com>
9139 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
9142 2017-11-04 Michael Clark <michaeljclark@mac.com>
9144 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
9145 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
9147 (*addsi3_extended): Likewise.
9148 (*addsi3_extended2): Likewise.
9149 (<optab>si3): Likewise.
9150 (<optab>di3): Likewise.
9151 (<optab><mode>3): Likewise.
9152 (<*optabe>si3_internal): Likewise.
9153 (zero_extendqi<SUPERQI:mode>2): Likewise.
9154 (*add<mode>hi3): Likewise.
9155 (*xor<mode>hi3): Likewise.
9156 (<optab>di3): Likewise.
9157 (*<optab>si3_extend): Likewise.
9158 (*sge<u>_<X:mode><GPR:mode>): Likewise.
9159 (*slt<u>_<X:mode><GPR:mode>): Likewise.
9160 (*sle<u>_<X:mode><GPR:mode>): Likewise.
9162 2017-11-04 Andrew Waterman <andrew@sifive.com>
9164 * config/riscv/riscv.c (riscv_option_override): Conditionally set
9165 TARGET_STRICT_ALIGN based upon -mtune argument.
9167 2017-11-04 Andrew Waterman <andrew@sifive.com>
9169 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
9171 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
9173 * config/i386/i386.c (choose_basereg): Use optional scratch
9174 register and add assertion.
9175 (x86_emit_outlined_ms2sysv_save): Use scratch register when
9176 needed, and don't allocate stack.
9177 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
9178 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
9179 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
9181 2017-11-03 Jeff Law <law@redhat.com>
9183 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
9184 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
9185 to probe at the start of a noreturn function.
9187 2017-11-03 Jakub Jelinek <jakub@redhat.com>
9189 PR tree-optimization/78821
9190 * gimple-ssa-store-merging.c: Update the file comment.
9191 (MAX_STORE_ALIAS_CHECKS): Define.
9192 (struct store_operand_info): New type.
9193 (store_operand_info::store_operand_info): New constructor.
9194 (struct store_immediate_info): Add rhs_code and ops data members.
9195 (store_immediate_info::store_immediate_info): Add rhscode, op0r
9196 and op1r arguments to the ctor, initialize corresponding data members.
9197 (struct merged_store_group): Add load_align_base and load_align
9199 (merged_store_group::merged_store_group): Initialize them.
9200 (merged_store_group::do_merge): Update them.
9201 (merged_store_group::apply_stores): Pick the constant for
9202 encode_tree_to_bitpos from one of the two operands, or skip
9203 encode_tree_to_bitpos if neither operand is a constant.
9204 (class pass_store_merging): Add process_store method decl. Remove
9205 bool argument from terminate_all_aliasing_chains method decl.
9206 (pass_store_merging::terminate_all_aliasing_chains): Remove
9207 var_offset_p argument and corresponding handling.
9208 (stmts_may_clobber_ref_p): New function.
9209 (compatible_load_p): New function.
9210 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
9211 if there is overlap and rhs_code is not INTEGER_CST. For
9212 non-overlapping stores terminate group if rhs is not mergeable.
9213 (get_alias_type_for_stmts): Change first argument from
9214 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
9215 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
9216 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
9218 (get_location_for_stmts): Change first argument from
9219 auto_vec<gimple *> & to vec<gimple *> &.
9220 (struct split_store): Remove orig_stmts data member, add orig_stores.
9221 (split_store::split_store): Create orig_stores rather than orig_stmts.
9222 (find_constituent_stmts): Renamed to ...
9223 (find_constituent_stores): ... this. Change second argument from
9224 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
9225 to info structures rather than the statements.
9226 (split_group): Rename ALLOW_UNALIGNED argument to
9227 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
9228 it. Adjust find_constituent_stores caller.
9229 (imm_store_chain_info::output_merged_store): Handle rhs_code other
9230 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
9231 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
9232 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
9233 (mem_valid_for_store_merging): New function.
9234 (handled_load): New function.
9235 (pass_store_merging::process_store): New method.
9236 (pass_store_merging::execute): Use process_store method. Adjust
9237 terminate_all_aliasing_chains caller.
9239 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9241 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9242 Return true for more constants, symbols and label references.
9243 (aarch64_valid_floating_const): Remove unused function.
9245 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9248 * config/aarch64/aarch64.c (aarch64_layout_frame):
9249 Undo forcing of LR at bottom of frame.
9251 2017-11-03 Jeff Law <law@redhat.com>
9254 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
9255 for int_registers_saved.
9257 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
9258 extracted from tree-ssa-dom.c.
9259 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
9260 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
9261 (record_equivalences_from_incoming_edge): Add additional argument
9262 to single_pred_edge_ignoring_loop_edges call.
9263 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
9264 (uncprop_dom_walker::before_dom_children): Add additional argument
9265 to single_pred_edge_ignoring_loop_edges call.
9266 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
9267 single_pred_edge_ignoring_loop_edges rather than open coding.
9268 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
9270 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
9272 * match.pd (-(-A)): Rewrite.
9274 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9276 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
9277 (rs6000_emit_int_cmove): New declaration.
9278 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
9279 (rs6000_emit_sISEL): Delete.
9280 (rs6000_emit_int_cmove): Make non-static.
9281 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
9282 instead of rs6000_emit_sISEL.
9284 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
9286 * asan.c (create_cond_insert_point): Maintain profile.
9287 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
9289 * basic-block.h (struct basic_block_def): Remove frequency.
9290 (EDGE_FREQUENCY): Use to_frequency
9291 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
9293 (find_traces): Update to use to_frequency.
9294 (find_traces_1_round): Likewise; use only IPA counts.
9295 (bb_to_key): Likewise.
9296 (connect_traces): Use IPA counts only.
9297 (copy_bb_p): Update to use to_frequency.
9298 (fix_up_crossing_landing_pad): Likewise.
9299 (sanitize_hot_paths): Likewise.
9300 * bt-load.c (basic_block_freq): Likewise.
9301 * cfg.c (init_flow): Set count_max to uninitialized.
9302 (check_bb_profile): Remove frequencies; check counts.
9303 (dump_bb_info): Do not dump frequencies.
9304 (update_bb_profile_for_threading): Update counts only.
9305 (scale_bbs_frequencies_int): Likewise.
9306 (MAX_SAFE_MULTIPLIER): Remove.
9307 (scale_bbs_frequencies_gcov_type): Update counts only.
9308 (scale_bbs_frequencies_profile_count): Update counts only.
9309 (scale_bbs_frequencies): Update counts only.
9310 * cfg.h (struct control_flow_graph): Add count-max.
9311 (update_bb_profile_for_threading): Update prototype.
9312 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
9313 (find_many_sub_basic_blocks): Likewise.
9314 * cfgcleanup.c (try_forward_edges): Likewise.
9315 (try_crossjump_to_edge): Likewise.
9316 * cfgexpand.c (expand_gimple_cond): Likewise.
9317 (expand_gimple_tailcall): Likewise.
9318 (construct_init_block): Likewise.
9319 (construct_exit_block): Likewise.
9320 * cfghooks.c (verify_flow_info): Check consistency of counts.
9321 (dump_bb_for_graph): Do not dump frequencies.
9322 (split_block_1): Do not update frequencies.
9323 (split_edge): Do not update frequencies.
9324 (make_forwarder_block): Do not update frequencies.
9325 (duplicate_block): Do not update frequencies.
9326 (account_profile_record): Do not update frequencies.
9327 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
9328 for global heuristics.
9329 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
9330 (expected_loop_iterations_unbounded): Use counts only.
9331 * cfgloopmanip.c (scale_loop_profile): Simplify.
9332 (create_empty_loop_on_edge): Simplify
9334 (duplicate_loop_to_header_edge): Simplify
9335 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
9336 (update_br_prob_note): Take care of removing note when profile
9338 (relink_block_chain): Do not dump frequency.
9339 (rtl_account_profile_record): Use to_frequency.
9340 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
9341 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
9342 (cgraph_update_edges_for_call_stmt_node): Likewise.
9343 (cgraph_edge::verify_count_and_frequency): Update.
9344 (cgraph_node::verify_node): Temporarily disable frequency verification.
9345 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
9346 to_cgraph_frequency.
9347 (cgraph_edge::rebuild_edges): Convert to ipa counts.
9348 * cgraphunit.c (init_lowered_empty_function): Do not initialize
9350 (cgraph_node::expand_thunk): Update profile.
9351 * except.c (dw2_build_landing_pads): Do not update frequency.
9352 * final.c (compute_alignments): Use to_frequency.
9353 (dump_basic_block_info): Do not dump frequency.
9354 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
9355 (dump_gimple_bb_header): Do not dump frequency.
9356 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
9358 * gimple-streamer-in.c (input_bb): Do not stream frequency.
9359 * gimple-streamer-out.c (output_bb): Do not stream frequency.
9360 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
9361 (init_before_recovery): Do not update frequency.
9362 (sched_create_recovery_edges): Do not update frequency.
9363 * hsa-gen.c (convert_switch_statements): Do not update frequency.
9364 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
9365 (ipa_cp_c_finalize): Set max_count to uninitialized.
9366 * ipa-fnsummary.c (get_minimal_bb): Use counts.
9367 (param_change_prob): Use counts.
9368 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
9370 * ipa-split.c (consider_split): Use to_frequency.
9371 (split_function): Use to_frequency.
9372 * ira-build.c (loop_compare_func): Likewise.
9373 (mark_loops_for_removal): Likewise.
9374 (mark_all_loops_for_removal): Likewise.
9375 * loop-doloop.c (doloop_modify): Do not update frequency.
9376 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
9378 * lto-streamer-in.c (input_function): Update count_max.
9379 * omp-expand.c (expand_omp_taskreg): Update count_max.
9380 * omp-simd-clone.c (simd_clone_adjust): Update profile.
9381 * predict.c (maybe_hot_frequency_p): Use to_frequency.
9382 (maybe_hot_count_p): Use ipa counts only.
9383 (maybe_hot_bb_p): Simplify.
9384 (maybe_hot_edge_p): Simplify.
9385 (probably_never_executed): Do not take frequency argument.
9386 (probably_never_executed_bb_p): Do not pass frequency.
9387 (probably_never_executed_edge_p): Likewise.
9388 (combine_predictions_for_bb): Check that profile is nonzero.
9389 (propagate_freq): Do not set frequency.
9390 (drop_profile): Simplify.
9391 (counts_to_freqs): Simplify.
9392 (expensive_function_p): Use to_frequency.
9393 (propagate_unlikely_bbs_forward): Simplify.
9394 (determine_unlikely_bbs): Simplify.
9395 (estimate_bb_frequencies): Add hack to silence graphite issues.
9396 (compute_function_frequency): Use ipa counts.
9397 (pass_profile::execute): Update.
9398 (rebuild_frequencies): Use counts only.
9399 (force_edge_cold): Use counts only.
9400 * profile-count.c (profile_count::dump): Dump new count types.
9401 (profile_count::differs_from_p): Check compatiblity.
9402 (profile_count::to_frequency): New function.
9403 (profile_count::to_cgraph_frequency): New function.
9404 * profile-count.h (struct function): Declare.
9405 (enum profile_quality): Add profile_guessed_local and
9406 profile_guessed_global0.
9407 (class profile_proability): Decrease number of bits to 29;
9408 update from_reg_br_prob_note and to_reg_br_prob_note.
9409 (class profile_count: Update comment; decrease number of bits
9410 to 61. Check compatibility.
9411 (profile_count::compatible_p): New private member function.
9412 (profile_count::ipa_p): New member function.
9413 (profile_count::operator<): Handle global zero correctly.
9414 (profile_count::operator>): Handle global zero correctly.
9415 (profile_count::operator<=): Handle global zero correctly.
9416 (profile_count::operator>=): Handle global zero correctly.
9417 (profile_count::nonzero_p): New member function.
9418 (profile_count::force_nonzero): New member function.
9419 (profile_count::max): New member function.
9420 (profile_count::apply_scale): Handle IPA scalling.
9421 (profile_count::guessed_local): New member function.
9422 (profile_count::global0): New member function.
9423 (profile_count::ipa): New member function.
9424 (profile_count::to_frequency): Declare.
9425 (profile_count::to_cgraph_frequency): Declare.
9426 * profile.c (OVERLAP_BASE): Delete.
9427 (compute_frequency_overlap): Delete.
9428 (compute_branch_probabilities): Do not use compute_frequency_overlap.
9429 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
9430 * sched-ebb.c (rank): Use counts only.
9431 * shrink-wrap.c (handle_simple_exit): Use counts only.
9432 (try_shrink_wrapping): Use counts only.
9433 (place_prologue_for_one_component): Use counts only.
9434 * tracer.c (find_best_predecessor): Use to_frequency.
9435 (find_trace): Use to_frequency.
9436 (tail_duplicate): Use to_frequency.
9437 * trans-mem.c (expand_transaction): Do not update frequency.
9438 * tree-call-cdce.c: Do not update frequency.
9439 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
9440 (gimple_merge_blocks): Likewise.
9441 (gimple_split_edge): Likewise.
9442 (gimple_duplicate_sese_region): Likewise.
9443 (gimple_duplicate_sese_tail): Likewise.
9444 (move_sese_region_to_fn): Likewise.
9445 (gimple_account_profile_record): Likewise.
9446 (insert_cond_bb): Likewise.
9447 * tree-complex.c (expand_complex_div_wide): Likewise.
9448 * tree-eh.c (lower_resx): Update profile.
9449 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
9451 (initialize_cfun): Do not initialize frequencies
9452 (freqs_to_counts): Delete.
9453 (copy_cfg_body): Ignore count parameter.
9454 (copy_body): Update.
9455 (expand_call_inline): Update count_max.
9456 (optimize_inline_calls): Update count_max.
9457 (tree_function_versioning): Update count_max.
9458 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
9459 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
9461 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
9462 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
9463 (try_peel_loop): Likewise.
9464 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
9466 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
9467 (tree_transform_and_unroll_loop): Do not use frequencies
9468 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
9469 Use reliable prediction only.
9470 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
9471 * tree-ssa-sink.c (select_best_block): Use to_frequency.
9472 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
9473 probability scaling.
9474 * tree-ssa-threadupdate.c (create_block_for_threading): Do
9475 not update frequency
9476 (any_remaining_duplicated_blocks): Likewise.
9477 (update_profile): Likewise.
9478 (estimated_freqs_path): Delete.
9479 (freqs_to_counts_path): Delete.
9480 (clear_counts_path): Delete.
9481 (ssa_fix_duplicate_block_edges): Likewise.
9482 (duplicate_thread_path): Likewise.
9483 * tree-switch-conversion.c (gen_inbound_check): Use counts.
9484 * tree-tailcall.c (decrease_profile): Do not update frequency.
9485 (eliminate_tail_call): Likewise.
9486 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
9487 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
9488 (optimize_mask_stores): Likewise.
9489 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9490 * ubsan.c (ubsan_expand_null_ifn): Update profile.
9491 (ubsan_expand_ptr_ifn): Update profile.
9492 * value-prof.c (gimple_ic): Simplify.
9493 * value-prof.h (gimple_ic): Update prototype.
9494 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
9495 * ipa-inline.c (compute_uninlined_call_time): Be sure that
9497 (want_inline_self_recursive_call_p): Likewise.
9498 (resolve_noninline_speculation): Only cummulate defined counts.
9499 (inline_small_functions): Use nonzero_p.
9500 (ipa_inline): Do not access freed node.
9502 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9504 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9505 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
9507 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
9509 * config/riscv/riscv.c (riscv_legitimize_move): Handle
9510 non-legitimate address.
9512 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9514 * config/rs6000/rs6000.md (*lt0_disi): Delete.
9515 (*lt0_<mode>di, *lt0_<mode>si): New.
9517 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9519 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
9520 TARGET_PAIRED_FLOAT.
9522 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
9523 Jim Wilson <jim.wilson@linaro.org>
9525 * config/aarch64/aarch64-cores.def (saphira): New CPU.
9526 * config/aarch64/aarch64-tune.md: Regenerated.
9527 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
9528 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
9530 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
9532 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
9533 (arc_expand_prologue): Restore blink for millicode.
9534 * config/arc/linux.h (LINK_EH_SPEC): Defined.
9536 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
9539 * config/i386/i386.c (ix86_vector_duplicate_value): Use
9540 gen_vec_duplicate after forcing the scalar into a register.
9542 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
9544 * combine (try_combine): Print the insns input to try_combine to the
9547 2017-11-02 Steve Ellcey <sellcey@cavium.com>
9550 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
9551 Remove second argument from aarch64_process_target_attr call.
9552 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
9554 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
9556 (aarch64_handle_attr_arch): Remove second argument.
9557 (aarch64_handle_attr_cpu): Ditto.
9558 (aarch64_handle_attr_tune): Ditto.
9559 (aarch64_handle_attr_isa_flags): Ditto.
9560 (aarch64_process_one_target_attr): Ditto.
9561 (aarch64_process_target_attr): Ditto.
9562 (aarch64_option_valid_attribute_p): Remove second argument.
9563 on aarch64_process_target_attr call.
9565 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9567 * diagnostic.c: Include "selftest-diagnostic.h".
9568 (selftest::assert_location_text): New function.
9569 (selftest::test_diagnostic_get_location_text): New function.
9570 (selftest::diagnostic_c_tests): Call it.
9572 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9574 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
9575 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
9576 (class selftest::test_diagnostic_context): Move to...
9577 * selftest-diagnostic.c: New file.
9578 * selftest-diagnostic.h: New file.
9580 2017-11-02 James Bowman <james.bowman@ftdichip.com>
9582 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
9583 offset range for FT32B.
9584 * config/ft32/ft32.h: option "mcompress" enables relaxation.
9585 * config/ft32/ft32.md: Add TARGET_NOPM.
9586 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
9587 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
9589 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
9591 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
9593 2017-11-02 Jeff Law <law@redhat.com>
9595 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
9596 virtual keyword on FINAL OVERRIDE members.
9598 * tree-ssa-propagate.h (ssa_propagation_engine): Group
9599 virtuals together. Add virtual destructor.
9600 (substitute_and_fold_engine): Similarly.
9602 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
9604 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
9606 2017-11-02 Richard Biener <rguenther@suse.de>
9608 PR tree-optimization/82795
9609 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
9611 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9613 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
9615 (gcc_GAS_CHECK_FEATURE): Remove.
9616 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
9617 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
9618 * configure: Regenerate.
9620 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
9622 * config/arc/arc.c (hwloop_optimize): Account for empty
9625 2017-11-02 Richard Biener <rguenther@suse.de>
9628 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
9629 Truncate ARRAY_REF index and element size.
9631 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9633 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
9635 2017-11-01 Jeff Law <law@redhat.com>
9637 * tree-ssa-ccp.c (ccp_folder): New class derived from
9638 substitute_and_fold_engine.
9639 (ccp_folder::get_value): New member function.
9640 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
9641 (ccp_fold_stmt): Remove prototype.
9642 (ccp_finalize): Call substitute_and_fold from the ccp_class.
9643 * tree-ssa-copy.c (copy_folder): New class derived from
9644 substitute_and_fold_engine.
9645 (copy_folder::get_value): Renamed from get_value.
9646 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
9647 * tree-vrp.c (vrp_folder): New class derived from
9648 substitute_and_fold_engine.
9649 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
9650 (vrp_folder::get_value): New member function.
9651 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
9652 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
9653 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
9654 provide a class interface to folder/substitute routines.
9655 (ssa_prop_fold_stmt_fn): Remove typedef.
9656 (ssa_prop_get_value_fn): Likewise.
9657 (subsitute_and_fold): Remove prototype.
9658 (replace_uses_in): Likewise.
9659 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
9660 Renamed from replace_uses_in. Call the virtual member function
9661 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
9662 (substitute_and_fold_dom_walker): Remove initialization of
9663 data member entries for calbacks. Add substitute_and_fold_engine
9664 member and initialize it.
9665 (substitute_and_fold_dom_walker::before_dom_children0: Use the
9666 member functions for get_value, replace_phi_args_in c
9667 replace_uses_in, and fold_stmt calls.
9668 (substitute_and_fold_engine::substitute_and_fold): Renamed from
9669 substitute_and_fold. Remove assert. Update ctor call.
9671 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
9672 (ssa_prop_visit_phi_fn): Likewise.
9673 (class ssa_propagation_engine): New class to provide an interface
9675 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
9677 (ssa_prop_visit_phi): Likewise.
9678 (ssa_propagation_engine::simulate_stmt): Moved into class.
9679 Call visit_phi/visit_stmt from the class rather than via
9680 file scoped static variables.
9681 (ssa_propagation_engine::simulate_block): Moved into class.
9682 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
9683 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
9684 set file scoped statics for the visit_stmt/visit_phi callbacks.
9685 * tree-complex.c (complex_propagate): New class derived from
9686 ssa_propagation_engine.
9687 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
9688 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
9689 (tree_lower_complex): Call ssa_propagate via the complex_propagate
9691 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
9692 ssa_propagation_engine.
9693 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
9694 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
9695 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
9696 * tree-ssa-copy.c (copy_prop): New class derived from
9697 ssa_propagation_engine.
9698 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
9699 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
9700 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
9701 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
9702 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
9703 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
9704 (execute_vrp): Call ssa_propagate from the vrp_prop class.
9706 2017-11-01 Jakub Jelinek <jakub@redhat.com>
9708 PR rtl-optimization/82778
9709 PR rtl-optimization/82597
9710 * compare-elim.c (struct comparison): Add in_a_setter field.
9711 (find_comparison_dom_walker::before_dom_children): Remove killed
9712 bitmap and df_simulate_find_defs call, instead walk the defs.
9713 Compute last_setter and initialize in_a_setter. Merge definitions
9714 with first initialization for a few variables.
9715 (try_validate_parallel): Use insn_invalid_p instead of
9716 recog_memoized. Return insn rather than just the pattern.
9717 (try_merge_compare): Fix up comment. Don't uselessly test if
9718 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
9720 (execute_compare_elim_after_reload): Remove df_chain_add_problem
9723 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9724 Alan Hayward <alan.hayward@arm.com>
9725 David Sherwood <david.sherwood@arm.com>
9727 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
9728 aarch64_hard_regno_nregs to get the number of registers
9731 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9732 Alan Hayward <alan.hayward@arm.com>
9733 David Sherwood <david.sherwood@arm.com>
9735 * config/aarch64/constraints.md (Upl): Rename to...
9737 * config/aarch64/aarch64.md
9738 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
9741 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9742 Alan Hayward <alan.hayward@arm.com>
9743 David Sherwood <david.sherwood@arm.com>
9745 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
9746 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
9749 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9750 Alan Hayward <alan.hayward@arm.com>
9751 David Sherwood <david.sherwood@arm.com>
9753 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
9754 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
9755 (aarch64_evpc_dup): Generate rtl direcly, rather than using
9757 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
9758 of a single element.
9759 * config/aarch64/iterators.md: Add a comment above the permute
9760 unspecs to say that they are generated directly by
9761 aarch64_expand_vec_perm_const.
9762 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
9764 2017-11-01 Nathan Sidwell <nathan@acm.org>
9766 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
9768 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9770 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
9771 and medany code models, and describe what they do.
9773 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9775 Revert accidental duplicate:
9777 * combine.c (can_change_dest_mode): Reject changes in
9778 REGMODE_NATURAL_SIZE.
9780 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9782 PR rtl-optimization/64682
9783 PR rtl-optimization/69567
9784 PR rtl-optimization/69737
9785 PR rtl-optimization/82683
9786 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9787 register mentioned in the note, drop the note, unless it came from I3,
9788 in which case it should go to I3 again.
9790 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9792 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9793 and return false if not.
9794 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9796 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9798 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9799 range is known to be empty.
9801 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9802 Alan Hayward <alan.hayward@arm.com>
9803 David Sherwood <david.sherwood@arm.com>
9805 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9806 and CONST_VECTOR_NUNITS instead of computing the number of units from
9807 the byte sizes of the vector and element.
9808 (simplify_binary_operation_1): Likewise.
9809 (simplify_const_binary_operation): Likewise.
9810 (simplify_ternary_operation): Likewise.
9812 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9813 Alan Hayward <alan.hayward@arm.com>
9814 David Sherwood <david.sherwood@arm.com>
9816 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9817 (int_mem_offset): ...this new function.
9818 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9819 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9820 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
9823 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9824 Alan Hayward <alan.hayward@arm.com>
9825 David Sherwood <david.sherwood@arm.com>
9827 * lower-subreg.c (interesting_mode_p): New function.
9828 (compute_costs, find_decomposable_subregs, decompose_register)
9829 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9830 (resolve_clobber, dump_choices): Use it.
9832 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9833 Alan Hayward <alan.hayward@arm.com>
9834 David Sherwood <david.sherwood@arm.com>
9836 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
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 * alias.c (find_base_value, find_base_term): Only process integer
9843 truncations. Check the precision rather than the size.
9845 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9846 Alan Hayward <alan.hayward@arm.com>
9847 David Sherwood <david.sherwood@arm.com>
9849 * machmode.h (is_narrower_int_mode): New function
9850 * optabs.c (expand_float, expand_fix): Use it.
9851 * dwarf2out.c (rotate_loc_descriptor): Likewise.
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 * rtl.h (narrower_subreg_mode): New function.
9858 * ira-color.c (update_costs_from_allocno): Use it.
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 * optabs-query.h (convert_optab_p): New function, split out from...
9865 (convert_optab_handler): ...here.
9866 (widening_optab_handler): Delete.
9867 (find_widening_optab_handler): Remove permit_non_widening parameter.
9868 (find_widening_optab_handler_and_mode): Likewise. Provide an
9869 override that operates on mode class wrappers.
9870 * optabs-query.c (widening_optab_handler): Delete.
9871 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9872 parameter. Assert that the two modes are the same class and that
9873 the "from" mode is narrower than the "to" mode. Use
9874 convert_optab_handler instead of widening_optab_handler.
9875 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9876 instead of widening_optab_handler.
9877 * expr.c (expand_expr_real_2): Update calls to
9878 find_widening_optab_handler.
9879 * optabs.c (expand_widen_pattern_expr): Likewise.
9880 (expand_binop_directly): Take the insn_code as a parameter.
9881 (expand_binop): Only call find_widening_optab_handler for
9882 conversion optabs; use optab_handler otherwise. Update calls
9883 to find_widening_optab_handler and expand_binop_directly.
9884 Use convert_optab_handler instead of widening_optab_handler.
9885 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9886 find_widening_optab_handler and use scalar_mode rather than
9888 (convert_plusminus_to_widen): Likewise.
9890 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9891 Alan Hayward <alan.hayward@arm.com>
9892 David Sherwood <david.sherwood@arm.com>
9894 * machmode.h (fixed_size_mode): New class.
9895 * rtl.h (get_pool_mode): Return fixed_size_mode.
9896 * gengtype.c (main): Add fixed_size_mode.
9897 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9898 (get_raw_arg_mode): Likewise.
9899 * doc/tm.texi: Regenerate.
9900 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9901 * targhooks.c (default_get_reg_raw_mode): Likewise.
9902 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9903 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9904 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9905 (msp430_get_raw_result_mode): Likewise.
9906 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9907 * dbxout.c (dbxout_parms): Require fixed-size modes.
9908 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9909 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9910 * omp-low.c (lower_oacc_reductions): Likewise.
9911 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9912 (simplify_subreg): Update accordingly.
9913 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9914 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9915 that aren't fixed-size.
9916 (get_pool_mode): Return a fixed_size_mode.
9917 (output_constant_pool_2): Take a fixed_size_mode.
9919 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9920 Alan Hayward <alan.hayward@arm.com>
9921 David Sherwood <david.sherwood@arm.com>
9923 * doc/rtl.texi (vec_series): Document.
9924 (const): Say that the operand can be a vec_series.
9925 * rtl.def (VEC_SERIES): New rtx code.
9926 * rtl.h (const_vec_series_p_1): Declare.
9927 (const_vec_series_p): New function.
9928 * emit-rtl.h (gen_const_vec_series): Declare.
9929 (gen_vec_series): Likewise.
9930 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9931 (gen_vec_series): Likewise.
9932 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9933 * simplify-rtx.c (simplify_unary_operation): Handle negations
9935 (simplify_binary_operation_series): New function.
9936 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9937 (test_vector_ops_series): New function.
9938 (test_vector_ops): Call it.
9939 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9940 gen_const_vec_series.
9941 (altivec_lvsr): Likewise.
9942 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9944 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9945 Alan Hayward <alan.hayward@arm.com>
9946 David Sherwood <david.sherwood@arm.com>
9948 * doc/rtl.texi (const): Update description of address constants.
9949 Say that vector constants are allowed too.
9950 * common.md (E, F): Use CONSTANT_P instead of checking for
9952 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9953 checking for CONST_VECTOR.
9954 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9956 * expr.c (expand_expr_real_2): Check for vector modes instead
9957 of checking for CONST_VECTOR.
9958 * rtl.h (const_vec_p): New function.
9959 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9960 (unwrap_const_vec_duplicate): Handle them here too.
9962 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9963 David Malcolm <dmalcolm@redhat.com>
9964 Alan Hayward <alan.hayward@arm.com>
9965 David Sherwood <david.sherwood@arm.com>
9967 * rtl.h (vec_duplicate_p): New function.
9968 * selftest-rtl.c (assert_rtx_eq_at): New function.
9969 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9970 (assert_rtx_eq_at): Declare.
9971 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9972 * selftest-run-tests.c (selftest::run_tests): Call it.
9973 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9974 (simplify_unary_operation_1): Recursively handle vector duplicates.
9975 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9977 (simplify_subreg): Handle subregs of vector duplicates.
9978 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9979 (selftest::simplify_rtx_c_tests): New functions.
9981 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9982 Alan Hayward <alan.hayward@arm.com>
9983 David Sherwood <david.sherwood@arm.com>
9985 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9986 (gen_vec_duplicate): Likewise.
9987 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9989 (gen_const_vector): ...here.
9990 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9991 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9992 whose elements are all equal.
9993 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9994 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9995 (simplify_relational_operation): Likewise.
9996 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9998 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9999 (aarch64_expand_vector_init): Likewise.
10000 * config/arm/arm.c (neon_vdup_constant): Likewise.
10001 (neon_expand_vector_init): Likewise.
10002 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
10003 (arm_block_set_unaligned_vect): Likewise.
10004 (arm_block_set_aligned_vect): Likewise.
10005 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
10006 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
10007 (expand_vec_perm_even_odd_pack): Likewise.
10008 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
10009 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
10010 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
10011 gen_const_vec_duplicate.
10012 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
10013 * config/mips/mips.c (mips_gen_const_int_vector): Use
10014 gen_const_vec_duplicate.
10015 (mips_expand_vector_init): Use CONST0_RTX.
10016 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
10017 (define_split): Use gen_const_vec_duplicate.
10018 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
10019 (define_split): Use gen_const_vec_duplicate.
10020 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
10021 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
10022 * config/spu/spu.c (spu_const): Likewise.
10024 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
10025 Alan Hayward <alan.hayward@arm.com>
10026 David Sherwood <david.sherwood@arm.com>
10028 * combine.c (can_change_dest_mode): Reject changes in
10029 REGMODE_NATURAL_SIZE.
10031 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
10033 * configure.ac (--enable-libssp): New.
10034 (gcc_cv_libc_provides_ssp): Check for explicit setting before
10035 trying to determine target-specific default. Adjust indentation.
10036 * configure: Regenerated.
10037 * doc/install.texi (Configuration): Expand --disable-libssp
10040 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
10042 config/i386/i386.c (ix86_expand_epilogue): Correct stack
10045 2017-10-31 Martin Jambor <mjambor@suse.cz>
10048 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
10050 2017-10-31 David Malcolm <dmalcolm@redhat.com>
10052 * auto-profile.c (autofdo_source_profile::read): Use
10053 UNKNOWN_LOCATION rather than 0.
10054 * diagnostic-core.h (warning_at_rich_loc): Rename to...
10055 (warning_at): ...this overload.
10056 (warning_at_rich_loc_n): Rename to...
10057 (warning_n): ...this overload.
10058 (error_at_rich_loc): Rename to...
10059 (error_at): ...this overload.
10060 (pedwarn_at_rich_loc): Rename to...
10061 (pedwarn): ...this overload.
10062 (permerror_at_rich_loc): Rename to...
10063 (permerror): ...this overload.
10064 (inform_at_rich_loc): Rename to...
10065 (inform): ...this overload.
10066 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
10067 (diagnostic_n_impl_richloc): Rename to...
10068 (diagnostic_n_impl): ...this rich_location *-based decl.
10069 (inform_at_rich_loc): Rename to...
10070 (inform): ...this, and add an assertion.
10071 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
10072 (warning_at_rich_loc): Rename to...
10073 (warning_at): ...this, and add an assertion.
10074 (warning_at_rich_loc_n): Rename to...
10075 (warning_n): ...this, and add an assertion.
10076 (warning_n): Update location_t-based implementation for removal of
10077 location_t-based diagnostic_n_impl.
10078 (pedwarn_at_rich_loc): Rename to...
10079 (pedwarn): ...this, and add an assertion.
10080 (permerror_at_rich_loc): Rename to...
10081 (permerror): ...this, and add an assertion.
10082 (error_n): Update for removal of location_t-based diagnostic_n_impl.
10083 (error_at_rich_loc): Rename to...
10084 (error_at): ...this, and add an assertion.
10085 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
10086 (driver::do_spec_on_infiles): Likewise.
10087 * substring-locations.c (format_warning_va): Update for renaming
10088 of inform_at_rich_loc.
10090 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
10092 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
10093 _Float<N>X built-in functions so that the variant without the
10094 "__builtin_" prefix is only enabled for the GNU C and Objective C
10095 languages when they are in non-strict ANSI/ISO mode.
10096 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
10097 * target.def (floatn_builtin_p): Add a target hook to control
10098 whether _Float<N> and _Float<N>X built-in functions without the
10099 "__builtin_" prefix are enabled, and return true for C and
10100 Objective C in the default hook. Include langhooks.h in
10102 * targhooks.h (default_floatn_builtin_p): Likewise.
10103 * targhooks.c (default_floatn_builtin_p): Likewise.
10104 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
10105 floatn_builtin_p target hook.
10106 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
10108 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
10109 Eric Botcazou <ebotcazou@adacore.com>
10111 PR rtl-optimization/81803
10112 * lra-constraints.c (curr_insn_transform): Also reload the whole
10113 register for a strict subreg no wider than a word if this is for
10114 a WORD_REGISTER_OPERATIONS target.
10116 2017-10-31 Jason Merrill <jason@redhat.com>
10118 * gdbinit.in: Skip over inlines from timevar.h.
10120 2017-10-31 Martin Liska <mliska@suse.cz>
10122 * doc/gcov.texi: Document new option.
10123 * gcov.c (print_usage): Likewise print it.
10124 (process_args): Support the argument.
10125 (format_count): New function.
10126 (format_gcov): Use the function.
10128 2017-10-31 Martin Liska <mliska@suse.cz>
10130 * gcov.c (struct name_map): do not use typedef.
10131 Define operator== and operator<.
10132 (name_search): Remove.
10133 (name_sort): Remove.
10134 (main): Do not allocate names.
10135 (process_file): Add vertical space.
10136 (generate_results): Use std::find.
10137 (release_structures): Do not release memory.
10138 (find_source): Use std::find.
10140 2017-10-31 Martin Liska <mliska@suse.cz>
10142 * gcov.c (struct line_info): Remove it's typedef.
10143 (line_info::line_info): Add proper ctor.
10144 (line_info::has_block): Do not use a typedef.
10145 (struct source_info): Do not use typedef.
10146 (circuit): Likewise.
10147 (get_cycles_count): Likewise.
10148 (output_intermediate_file): Iterate via vector iterator.
10149 (add_line_counts): Use std::vector methods.
10150 (accumulate_line_counts): Likewise.
10151 (output_lines): Likewise.
10153 2017-10-31 Martin Liska <mliska@suse.cz>
10155 * gcov.c (struct source_info): Remove typedef.
10156 (source_info::source_info): Add proper ctor.
10157 (accumulate_line_counts): Use struct, not it's typedef.
10158 (output_gcov_file): Likewise.
10159 (output_lines): Likewise.
10160 (main): Do not allocate an array.
10161 (output_intermediate_file): Use size of vector container.
10162 (process_file): Resize the vector.
10163 (generate_results): Do not preallocate, use newly added vector
10165 (release_structures): Do not release sources.
10166 (find_source): Use vector methods.
10167 (add_line_counts): Do not use typedef.
10169 2017-10-31 Martin Liska <mliska@suse.cz>
10171 * doc/gcov.texi: Document that.
10172 * gcov.c (add_line_counts): Mark lines with a non-executed
10174 (output_line_beginning): Handle such lines.
10175 (output_lines): Pass new argument.
10176 (output_intermediate_file): Print it in intermediate format.
10178 2017-10-31 Martin Liska <mliska@suse.cz>
10180 * color-macros.h: New file.
10181 * diagnostic-color.c: Factor out color related to macros to
10183 * doc/gcov.texi: Document -k option.
10184 * gcov.c (INCLUDE_STRING): Include string.h.
10185 (print_usage): Add -k option.
10186 (process_args): Parse it.
10187 (pad_count_string): New function.
10188 (output_line_beginning): Likewise.
10189 (DEFAULT_LINE_START): New macro.
10190 (output_lines): Support color output.
10192 2017-10-31 Martin Liska <mliska@suse.cz>
10194 PR gcov-profile/82633
10195 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
10196 their interaction with GCOV infrastructure.
10197 * configure.ac: Add -fkeep-{inline,static}-functions to
10199 * configure: Regenerate.
10201 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
10204 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
10206 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
10209 * config/rs6000/rs6000.md (allocate_stack): Force update interval
10210 into a register if it does not fit into an immediate offset field.
10212 2017-10-31 Olivier Hainque <hainque@adacore.com>
10214 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
10216 2017-10-31 Julia Koval <julia.koval@intel.com>
10218 * config.gcc: Add gfniintrin.h.
10219 * config/i386/gfniintrin.h: New.
10220 * config/i386/i386-builtin-types.def
10221 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
10222 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
10223 __builtin_ia32_vgf2p8affineinvqb_v32qi,
10224 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
10225 __builtin_ia32_vgf2p8affineinvqb_v16qi,
10226 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
10227 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
10228 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
10229 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
10230 V64QI_FTYPE_V64QI_V64QI_INT): New types.
10231 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
10232 * config/i386/immintrin.h: Include gfniintrin.h.
10233 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
10235 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
10237 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
10239 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10241 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
10242 (arm_ashldi3_1bit): Remove pattern.
10243 (ashrdi3): Remove shift by 1 expansion.
10244 (arm_ashrdi3_1bit): Remove pattern.
10245 (lshrdi3): Remove shift by 1 expansion.
10246 (arm_lshrdi3_1bit): Remove pattern.
10247 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
10248 cost of ashldi3 by 1.
10249 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
10250 (<shift>di3_neon): Likewise.
10252 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10254 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
10255 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
10256 and (*aarch64_simd_mov<VQ:mode>).
10257 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
10258 pattern alternative.
10259 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
10260 attributes to match pattern alternative.
10262 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
10264 * config.gcc (powerpc*-*-*): Add emmintrin.h.
10265 * config/rs6000/emmintrin.h: New file.
10266 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
10268 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10270 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
10271 * (movdi_vfp_cortexa8): Remove pattern.
10273 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10275 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
10277 (Specific, alpha*-dec-osf5.1): Remove.
10278 (Specific, mips-sgi-irix5): Remove.
10279 (Specific, mips-sgi-irix6): Remove.
10281 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10283 PR middle-end/22141
10284 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
10285 arguments to clear_bit_region_be.
10287 2017-10-30 Jim Wilson <wilson@tuliptree.org>
10289 * gimplify.c: Include memmodel.h.
10291 2017-10-30 Martin Jambor <mjambor@suse.cz>
10293 * omp-grid.c (grid_attempt_target_gridification): Also insert a
10294 condition whether loop should be executed at all.
10296 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
10298 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
10299 gimple folding of vec_madd() intrinsics.
10300 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
10301 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
10302 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
10304 2017-10-30 Richard Biener <rguenther@suse.de>
10306 PR tree-optimization/82762
10308 2017-10-23 Richard Biener <rguenther@suse.de>
10310 PR tree-optimization/82129
10312 2017-08-01 Richard Biener <rguenther@suse.de>
10314 PR tree-optimization/81181
10315 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10316 (compute_antic): ... end of iteration here.
10318 2017-10-30 Joseph Myers <joseph@codesourcery.com>
10320 * doc/invoke.texi (C Dialect Options): Document -std=c17,
10321 -std=iso9899:2017 and -std=gnu17.
10322 * doc/standards.texi (C Language): Document C17 support.
10323 * doc/cpp.texi (Overview): Mention -std=c17.
10324 (Standard Predefined Macros): Document C11 and C17 values of
10325 __STDC_VERSION__. Do not refer to C99 support as incomplete.
10326 * doc/extend.texi (Inline): Do not list individual options for
10327 standards newer than C99.
10328 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
10330 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
10333 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
10335 * asan.c (asan_finish_file): Align asan globals array by shadow
10338 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10340 PR middle-end/22141
10341 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
10342 (struct store_immediate_info): Add bitregion_start and bitregion_end
10344 (store_immediate_info::store_immediate_info): Add brs and bre
10345 arguments and initialize bitregion_{start,end} from those.
10346 (struct merged_store_group): Add bitregion_start, bitregion_end,
10347 align_base and mask fields. Drop unnecessary struct keyword from
10348 struct store_immediate_info. Add do_merge method.
10349 (clear_bit_region_be): Use memset instead of loop storing zeros.
10350 (merged_store_group::do_merge): New method.
10351 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
10352 stores as long as the surrounding bitregions have no gaps.
10353 (merged_store_group::merge_overlapping): Use do_merge.
10354 (merged_store_group::apply_stores): Test that bitregion_{start,end}
10355 is byte aligned, rather than requiring that start and width are
10356 byte aligned. Drop unnecessary struct keyword from
10357 struct store_immediate_info. Allocate and populate also mask array.
10358 Make start of the arrays relative to bitregion_start rather than
10359 start and size them according to bitregion_{end,start} difference.
10360 (struct imm_store_chain_info): Drop unnecessary struct keyword from
10361 struct store_immediate_info.
10362 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
10363 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
10364 struct keyword from struct store_immediate_info.
10365 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
10366 between stores as long as the surrounding bitregions have no gaps.
10368 (struct split_store): Add orig non-static data member.
10369 (split_store::split_store): Initialize orig to false.
10370 (find_constituent_stmts): Return store_immediate_info *, non-NULL
10371 if there is exactly a single original stmt. Change stmts argument
10372 to pointer from reference, if NULL, don't push anything to it. Add
10373 first argument, use it to optimize skipping over orig stmts that
10374 are known to be before bitpos already. Simplify.
10375 (split_group): Return unsigned int count how many stores are or
10376 would be needed rather than a bool. Add allow_unaligned argument.
10377 Change split_stores argument from reference to pointer, if NULL,
10378 only do a dry run computing how many stores would be produced.
10379 Rewritten algorithm to use both alignment and misalign if
10380 !allow_unaligned and handle bitfield stores with gaps.
10381 (imm_store_chain_info::output_merged_store): Set start_byte_pos
10382 from bitregion_start instead of start. Compute allow_unaligned
10383 here, if true, do 2 split_group dry runs to compute which one
10384 produces fewer stores and prefer aligned if equal. Punt if
10385 new count is bigger or equal than original before emitting any
10386 statements, rather than during that. Remove no longer needed
10387 new_ssa_names tracking. Replace num_stmts with
10388 split_stores.length (). Use 32-bit stack allocated entries
10389 in split_stores auto_vec. Try to reuse original store lhs/rhs1
10390 if possible. Handle bitfields with gaps.
10391 (pass_store_merging::execute): Ignore bitsize == 0 stores.
10392 Compute bitregion_{start,end} for the stores and construct
10393 store_immediate_info with that. Formatting fixes.
10395 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
10398 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
10399 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
10401 2017-10-29 Jim Wilson <wilson@tuliptree.org>
10403 * gimplify.c: Include tm_p.h.
10405 * common.opt (gcoff): Re-add as ignored option.
10406 (gcoff1, gcoff2, gcoff3): Likewise.
10408 * Makefile.in (OBJS): Delete sdbout.o.
10409 (GTFILES): Delete $(srcdir)/sdbout.c.
10410 * debug.h: Delete sdb_debug_hooks.
10411 * final.c: Delete sdbout.h include.
10412 (final_scan_insn): Delete SDB_DEBUG check.
10413 (rest_of_clean_state): Likewise.
10414 * output.h: Delete sdb_begin_function_line.
10415 * sdbout.c: Delete.
10416 * sdbout.h: Delete.
10417 * toplev.c: Delete sdbout.h include.
10418 (process_options): Delete SDB_DEBUG check.
10419 * tree-core.h (tree_type_common): Delete pointer field of
10421 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
10422 TYPE_SYMTAB_POINTER.
10423 * tree.h (TYPE_SYMTAB_POINTER): Delete.
10424 (TYPE_SYMTAB_IS_POINTER): Delete.
10425 (TYPE_SYMTAB_IS_DIE): Renumber.
10426 * xcoffout.c: Refer to former sdbout.c file.
10427 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
10429 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
10430 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
10431 (Debugging Options): Delete -gcoff.
10432 (-gstabs): Delete SDB reference.
10434 (-gcoff@var{level}): Delete.
10435 * doc/passes.texi (Debugging information output): Delete SDB and
10436 sdbout.c references.
10437 * doc/tm.texi: Regenerate.
10438 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
10439 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
10441 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
10442 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
10443 (SDB_DEBUGGING_INFO): Delete.
10444 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
10445 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
10446 * target.def (output_source_filename): Delete COFF reference.
10448 * common.opt (gcoff): Delete.
10449 (gxcoff+): Update Negative chain.
10450 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
10452 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
10453 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
10454 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
10455 and SDB references.
10456 (expand_function_start): Change sdb reference to past tense.
10457 (expand_function_end): Change sdb reference to past tense.
10458 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
10459 * opts.c (debug_type_names): Delete coff entry.
10460 (common_handle_option): Delete OPT_gcoff case.
10461 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
10463 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
10464 * config/cris/cris.h: Delete SDB reference in comment.
10465 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
10466 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
10467 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
10468 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
10470 (ix86_expand_prologue): Likewise.
10471 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
10472 * config/ia64/ia64.h: Likewise.
10473 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
10474 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
10476 * config/mmix/mmix.h: Likewise.
10477 * config/nds32/nds32.c: Likewise.
10478 * config/stormy/storym16.h: Likewise.
10479 * config/visium/visium.h: Likewise.
10480 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
10482 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
10484 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
10485 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
10486 FRAME_POINTER_REGNUM point at high end of local var area.
10488 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
10490 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
10491 Move comment around. Do not reset best_edge for a copiable
10492 destination if the copy would cause a partition change.
10493 (better_edge_p): Remove redundant check.
10495 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10497 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
10499 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10501 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
10502 for math functions that have _Float<N> and _Float<N>X variants.
10503 (mathfn_built_in_2): Add support for math functions that have
10504 _Float<N> and _Float<N>X variants.
10505 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
10506 (expand_builtin_mathfn_ternary): Add support for fma with
10507 _Float<N> and _Float<N>X variants.
10508 (expand_builtin): Likewise.
10509 (fold_builtin_3): Likewise.
10510 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
10511 create math function _Float<N> and _Float<N>X variants as external
10513 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
10514 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
10515 the __builtin_ prefix and if not strict ansi, without the prefix.
10516 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
10517 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
10518 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
10519 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
10520 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
10521 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
10522 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
10523 function signatures for fma _Float<N> and _Float<N>X variants.
10524 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
10525 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
10526 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
10527 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
10528 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
10529 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
10530 * gencfn-macros.c (print_case_cfn): Add support for math functions
10531 that have _Float<N> and _Float<N>X variants.
10532 (print_define_operator_list): Likewise.
10533 (fltfn_suffixes): Likewise.
10535 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
10536 for math functions that have _Float<N> and _Float<N>X variants.
10537 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
10538 and _Float<N>X variants.
10539 (COPYSIGN): Likewise.
10542 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
10543 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
10545 (integer_valued_read_call_p): Likewise.
10546 * fold-const-call.c (fold_const_call_ss): Likewise.
10547 (fold_const_call_sss): Add support for copysign, fmin, and fmax
10548 _Float<N> and _Float<N>X variants.
10549 (fold_const_call_ssss): Add support for fma _Float<N> and
10550 _Float<N>X variants.
10551 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
10552 support for copysign and fma _Float<N> and _Float<N>X variants.
10553 (backprop::process_builtin_call_use): Likewise.
10554 * tree-call-cdce.c (can_test_argument_range); Add support for
10555 sqrt _Float<N> and _Float<N>X variants.
10556 (edom_only_function): Likewise.
10557 (get_no_error_domain): Likewise.
10558 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
10559 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
10560 copysign _Float<N> and _Float<N>X variants.
10561 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
10562 handled by machine independent code.
10563 (FMAF128): Likewise.
10564 * doc/cpp.texi (Common Predefined Macros): Document defining
10565 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
10566 fma _Float<N> and _Float<N>X variants.
10568 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10571 * config/i386/i386-modes.def (CCFPU): Remove definition.
10572 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
10573 (ix86_cc_modes_compatible): Ditto.
10574 (ix86_expand_carry_flag_compare): Ditto.
10575 (ix86_expand_int_movcc): Ditto.
10576 (ix86_expand_int_addcc): Ditto.
10577 (ix86_reverse_condition): Ditto.
10578 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
10579 Return true/false for unordered/ordered fp comparisons.
10580 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
10581 (ix86_prepare_fp_compare_args): Update for rename.
10582 (ix86_expand_fp_compare): Update for rename. Generate unordered
10583 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
10584 (ix86_expand_sse_compare_and_jump): Ditto.
10585 * config/i386/predicates.md (fcmov_comparison_operator):
10586 Remove CCFPU mode handling.
10587 (ix86_comparison_operator): Ditto.
10588 (ix86_carry_flag_operator): Ditto.
10589 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
10590 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
10591 (*cmpu<mode>_cc_i387): Ditto.
10592 (FPCMP): Remove mode iterator.
10593 (unord): Remove mode attribute.
10594 (unord_subst): New define_subst transformation
10595 (unord): New define_subst attribute.
10596 (unordered): Ditto.
10597 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
10598 (*cmpi<unord>xf_i387): Ditto.
10599 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
10600 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
10601 using unord_subst transformation.
10602 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
10603 (round_saeonly): Also handle CCFP mode.
10604 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
10605 Remove UNSPEC_SAHF unspec handling.
10607 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10609 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
10611 2017-10-27 Jeff Law <law@redhat.com>
10613 * gimple-ssa-sprintf.c: Include domwalk.h.
10614 (class sprintf_dom_walker): New class, derived from dom_walker.
10615 (sprintf_dom_walker::before_dom_children): New function.
10616 (struct call_info): Moved into sprintf_dom_walker class
10617 (compute_formath_length, handle_gimple_call): Likewise.
10618 (sprintf_length::execute): Call the dominator walker rather
10619 than walking the statements.
10621 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
10622 gimple statement locations.
10623 (check_array_bounds): Corresponding changes. Get the statement's
10624 location directly from wi->stmt.
10626 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
10629 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
10631 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10633 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
10634 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
10636 2017-10-27 Jakub Jelinek <jakub@redhat.com>
10639 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
10640 * config/i386/i386.c (maybe_get_pool_constant): Removed.
10641 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
10642 maybe_get_pool_constant.
10643 * config/i386/predicates.md (zero_extended_scalar_load_operand):
10646 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10648 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
10649 2.26 caveat. Update gas and gld versions.
10650 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
10653 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10655 * cgraph.h (set_malloc_flag): Declare.
10656 * cgraph.c (set_malloc_flag_1): New function.
10657 (set_malloc_flag): Likewise.
10658 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
10659 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
10661 (read_ipa_call_summary): Add support for reading is_return_callee.
10662 (write_ipa_call_summary): Stream is_return_callee.
10663 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
10664 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
10665 ipa-prop.h, ipa-fnsummary.h.
10666 (pure_const_names): Change to static.
10667 (malloc_state_e): Define.
10668 (malloc_state_names): Define.
10669 (funct_state_d): Add field malloc_state.
10670 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
10671 (check_retval_uses): New function.
10672 (malloc_candidate_p): Likewise.
10673 (analyze_function): Add support for malloc attribute.
10674 (pure_const_write_summary): Stream malloc_state.
10675 (pure_const_read_summary): Add support for reading malloc_state.
10676 (dump_malloc_lattice): New function.
10677 (propagate_malloc): New function.
10678 (warn_function_malloc): New function.
10679 (ipa_pure_const::execute): Call propagate_malloc and
10680 ipa_free_fn_summary.
10681 (pass_local_pure_const::execute): Add support for malloc attribute.
10682 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
10683 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
10685 2017-10-27 Martin Liska <mliska@suse.cz>
10687 PR gcov-profile/82457
10688 * doc/invoke.texi: Document that one needs a non-strict ISO mode
10689 for fork-like functions to be properly instrumented.
10691 2017-10-27 Richard Biener <rguenther@suse.de>
10693 PR middle-end/81659
10694 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
10695 info when we redirected EH.
10697 2017-10-26 Michael Collison <michael.collison@arm.com>
10699 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
10701 (<optab>_trunchf<GPI:mode>2: New pattern.
10702 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
10703 * config/aarch64/iterators.md (wv): New mode attribute.
10704 (vf, VF): New mode attributes.
10705 (vgp, VGP): New mode attributes.
10706 (s): Update attribute with SImode and DImode prefixes.
10708 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10710 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
10711 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
10712 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
10713 (nios2_option_overide): Initialize it. Don't allow R0-relative
10714 addressing with PIC.
10715 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
10716 (nios2_symbolic_constant_p): Likewise.
10717 (nios2_legitimate_address_p): Likewise.
10718 (nios2_r0rel_section_name_p): New.
10719 (nios2_symbol_ref_in_r0rel_data_p): New.
10720 (nios2_emit_move_sequence): Handle r0rel_constant_p.
10721 (r0rel_constant_p): New.
10722 (nios2_print_operand_address): Handle r0rel_constant_p.
10723 (nios2_cdx_narrow_form_p): Likewise.
10724 * config/nios2/nios2.opt (mr0rel-sec=): New option.
10725 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
10726 (Nios II Options): Document -mr0rel-sec.
10728 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10730 * config/nios2/nios2.c: Include xregex.h.
10731 (nios2_gprel_sec_regex): New.
10732 (nios2_option_overide): Initialize it. Don't allow GP-relative
10733 addressing with PIC.
10734 (nios2_small_section_name_p): Check for regex match.
10735 * config/nios2/nios2.opt (mgprel-sec=): New option.
10736 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
10737 (Nios II Options): Document -mgprel-sec.
10739 2017-10-26 Jim Wilson <wilson@tuliptree.org>
10741 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
10743 2017-10-26 Tom de Vries <tom@codesourcery.com>
10745 PR tree-optimization/82707
10746 * gimple.c (gimple_copy): Fix unsharing of
10747 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
10749 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
10751 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
10752 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
10753 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
10754 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
10755 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
10756 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
10757 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
10758 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
10759 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
10760 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
10761 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
10762 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
10763 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
10764 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
10765 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
10766 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
10767 _mm512_mask_cmpunord_ps_mask): New intrinsics.
10769 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10771 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
10773 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
10774 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10775 warning to rs6000.c. Remove the Undocumented flag, since it has
10777 (-mabi=ibmlongdouble): Likewise.
10778 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10779 already set, set the default format for long double.
10780 (rs6000_debug_reg_global): Print whether long double is IBM or
10782 (rs6000_option_override_internal): Rework setting long double
10783 format. Only warn if the user is changing the long double default
10784 and they did not use -Wno-psabi.
10785 * doc/invoke.texi (PowerPC options): Update the documentation for
10786 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10788 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10789 Alan Hayward <alan.hayward@arm.com>
10790 David Sherwood <david.sherwood@arm.com>
10792 * rtl.h (wider_subreg_mode): New function.
10793 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10794 rather than an unsigned int *.
10795 * ira-color.c (regno_max_ref_width): Replace with...
10796 (regno_max_ref_mode): ...this new variable.
10797 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10798 Use wider_subreg_mode.
10799 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10800 rather than an unsigned int *.
10801 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10802 (process_alt_operands): Likewise.
10803 (invariant_p): Likewise.
10804 * lra-spills.c (assign_mem_slot): Likewise.
10805 (add_pseudo_to_slot): Likewise.
10806 * lra.c (collect_non_operand_hard_regs): Likewise.
10807 (add_regs_to_insn_regno_info): Likewise.
10808 * reload1.c (regno_max_ref_width): Replace with...
10809 (regno_max_ref_mode): ...this new variable.
10810 (reload): Update accordingly. Update call to
10811 ira_sort_regnos_for_alter_reg.
10812 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10813 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10814 (scan_paradoxical_subregs): Likewise.
10816 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10818 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10819 (aarch64_frame): Add emit_frame_chain boolean.
10820 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10821 Move eh_return case to aarch64_layout_frame.
10822 (aarch64_layout_frame): Initialize emit_frame_chain.
10823 (aarch64_expand_prologue): Use emit_frame_chain.
10825 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10827 * config/aarch64/aarch64.c (aarch64_layout_frame):
10828 Ensure LR is always stored at the bottom of the callee-saves.
10829 Remove rarely used frame layout which saves callee-saves at top of
10830 frame, so the store of LR can be used as a valid probe in all cases.
10832 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10834 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10835 Improve unaligned TImode/TFmode base/offset split.
10837 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10838 Alan Hayward <alan.hayward@arm.com>
10839 David Sherwood <david.sherwood@arm.com>
10841 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10842 * combine.c (find_single_use_1): Likewise.
10843 (expand_field_assignment): Likewise.
10844 (move_deaths): Likewise.
10845 * lra-constraints.c (simplify_operand_subreg): Likewise.
10846 (curr_insn_transform): Likewise.
10847 * lra.c (collect_non_operand_hard_regs): Likewise.
10848 (add_regs_to_insn_regno_info): Likewise.
10849 * rtlanal.c (reg_referenced_p): Likewise.
10850 (covers_regno_no_parallel_p): Likewise.
10852 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10854 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10855 Don't print any bits outside the precision of the value.
10856 * wide-int.cc (test_printing): Add some new tests.
10858 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10860 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10861 supports -xbrace_comment option.
10862 * configure: Regenerate.
10863 * config.in: Regenerate.
10864 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10865 (ASM_CPU_SPEC): Use it.
10867 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10869 * target.def (static_rtx_alignment): New hook.
10870 * targhooks.h (default_static_rtx_alignment): Declare.
10871 * targhooks.c (default_static_rtx_alignment): New function.
10872 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10873 * doc/tm.texi: Regenerate.
10874 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10875 instead of targetm.constant_alignment. Remove call to
10876 set_mem_attributes.
10877 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10878 (cris_preferred_mininum_alignment): New function, split out from...
10879 (cris_constant_alignment): ...here.
10880 (cris_static_rtx_alignment): New function.
10881 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10883 (ix86_constant_alignment): ...here.
10884 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10885 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10886 (mmix_static_rtx_alignment): New function.
10887 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10888 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10890 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10893 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10894 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10896 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10898 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10899 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10900 mark as a sign-extending load.
10901 (local_pic_load_u): Define.
10903 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10905 PR middle-end/82062
10906 * fold-const.c (operand_equal_for_comparison_p): Also return true
10907 if ARG0 is a simple variant of ARG1 with narrower precision.
10908 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10910 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10912 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10914 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10915 scatter_static, scatter_per_elt.
10916 * x86-tune-costs.h: Add new cost entries.
10918 2017-10-25 Richard Biener <rguenther@suse.de>
10920 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10921 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10922 class pass_fre): Move to ...
10923 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10924 class pass_fre): ... here and adjust for statistics.
10926 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10929 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10930 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10931 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10932 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10935 2017-10-25 Richard Biener <rguenther@suse.de>
10937 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10938 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10939 eliminate_push_avail, eliminate_insert): Move inside...
10940 (class eliminate_dom_walker): ... this class in preparation
10942 (fini_eliminate): Remove by merging with ...
10943 (eliminate): ... this function. Adjust for class changes.
10944 (pass_pre::execute): Remove fini_eliminate call.
10945 (pass_fre::execute): Likewise.
10947 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10950 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10951 (VPERMI2, VPERMI2I): New mode iterators.
10952 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10953 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10955 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10956 mode iterator. Remove 3 old define_insn patterns.
10957 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10958 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10959 VPERMI2 mode iterator, remove the other two expanders.
10960 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10961 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10962 instructions, remove the other two patterns.
10963 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10964 mode iterator, remove the other two patterns.
10965 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10966 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10967 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10968 and adjust argument order accordingly.
10969 (ix86_expand_vec_perm): Adjust caller.
10970 (expand_vec_perm_1): Likewise.
10971 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10972 (expand_vec_perm_vpermt2_vpshub2): ... this.
10973 (ix86_expand_vec_perm_const_1): Adjust caller.
10974 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10977 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10978 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10979 (vec_shl_<mode>): Remove unused expander.
10980 (avx512bw_<shift_insn><mode>3): New define_insn.
10981 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10982 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10984 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10987 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10990 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10992 PR rtl-optimization/82396
10993 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10994 (autopref_multipass_init): Simplify initialization.
10995 (autopref_rank_data): Simplify sort order.
10996 * gcc/sched-int.h (autopref_multipass_data_): Remove
10997 multi_mem_insn_p, min_offset and max_offset.
10999 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
11001 PR middle-end/60580
11002 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
11003 Check special value of flag_omit_frame_pointer.
11004 (aarch64_can_eliminate): Likewise.
11005 (aarch64_override_options_after_change_1): Simplify handling of
11006 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
11008 2017-10-24 Richard Biener <rguenther@suse.de>
11010 PR tree-optimization/82697
11011 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
11012 zero for conditional load and unconditional store.
11014 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
11016 * doc/install.texi: Document bootstrap-cet.
11018 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
11021 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
11022 ENDBR instruction at function entrance if function is only
11025 2017-10-24 Jakub Jelinek <jakub@redhat.com>
11028 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
11029 patterns to better describe from which operation the CF is computed.
11030 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
11031 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
11032 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
11033 is 0, use _0 suffixed expanders instead of emitting a comparison
11036 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
11038 * config/i386/i386.md(*movsf_internal, *movdf_internal):
11039 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
11041 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
11043 PR middle-end/82569
11044 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
11045 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
11046 * loop-iv.c (iv_get_reaching_def): Likewise.
11047 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
11048 variable is promoted and the partition contains undefined values.
11050 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11052 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
11054 (nios2_address_cost): Define.
11055 (nios2_legitimize_address): Recognize (exp + constant) directly.
11056 (TARGET_ADDRESS_COST): Define.
11058 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11060 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
11061 (nios2_symbolic_memory_operand_p): Declare.
11062 (nios2_split_large_constant): Declare.
11063 (nios2_split_symbolic_memory_operand): Declare.
11064 * config/nios2/nios2.c: Adjust includes.
11065 (nios2_symbolic_constant_allowed): New.
11066 (nios2_symbolic_constant_p): New.
11067 (nios2_plus_symbolic_constant_p): New.
11068 (nios2_valid_addr_expr_p): Recognize addresses involving
11069 symbolic constants.
11070 (nios2_legitimate_address_p): Likewise, also LO_SUM.
11071 (nios2_symbolic_memory_operand_p): New.
11072 (nios2_large_constant_p): New.
11073 (nios2_split_large_constant): New.
11074 (nios2_split_plus_large_constant): New.
11075 (nios2_split_symbolic_memory_operand): New.
11076 (nios2_legitimize_address): Code refactoring. Handle addresses
11077 involving symbolic constants.
11078 (nios2_emit_move_sequence): Likewise.
11079 (nios2_print_operand): Improve error output.
11080 (nios2_print_operand_address): Handle LO_SUM.
11081 (nios2_cdx_narrow_form_p): Likewise.
11082 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
11083 operands involving symbolic constants.
11084 (movhi_internal, movsi_internal): Likewise.
11085 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
11086 (extendhisi2, extendqi<mode>2): Likewise.
11088 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11090 * tree-pass.h (PROP_rtl_split_insns): Define.
11091 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
11093 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11095 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
11097 2017-10-23 Jakub Jelinek <jakub@redhat.com>
11100 * target.def (const_not_ok_for_debug_p): Default to
11101 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
11102 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
11103 * targhooks.c (default_const_not_ok_for_debug_p): New function.
11104 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
11105 which targetm.const_not_ok_for_debug_p returned true.
11106 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
11108 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
11110 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
11111 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
11112 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
11114 (ix86_const_not_ok_for_debug_p): New function.
11115 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
11116 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
11118 2017-10-23 David Malcolm <dmalcolm@redhat.com>
11121 * system.h: Conditionally include "unique-ptr.h" if
11122 INCLUDE_UNIQUE_PTR is defined.
11123 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
11124 of defining INCLUDE_UNIQUE_PTR before including "system.h".
11126 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
11128 * config/rl78/rl78.md: New define_expand "subdi3".
11130 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
11133 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
11134 DF_REF_INSN if DF_REF_INSN_INFO is false.
11136 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
11138 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
11139 xmm_move instead of sse_move.
11140 (sse_store_index): New function.
11141 (ix86_register_move_cost): Be more sensible about mismatch stall;
11142 model AVX moves correctly; make difference between sse->integer and
11144 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
11145 moves; make difference between SSE and AVX.
11146 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
11147 and zmm_move. Increase size of sse load and store tables;
11148 add unaligned load and store tables; add ssemmx_to_integer.
11149 * x86-tune-costs.h: Update all entries according to real
11150 move latencies from Agner Fog's manual and chip documentation.
11152 2017-10-23 Jakub Jelinek <jakub@redhat.com>
11155 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
11156 * config/i386/constraints.md (Wf): New constraint.
11157 * config/i386/i386.md (UNSPEC_SBB): New unspec.
11158 (cmp<dwi>_doubleword): Removed.
11159 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
11160 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
11161 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
11162 expand with cmp<dwi>_doubleword. For LTU and GEU use
11163 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
11165 * common.opt (gcolumn-info): Enable by default.
11166 * doc/invoke.texi (gcolumn-info): Document new default.
11168 2017-10-23 Richard Biener <rguenther@suse.de>
11170 PR tree-optimization/82672
11171 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
11172 Fold the stmt if we propagated into it.
11174 2017-10-23 Richard Biener <rguenther@suse.de>
11176 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
11177 (bitmap_remove_expr_from_set): ... this. All callers call this
11178 for non-constant values.
11179 (bitmap_set_subtract): Rename to...
11180 (bitmap_set_subtract_expressions): ... this. Adjust and
11182 (bitmap_set_contains_value): Remove superfluous check.
11183 (bitmap_set_replace_value): Inline into single caller ...
11184 (bitmap_value_replace_in_set): ... here and simplify.
11185 (dependent_clean): Merge into ...
11186 (clean): ... this using an overload. Adjust.
11187 (prune_clobbered_mems): Adjust.
11188 (compute_antic_aux): Likewise.
11189 (compute_partial_antic_aux): Likewise.
11191 2017-10-23 Richard Biener <rguenther@suse.de>
11193 PR tree-optimization/82129
11195 2017-08-01 Richard Biener <rguenther@suse.de>
11197 PR tree-optimization/81181
11198 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
11199 (compute_antic): ... end of iteration here.
11201 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11203 * target.def (starting_frame_offset): New hook.
11204 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
11205 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
11206 * doc/tm.texi.in: Regenerate.
11207 * hooks.h (hook_hwi_void_0): Declare.
11208 * hooks.c (hook_hwi_void_0): New function.
11209 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
11210 STARTING_FRAME_OFFSET.
11211 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
11212 * reload1.c (reload): Likewise.
11213 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
11214 instead of STARTING_FRAME_OFFSET.
11215 * function.c (try_fit_stack_local): Likewise.
11216 (assign_stack_local_1): Likewise
11217 (instantiate_virtual_regs): Likewise.
11218 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
11219 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
11220 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
11221 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
11222 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
11223 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
11224 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
11225 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
11226 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
11227 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
11228 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
11229 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
11230 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
11231 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
11232 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
11233 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
11234 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
11235 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
11236 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
11237 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
11238 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
11239 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
11240 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
11241 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
11242 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
11243 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
11244 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
11245 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
11246 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
11247 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
11248 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
11249 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
11250 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
11251 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
11252 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
11253 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
11254 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
11255 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
11256 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
11257 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
11258 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
11259 * config/avr/avr.c (avr_starting_frame_offset): Make static and
11260 return a HOST_WIDE_INT.
11261 (avr_builtin_setjmp_frame_value): Use it instead of
11262 STARTING_FRAME_OFFSET.
11263 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11264 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
11265 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
11267 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11268 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
11269 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
11270 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11271 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
11272 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
11273 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11274 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
11275 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
11276 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11277 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
11278 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
11280 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11281 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
11282 * config/mips/mips.c (mips_compute_frame_info): Refer to
11283 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11284 (mips_starting_frame_offset): New function.
11285 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11286 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
11287 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
11288 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
11289 and return a HOST_WIDE_INT.
11290 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11291 (mmix_initial_elimination_offset): Refer to
11292 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11293 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
11294 * config/pa/pa.c (pa_starting_frame_offset): New function.
11295 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
11296 (pa_expand_prologue): Likewise.
11297 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11298 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
11299 !FRAME_GROWS_DOWNWARD handling to...
11300 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11301 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
11302 !FRAME_GROWS_DOWNWARD handling to...
11303 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11304 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
11305 !FRAME_GROWS_DOWNWARD handling to...
11306 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11307 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
11309 (rs6000_starting_frame_offset): New function.
11310 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
11311 !FRAME_GROWS_DOWNWARD handling to...
11312 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11313 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
11314 !FRAME_GROWS_DOWNWARD handling to...
11315 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11316 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
11317 !FRAME_GROWS_DOWNWARD handling to...
11318 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11319 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
11320 (rs6000_starting_frame_offset): New function.
11321 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
11322 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
11323 * config/vax/vax.c (vax_starting_frame_offset): New function.
11324 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
11325 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11326 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
11327 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
11328 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11329 * system.h (STARTING_FRAME_OFFSET): Poison.
11331 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11333 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11334 SCALAR_TYPE_MODE instead of TYPE_MODE.
11336 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11337 Alan Hayward <alan.hayward@arm.com>
11338 David Sherwood <david.sherwood@arm.com>
11340 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
11342 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11343 Alan Hayward <alan.hayward@arm.com>
11344 David Sherwood <david.sherwood@arm.com>
11346 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
11348 2017-10-23 Richard Biener <rguenther@suse.de>
11350 PR tree-optimization/82129
11351 * tree-ssa-pre.c (bitmap_set_and): Remove.
11352 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
11353 canonicalizing expressions in the set to those with lowest
11354 ID rather than taking that from the first edge.
11356 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11358 * combine.c (rtx_equal_for_field_assignment_p): Use
11359 byte_lowpart_offset.
11361 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11362 Alan Hayward <alan.hayward@arm.com>
11363 David Sherwood <david.sherwood@arm.com>
11365 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
11366 to a SUBREG_PROMOTED_VAR.
11368 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11369 Alan Hayward <alan.hayward@arm.com>
11370 David Sherwood <david.sherwood@arm.com>
11372 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
11373 (expand_debug_source_expr): Likewise.
11374 * combine.c (combine_simplify_rtx): Likewise.
11375 * cse.c (fold_rtx): Likewise.
11376 * optabs.c (expand_float): Likewise.
11377 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11378 (simplify_binary_operation_1): Likewise.
11380 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11381 Alan Hayward <alan.hayward@arm.com>
11382 David Sherwood <david.sherwood@arm.com>
11384 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
11385 (record_promoted_value): Likewise.
11386 * expr.c (expand_expr_real_2): Likewise.
11387 * ree.c (update_reg_equal_equiv_notes): Likewise.
11388 (combine_set_extension): Likewise.
11389 * rtlanal.c (low_bitmask_len): Likewise.
11390 * simplify-rtx.c (neg_const_int): Likewise.
11391 (simplify_binary_operation_1): Likewise.
11393 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11394 Alan Hayward <alan.hayward@arm.com>
11395 David Sherwood <david.sherwood@arm.com>
11397 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
11398 * regcprop.c (maybe_mode_change): Likewise.
11399 * reload1.c (alter_reg): Likewise.
11401 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11403 * inchash.h (inchash::hash::add_wide_int): New function.
11404 * lto-streamer-out.c (hash_tree): Use it.
11406 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11408 * inchash.h (inchash::hash::add_wide_int): Rename to...
11409 (inchash::hash::add_hwi): ...this.
11410 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
11411 (polymorphic_call_target_hasher::hash): Likewise.
11412 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
11413 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
11414 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
11415 * lto-streamer-out.c (hash_tree): Likewise.
11416 * optc-save-gen.awk: Likewise.
11417 * tree.c (add_expr): Likewise.
11419 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11422 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
11423 for ordered inequality comparisons even with TARGET_IEEE_FP.
11425 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11428 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
11429 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11430 Expand with cmp<dwi>_doubleword.
11432 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11434 * extend.texi: Add x86 specific to 'nocf_check' attribute.
11435 List CET intrinsics.
11436 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
11437 specific to -fcf-protection option.
11439 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11441 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
11442 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
11443 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
11444 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
11445 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
11446 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
11447 (extra_objs): Add cet.o for Linux/x86 targets.
11448 (tmake_file): Add i386/t-cet for Linux/x86 targets.
11449 * config/i386/cet.c: New file.
11450 * config/i386/cetintrin.h: Likewise.
11451 * config/i386/t-cet: Likewise.
11452 * config/i386/cpuid.h (bit_SHSTK): New.
11453 (bit_IBT): Likewise.
11454 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
11455 pass IBT and SHSTK bits.
11456 * config/i386/i386-builtin-types.def
11457 (VOID_FTYPE_UNSIGNED_PVOID): New.
11458 (VOID_FTYPE_UINT64_PVOID): Likewise.
11459 * config/i386/i386-builtin.def: Add CET intrinsics.
11460 * config/i386/i386-c.c (ix86_target_macros_internal): Add
11461 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
11462 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
11463 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
11465 * config/i386/i386.c (rest_of_insert_endbranch): New.
11466 (pass_data_insert_endbranch): Likewise.
11467 (pass_insert_endbranch): Likewise.
11468 (make_pass_insert_endbranch): Likewise.
11469 (ix86_notrack_prefixed_insn_p): Likewise.
11470 (ix86_target_string): Add -mibt, -mshstk flags.
11471 (ix86_option_override_internal): Add flag_cf_protection
11473 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
11474 (ix86_print_operand): Add 'notrack' prefix output.
11475 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
11476 (ix86_expand_builtin): Expand CET intrinsics.
11477 (x86_output_mi_thunk): Add 'endbranch' instruction.
11478 * config/i386/i386.h (TARGET_IBT): New.
11479 (TARGET_IBT_P): Likewise.
11480 (TARGET_SHSTK): Likewise.
11481 (TARGET_SHSTK_P): Likewise.
11482 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
11483 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
11484 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
11485 (builtin_setjmp_setup): New pattern.
11486 (builtin_longjmp): Likewise.
11487 (rdssp<mode>): Likewise.
11488 (incssp<mode>): Likewise.
11489 (saveprevssp): Likewise.
11490 (rstorssp): Likewise.
11491 (wrss<mode>): Likewise.
11492 (wruss<mode>): Likewise.
11493 (setssbsy): Likewise.
11494 (clrssbsy): Likewise.
11495 (nop_endbr): Likewise.
11496 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
11498 * config/i386/immintrin.h: Include <cetintrin.h>.
11499 * config/i386/linux-common.h
11500 (file_end_indicate_exec_stack_and_cet): New prototype.
11501 (TARGET_ASM_FILE_END): New.
11503 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11505 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
11506 latencies instead of having separate table; make difference between
11507 integer and float costs.
11508 * i386.h (processor_costs): Remove scalar_stmt_cost,
11509 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
11510 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
11512 * x86-tune-costs.h: Remove entries which has been removed in
11513 procesor_costs from all tables; make cond_taken_branch_cost
11514 and cond_not_taken_branch_cost COST_N_INSNS based.
11516 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11518 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
11520 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11522 * config/i386/i386.md (isa): Remove fma_avx512f.
11523 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
11524 <avx512>_fmadd_<mode>_mask3<round_name>,
11525 <avx512>_fmsub_<mode>_mask<round_name>,
11526 <avx512>_fmsub_<mode>_mask3<round_name>,
11527 <avx512>_fnmadd_<mode>_mask<round_name>,
11528 <avx512>_fnmadd_<mode>_mask3<round_name>,
11529 <avx512>_fnmsub_<mode>_mask<round_name>,
11530 <avx512>_fnmsub_<mode>_mask3<round_name>,
11531 <avx512>_fmaddsub_<mode>_mask<round_name>,
11532 <avx512>_fmaddsub_<mode>_mask3<round_name>,
11533 <avx512>_fmsubadd_<mode>_mask<round_name>,
11534 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
11535 (*vec_widen_umult_even_v16si<mask_name>,
11536 *vec_widen_smult_even_v16si<mask_name>): Likewise.
11537 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
11539 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11541 * extend.texi: Add 'nocf_check' documentation.
11542 * gimple.texi: Add second parameter to
11543 gimple_build_call_from_tree.
11544 * invoke.texi: Add -fcf-protection documentation.
11545 * rtl.texi: Add REG_CALL_NOTRACK documenation.
11547 2017-10-20 Richard Biener <rguenther@suse.de>
11549 PR tree-optimization/82473
11550 * tree-vect-loop.c (vectorizable_reduction): Properly get at
11551 the largest input type.
11553 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11555 * c-attribs.c (handle_nocf_check_attribute): New function.
11556 (c_common_attribute_table): Add 'nocf_check' handling.
11557 * gimple-parser.c: Add second argument NULL to
11558 gimple_build_call_from_tree.
11559 * attrib.c (comp_type_attributes): Check nocf_check attribute.
11560 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
11562 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
11563 * common.opt: Add fcf-protection flag.
11564 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
11565 * flag-types.h: Add enum cf_protection_level.
11566 * gimple.c (gimple_build_call_from_tree): Add second parameter.
11567 Add 'nocf_check' attribute propagation to gimple call.
11568 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
11569 (gimple_build_call_from_tree): Update prototype.
11570 (gimple_call_nocf_check_p): New function.
11571 (gimple_call_set_nocf_check): Likewise.
11572 * gimplify.c: Add second argument to gimple_build_call_from_tree.
11573 * ipa-icf.c: Add nocf_check attribute in statement hash.
11574 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
11575 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
11576 * toplev.c (process_options): Add flag_cf_protection handling.
11578 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11580 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
11582 2017-10-20 Richard Biener <rguenther@suse.de>
11584 PR tree-optimization/82603
11585 * tree-if-conv.c (predicate_mem_writes): Make sure to only
11586 remove false predicated stores.
11588 2017-10-20 Richard Biener <rguenther@suse.de>
11590 * graphite-isl-ast-to-gimple.c
11591 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
11592 Remove return value and simplify, dump copied stmt after lhs
11594 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
11595 Reduce dump verbosity.
11596 (gsi_insert_earliest): Likewise.
11597 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
11598 * graphite.c (print_global_statistics): Adjust dumping.
11599 (print_graphite_scop_statistics): Likewise.
11600 (print_graphite_statistics): Do not dump loops here.
11601 (graphite_transform_loops): But here.
11603 2017-10-20 Nicolas Roche <roche@adacore.com>
11605 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
11606 * configure: Regenerate.
11608 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11611 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
11612 functions when optimizing replace GIMPLE_RETURN stmts with
11613 calls to __builtin_unreachable ().
11616 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
11617 for -fsanitize=thread link of executables.
11618 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
11619 link of executables.
11622 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
11623 New mode iterators.
11624 (<shift_insn><mode>3<mask_name>): Change the last of the 3
11625 define_insns for logical vector shifts to use VI248_AVX512BW
11626 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
11627 condition, useless isa and prefix attributes. Change the first
11628 2 of these define_insns to ...
11629 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
11630 define_insn for avx512vl.
11631 (<shift_insn><mode>3): ... and this, new define_insn without
11632 masking for non-avx512vl.
11635 * config/i386/sse.md (*andnot<mode>3,
11636 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
11637 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
11638 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
11639 not applied use empty suffix even for TARGET_AVX512VL.
11640 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
11641 is applied, supply evex,evex or evex,evex,evex instead of just
11644 2017-10-20 Julia Koval <julia.koval@intel.com>
11646 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
11647 (OPTION_MASK_ISA_GFNI_UNSET): New.
11648 (ix86_handle_option): Handle OPT_mgfni.
11649 * config/i386/cpuid.h (bit_GFNI): New.
11650 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
11651 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
11652 * config/i386/i386.c (ix86_target_string): Add -mgfni.
11653 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
11654 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
11655 * config/i386/i386.opt: Add mgfni.
11657 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
11659 * config/msp430/msp430.c (msp430_option_override): Disable
11660 -fdelete-null-pointer-checks.
11661 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
11663 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11665 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
11666 of x87 and SSE instructions.
11668 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11670 * asan.c (create_cond_insert_point): Do not update edge count.
11671 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
11672 (afdo_propagate_circuit): Likewise.
11673 (afdo_calculate_branch_prob): Likewise.
11674 (afdo_annotate_cfg): Likewise.
11675 * basic-block.h (struct edge_def): Remove count.
11676 (edge_def::count): New accessor.
11677 * bb-reorder.c (rotate_loop): Update.
11678 (find_traces_1_round): Update.
11679 (connect_traces): Update.
11680 (sanitize_hot_paths): Update.
11681 * cfg.c (unchecked_make_edge): Update.
11682 (make_single_succ_edge): Update.
11683 (check_bb_profile): Update.
11684 (dump_edge_info): Update.
11685 (update_bb_profile_for_threading): Update.
11686 (scale_bbs_frequencies_int): Update.
11687 (scale_bbs_frequencies_gcov_type): Update.
11688 (scale_bbs_frequencies_profile_count): Update.
11689 (scale_bbs_frequencies): Update.
11690 * cfganal.c (connect_infinite_loops_to_exit): Update.
11691 * cfgbuild.c (compute_outgoing_frequencies): Update.
11692 (find_many_sub_basic_blocks): Update.
11693 * cfgcleanup.c (try_forward_edges): Update.
11694 (try_crossjump_to_edge): Update
11695 * cfgexpand.c (expand_gimple_cond): Update
11696 (expand_gimple_tailcall): Update
11697 (construct_exit_block): Update
11698 * cfghooks.c (verify_flow_info): Update
11699 (redirect_edge_succ_nodup): Update
11700 (split_edge): Update
11701 (make_forwarder_block): Update
11702 (duplicate_block): Update
11703 (account_profile_record): Update
11704 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11705 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11706 * cfgloopmanip.c (scale_loop_profile): Update.
11708 (lv_adjust_loop_entry_edge): Update.
11709 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11710 (force_nonfallthru_and_redirect): Update.
11711 (purge_dead_edges): Update.
11712 (rtl_flow_call_edges_add): Update.
11713 * cgraphunit.c (init_lowered_empty_function): Update.
11714 (cgraph_node::expand_thunk): Update.
11715 * gimple-pretty-print.c (dump_probability): Update.
11716 (dump_edge_probability): Update.
11717 * gimple-ssa-isolate-paths.c (isolate_path): Update.
11718 * haifa-sched.c (sched_create_recovery_edges): Update.
11719 * hsa-gen.c (convert_switch_statements): Update.
11720 * ifcvt.c (dead_or_predicable): Update.
11721 * ipa-inline-transform.c (inline_transform): Update.
11722 * ipa-split.c (split_function): Update.
11723 * ipa-utils.c (ipa_merge_profiles): Update.
11724 * loop-doloop.c (add_test): Update.
11725 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
11726 * lto-streamer-in.c (input_cfg): Update.
11727 (input_function): Update.
11728 * lto-streamer-out.c (output_cfg): Update.
11729 * modulo-sched.c (sms_schedule): Update.
11730 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11731 * predict.c (maybe_hot_edge_p): Update.
11732 (unlikely_executed_edge_p): Update.
11733 (probably_never_executed_edge_p): Update.
11734 (dump_prediction): Update.
11735 (drop_profile): Update.
11736 (propagate_unlikely_bbs_forward): Update.
11737 (determine_unlikely_bbs): Update.
11738 (force_edge_cold): Update.
11739 * profile.c (compute_branch_probabilities): Update.
11740 * reg-stack.c (better_edge): Update.
11741 * shrink-wrap.c (handle_simple_exit): Update.
11742 * tracer.c (better_p): Update.
11743 * trans-mem.c (expand_transaction): Update.
11744 (split_bb_make_tm_edge): Update.
11745 * tree-call-cdce.c: Update.
11746 * tree-cfg.c (gimple_find_sub_bbs): Update.
11747 (gimple_split_edge): Update.
11748 (gimple_duplicate_sese_region): Update.
11749 (gimple_duplicate_sese_tail): Update.
11750 (gimple_flow_call_edges_add): Update.
11751 (insert_cond_bb): Update.
11752 (execute_fixup_cfg): Update.
11753 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
11754 * tree-complex.c (expand_complex_div_wide): Update.
11755 * tree-eh.c (lower_resx): Update.
11756 (unsplit_eh): Update.
11757 (cleanup_empty_eh_move_lp): Update.
11758 * tree-inline.c (copy_edges_for_bb): Update.
11759 (freqs_to_counts): Update.
11760 (copy_cfg_body): Update.
11761 * tree-ssa-dce.c (remove_dead_stmt): Update.
11762 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11763 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11764 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11765 (unloop_loops): Update.
11766 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11767 * tree-ssa-loop-split.c (connect_loops): Update.
11768 (split_loop): Update.
11769 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11770 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11771 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11772 * tree-ssa-reassoc.c (branch_fixup): Update.
11773 * tree-ssa-tail-merge.c (replace_block_by): Update.
11774 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11775 (compute_path_counts): Update.
11776 (update_profile): Update.
11777 (recompute_probabilities): Update.
11778 (update_joiner_offpath_counts): Update.
11779 (estimated_freqs_path): Update.
11780 (freqs_to_counts_path): Update.
11781 (clear_counts_path): Update.
11782 (ssa_fix_duplicate_block_edges): Update.
11783 (duplicate_thread_path): Update.
11784 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11785 (case_bit_test_cmp): Update.
11786 (collect_switch_conv_info): Update.
11787 (gen_inbound_check): Update.
11788 (do_jump_if_equal): Update.
11789 (emit_cmp_and_jump_insns): Update.
11790 * tree-tailcall.c (decrease_profile): Update.
11791 (eliminate_tail_call): Update.
11792 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11793 (vect_do_peeling): Update.
11794 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11795 * ubsan.c (ubsan_expand_null_ifn): Update.
11796 (ubsan_expand_ptr_ifn): Update.
11797 * value-prof.c (gimple_divmod_fixed_value): Update.
11798 (gimple_mod_pow2): Update.
11799 (gimple_mod_subtract): Update.
11800 (gimple_ic): Update.
11801 (gimple_stringop_fixed_value): Update.
11803 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11806 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11808 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11810 PR rtl-optimization/82395
11811 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11812 based on non_spilled_static_chain_regno_p.
11814 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11816 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11817 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11818 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11820 2017-10-19 Martin Sebor <msebor@redhat.com>
11822 PR tree-optimization/82596
11823 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11825 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11827 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11828 (get_mem_refs_of_builtin_call): Likewise.
11829 * builtins.c (expand_builtin_apply): Adjust call to
11830 allocate_dynamic_stack_space.
11831 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11832 the third argument to allocate_dynamic_stack_space, otherwise -1.
11833 (expand_builtin): Deal with all alloca variants.
11834 (is_inexpensive_builtin): Likewise.
11835 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11836 * calls.c (special_function_p): Deal with all alloca variants.
11837 (initialize_argument_information): Adjust call to
11838 allocate_dynamic_stack_space.
11839 (expand_call): Likewise.
11840 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11841 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11842 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11843 use it for the stack usage computation.
11844 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11845 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11846 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11847 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11848 (in_loop_p): Remove first argument and useless check.
11849 (pass_walloca::execute): Remove useless test and adjust call to above.
11850 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11851 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11852 (gimplify_call_expr): Deal with all alloca variants.
11853 * hsa-gen.c (gen_hsa_alloca): Likewise.
11854 (gen_hsa_insns_for_call): Likewise.
11855 * ipa-pure-const.c (special_builtin_state): Likewise.
11856 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11857 * tree-object-size.c (alloc_object_size): Likewise.
11858 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11859 (call_may_clobber_ref_p_1): Likewise.
11860 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11861 (ccp_fold_stmt): Likewise.
11862 (optimize_stack_restore): Likewise.
11863 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11864 (mark_all_reaching_defs_necessary_1): Likewise.
11865 (propagate_necessity): Likewise.
11866 (eliminate_unnecessary_stmts): Likewise.
11867 * tree.c (build_common_builtin_nodes): Build
11868 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11869 (build_alloca_call_expr): New function.
11870 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11871 (CASE_BUILT_IN_ALLOCA): Likewise.
11872 (build_alloca_call_expr): Declare.
11873 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11875 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11878 * dwarf2out.c (new_die_raw): New static inline function.
11879 (new_die): Use it to create the DIE.
11880 (add_AT_external_die_ref): Likewise.
11881 (clone_die): Likewise.
11882 (clone_as_declaration): Likewise.
11883 (dwarf2out_vms_debug_main_pointer): Likewise.
11884 (base_type_die): Likewise. Remove early return for corner cases.
11885 Do not call add_pubtype on the DIE here.
11886 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11887 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11888 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11889 native order exists for base types, attach the DIE manually and call
11890 add_pubtype on it. Do not equate a reverse order DIE to the type.
11892 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11894 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11895 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11897 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11899 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11901 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11902 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11903 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11904 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11905 * opts.c (sanitizer_opts): Add builtin.
11906 * ubsan.c (instrument_builtin): New function.
11907 (pass_ubsan::execute): Call it.
11908 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11909 * doc/invoke.texi: Document -fsanitize=builtin.
11911 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11912 builtins, store max (log2 (align), 0) into uchar field instead of
11913 align into uptr field.
11914 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11915 store uchar 0 field instead of uptr 0 field.
11916 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11917 instead of passing one address of struct with 2 locations pass
11918 two addresses of structs with 1 location each.
11919 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11920 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11921 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11922 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11923 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11924 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11925 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11926 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11928 2017-10-19 Martin Liska <mliska@suse.cz>
11931 * file-find.c (remove_prefix): Remove.
11932 * file-find.h (remove_prefix): Likewise.
11933 * gcc-ar.c: Remove smartness of lookup.
11935 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11937 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11938 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11939 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11941 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11944 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11945 (setcc + and to xor + setcc): New peephole2.
11947 2017-10-19 Tom de Vries <tom@codesourcery.com>
11949 * doc/sourcebuild.texi (Test Directives, Variants of
11950 dg-require-support): Add dg-require-stack-size.
11952 2017-10-19 Martin Liska <mliska@suse.cz>
11955 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11956 that have a large alignment.
11957 (gimplify_target_expr): Likewise.
11959 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11961 PR rtl-optimization/82602
11962 * ira.c (rtx_moveable_p): Return false for volatile asm.
11964 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11967 * config/i386/i386-modes.def (CCGZ): New CC mode.
11968 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11969 * config/i386/predicates.md (ix86_comparison_operator):
11971 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11972 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11973 with double-word subtraction.
11974 (put_condition_code): Handle CCGZmode.
11976 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11978 * wide-int.cc (debug (const wide_int &)): New.
11979 (debug (const wide_int *)): New.
11980 (debug (const widest_int &)): New.
11981 (debug (const widest_int *)): New.
11983 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11985 PR middle-end/82556
11986 * lra-constraints.c (curr_insn_transform): Use non-input operand
11987 instead of output one for matched reload.
11989 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11991 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11992 (tree-ssa-loop-ivopts.h): New header file.
11993 (struct builtin_info): New fields.
11994 (classify_builtin_1): Compute and record base and offset parts for
11995 memset builtin partition by calling strip_offset.
11996 (offset_cmp, fuse_memset_builtins): New functions.
11997 (finalize_partitions): Fuse adjacent memset partitions by calling
11999 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
12000 Expose the interface.
12001 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
12003 2017-10-18 Bin Cheng <bin.cheng@arm.com>
12005 PR tree-optimization/82574
12006 * tree-loop-distribution.c (find_single_drs): New parameter. Check
12007 that data reference must be executed exactly once per iteration
12008 against the outermost loop in nest.
12009 (classify_partition): Update call to above function.
12011 2017-10-18 Richard Biener <rguenther@suse.de>
12013 PR tree-optimization/82591
12014 * graphite.c (graphite_transform_loops): Move code gen message
12016 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12017 Here. Handle scop_to_isl_ast failing.
12018 (scop_to_isl_ast): Limit the number of ISL operations.
12020 2017-10-18 Richard Biener <rguenther@suse.de>
12022 * graphite-isl-ast-to-gimple.c
12023 (translate_isl_ast_to_gimple::set_rename): Simplify.
12024 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
12025 (graphite_copy_stmts_from_block): ... here.
12026 (copy_bb_and_scalar_dependences): Simplify.
12027 (add_parameters_to_ivs_params): Canonicalize.
12028 (generate_entry_out_of_ssa_copies): Simplify.
12029 * graphite-sese-to-poly.c (extract_affine_name): Simplify
12030 by passing in ISL dimension.
12031 (parameter_index_in_region_1): Rename to ...
12032 (parameter_index_in_region): ... this.
12033 (extract_affine): Adjust assert, pass down parameter index.
12034 (add_param_constraints): Use range-info when available.
12035 (build_scop_context): Adjust.
12036 * sese.c (new_sese_info): Adjust.
12037 (free_sese_info): Likewise.
12038 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
12039 Remove unused typedefs.
12040 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
12042 2017-10-18 Martin Liska <mliska@suse.cz>
12044 * combine.c (simplify_compare_const): Add gcc_fallthrough.
12046 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
12048 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
12049 (s390_sched_init): Do not reset s390_sched_state if we entered the
12050 current basic block via a fallthru edge and all others are unlikely.
12052 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
12054 * config/s390/s390.c (NUM_SIDES): New variable.
12055 (LONGRUNNING_THRESHOLD): New variable.
12056 (LATENCY_FACTOR): New variable.
12057 (s390_sched_score): Decrease score for long-running instructions on
12059 (s390_sched_variable_issue): Perform bookkeeping for long-running
12062 2017-10-18 Richard Biener <rguenther@suse.de>
12064 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
12065 Simplify with removal of the parameter rename map.
12066 (set_rename): Likewise.
12067 (should_copy_to_new_region): Likewise.
12068 (graphite_copy_stmts_from_block): Likewise.
12069 (copy_bb_and_scalar_dependences): Remove initialization of
12070 unused copied_bb_map.
12071 (copy_def): Remove.
12072 (copy_internal_parameters): Likewise.
12073 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
12074 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
12075 Use INTEGRAL_TYPE_P.
12076 (parameter_index_in_region_1): Rename to ...
12077 (assign_parameter_index_in_region): ... this. Assert we have
12078 a parameter we handle.
12079 (scan_tree_for_params): Adjust.
12080 * sese.h (parameter_rename_map_t): Remove.
12081 (struct sese_info_t): Remove unused parameter_rename_map and
12082 copied_bb_map members.
12083 * sese.c (new_sese_info): Adjust.
12084 (free_sese_info): Likewise.
12086 2017-10-18 Martin Liska <mliska@suse.cz>
12089 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
12090 on an abnormal edge.
12092 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
12094 * doc/invoke.texi (ffunction-sections and fdata-sections):
12097 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
12099 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
12100 the use statement can throw internally.
12102 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
12104 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
12105 any RTX present on the RHS of a SET.
12106 * compare-elim.c (try_eliminate_compare): Restore comment.
12108 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12110 * langhooks.h (struct lang_hooks): Document that tree_size langhook
12111 may be also called on tcc_type nodes.
12112 * langhooks.c (lhd_tree_size): Likewise.
12114 2017-10-17 David Malcolm <dmalcolm@redhat.com>
12116 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
12117 format_warning_at_substring.
12118 (maybe_warn): Convert source_range * param to a location_t. Pass
12119 UNKNOWN_LOCATION rather than NULL to fmtwarn.
12120 (format_directive): Remove code to extract source_ranges and
12121 source_range * in favor of just a location_t.
12122 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
12124 * substring-locations.c (format_warning_va): Convert
12125 source_range * param to a location_t.
12126 (format_warning_at_substring): Likewise.
12127 * substring-locations.h (format_warning_va): Likewise.
12128 (format_warning_at_substring): Likewise.
12130 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
12132 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
12134 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
12135 and scatter/gather ops.
12137 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
12138 vec_gather_load and vec_scatter_store.
12139 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
12140 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
12141 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
12142 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
12143 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
12144 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
12146 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
12148 * reg-stack.c (compare_for_stack_reg): Add bool argument.
12149 Detect FTST instruction and handle its register pops. Only pop
12150 second operand if can_pop_second_op is true.
12151 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
12152 set can_pop_second_op to false in the compare_for_stack_reg call.
12154 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
12155 output_fp_compare for stack register operands.
12156 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
12157 instructions here. Do not emit stack register pops here. Assert
12158 that FCOMPP pops next to top stack register. Rewrite function.
12160 2017-10-17 Nathan Sidwell <nathan@acm.org>
12162 PR middle-end/82577
12163 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
12164 use DECL_ASSEMBLER_NAME_RAW.
12166 PR middle-end/82546
12167 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
12170 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
12171 Wilco Dijkstra <wilco.dijkstra@arm.com>
12173 * builtins.c (expand_builtin_update_setjmp_buf): Add a
12174 converstion to Pmode from the buf_addr.
12176 2017-10-17 Richard Biener <rguenther@suse.de>
12178 * graphite-dependences.c (scop_get_reads_and_writes): Change
12179 output parameters to references.
12181 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
12183 PR 71026/tree-optimization
12184 * fold-const.c (distribute_real_division): Removed.
12185 (fold_binary_loc): Remove calls to distribute_real_divison.
12187 2017-10-17 Richard Biener <rguenther@suse.de>
12189 * graphite-scop-detection.c
12190 (scop_detection::stmt_has_simple_data_refs_p): Always use
12191 the full nest as region.
12192 (try_generate_gimple_bb): Likewise.
12193 * sese.c (scalar_evolution_in_region): Simplify now that
12194 SCEV can handle instantiation in regions.
12195 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
12196 in the non-loop part of a function if requested.
12198 2017-10-17 Richard Biener <rguenther@suse.de>
12200 PR tree-optimization/82563
12201 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
12203 (graphite_regenerate_ast_isl): Call it.
12204 * graphite-scop-detection.c (build_scops): Remove entry edge split.
12206 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12208 PR tree-optimization/82549
12209 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
12210 Formatting fixes. Instead of calling make_bit_field_ref with negative
12213 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
12215 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
12216 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
12217 _mm_maskz_reduce_ss): New.
12218 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
12219 __builtin_ia32_reducess_mask): Ditto..
12220 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
12221 * config/i386/sse.md (reduces<mode>): Renamed to ...
12222 (reduces<mode><mask_scalar_name>): ... this.
12223 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
12225 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
12226 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
12228 2017-10-16 David Malcolm <dmalcolm@redhat.com>
12230 * Makefile.in (OBJS): Add unique-ptr-tests.o.
12231 * selftest-run-tests.c (selftest::run_tests): Call
12232 selftest::unique_ptr_tests_cc_tests.
12233 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
12234 * unique-ptr-tests.cc: New file.
12236 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
12239 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12241 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12242 (make_hard_regno_born, make_hard_regno_dead): Update
12243 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
12245 2017-10-16 Jeff Law <law@redhat.com>
12247 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12249 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
12251 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
12253 2017-10-16 Olivier Hainque <hainque@adacore.com>
12255 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
12256 with_cpu if we were configured for an e500v2 target cpu name.
12258 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
12260 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
12261 * doc/invoke.texi: Document +nodsp as a valid extension for
12264 2017-10-16 Martin Liska <mliska@suse.cz>
12266 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
12267 (test_set_range): Likewise.
12268 (test_range_functions): Rename to ...
12269 (test_bit_in_range): ... this.
12270 (sbitmap_c_tests): Add new test.
12272 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12274 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
12276 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
12277 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
12279 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12281 * config/aarch64/aarch64-builtins.c
12282 (aarch64_types_quadopu_lane_qualifiers): New.
12283 (TYPES_QUADOPU_LANE): New.
12284 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
12285 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
12286 (aarch64_<sur>dot_laneq<vsi2qi>): New.
12287 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
12288 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
12289 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
12290 (Vdottype, DOTPROD): New.
12291 (sur): Add SDOT and UDOT.
12293 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12295 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
12296 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
12297 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
12298 Add TARGET_DOTPROD.
12299 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
12300 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
12301 Enable TARGET_DOTPROD.
12302 (cortex-a75.cortex-a55): Likewise.
12303 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
12305 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12307 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
12308 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
12310 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
12312 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
12313 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
12314 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
12315 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
12316 * config/arm/types.md (neon_dot, neon_dot_q): New.
12317 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
12319 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12321 * config/arm/arm.h (TARGET_DOTPROD): New.
12322 * config/arm/arm.c (arm_arch_dotprod): New.
12323 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
12324 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
12325 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
12326 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
12327 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
12328 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
12329 * doc/invoke.texi (armv8.2-a): Document dotprod
12331 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12333 * i386.c (ix86_vec_cost): New function.
12334 (ix86_rtx_costs): Handle vector operations better.
12335 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
12336 * x86-tune-costs.h: Add new costs to all tables.
12338 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12340 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
12342 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
12343 divsd, sqrtss and sqrtsd
12344 * x86-tune-costs.h: Add new entries to all costs.
12345 (znver1_cost): Fix to match real instruction latencies.
12347 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12348 Michael Collison <michael.collison@arm.com>
12350 * compare-elim.c: Include emit-rtl.h.
12351 (can_merge_compare_into_arith): New function.
12352 (try_validate_parallel): Likewise.
12353 (try_merge_compare): Likewise.
12354 (try_eliminate_compare): Call the above when no previous clobber
12356 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
12359 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12361 PR middle-end/62263
12362 PR middle-end/82498
12363 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
12364 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
12366 PR middle-end/62263
12367 PR middle-end/82498
12368 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
12369 to be any operand_equal_p operands. For & (B - 1) require
12370 B to be power of 2. Recognize
12371 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
12373 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12376 * optabs.c (expand_memory_blockage): Fix call of
12377 targetm.have_memory_blockage.
12379 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12382 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
12383 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
12384 objects to extra_objs instead of overwriting it.
12386 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12388 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
12389 Use any_fp_register_operand as operand[3] predicate. Simplify
12390 equality test for operands[2] and operands[4] memory location.
12391 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
12392 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
12393 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
12394 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
12395 any_fp_register_operand as operand[1] predicate. Simplify
12396 equality test for operands[0] and operands[3] memory location.
12397 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
12398 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
12399 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
12401 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12403 * target-insns.def: Add memory_blockage.
12404 * optabs.c (expand_memory_blockage): New function.
12405 (expand_asm_memory_barrier): Rename ...
12406 (expand_asm_memory_blockage): ... to this.
12407 (expand_mem_thread_fence): Call expand_memory_blockage
12408 instead of expand_asm_memory_barrier.
12409 (expand_mem_singnal_fence): Ditto.
12410 (expand_atomic_load): Ditto.
12411 (expand_atomic_store): Ditto.
12412 * doc/md.texi (Standard Pattern Names For Generation):
12413 Document memory_blockage instruction pattern.
12415 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
12417 * config/rl78/rl78.c (rl78_emit_libcall): New function.
12418 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
12419 * config/rl78/rl78.md: New define_expand "adddi3".
12421 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12423 * cfghooks.c (verify_flow_info): Disable check that all probabilities
12426 2017-10-13 Jeff Law <law@redhat.com>
12428 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
12430 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12433 * internal-fn.c (expand_mul_overflow): If both operands have
12434 the same highpart of -1 or 0 and the topmost bit of lowpart
12435 is different, overflow is if res <= 0 rather than res < 0.
12437 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
12439 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
12440 TARGET_P9_VECTOR code for unaligned_load case.
12442 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12444 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
12446 2017-10-13 Nathan Sidwell <nathan@acm.org>
12448 * tree-core.h (tree_contains_struct): Make bool.
12449 * tree.c (tree_contains_struct): Likewise.
12450 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
12451 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
12452 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
12453 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
12454 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
12456 2017-10-13 Richard Biener <rguenther@suse.de>
12458 * graphite-isl-ast-to-gimple.c
12459 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
12460 parameters and dominance check.
12461 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
12462 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
12463 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
12464 Do not update SSA form here or do intermediate IL verification.
12465 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
12466 (graphite_initialize): Remove check on the number of loops in
12467 the function and inline into graphite_transform_loops.
12468 (graphite_finalize): Inline into graphite_transform_loops.
12469 (graphite_transform_loops): Perform SSA update and IL verification
12471 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
12473 2017-10-13 Richard Biener <rguenther@suse.de>
12475 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
12476 graphite_expression_type_precision): Avoid global constructor
12478 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
12479 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
12480 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
12481 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
12482 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
12483 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
12485 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
12488 * config/i386/i386.h (ix86_red_zone_size): New.
12489 * config/i386/i386.md (push peephole2s): Replace
12490 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
12492 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12493 Alan Hayward <alan.hayward@arm.com>
12494 David Sherwood <david.sherwood@arm.com>
12496 * combine.c (can_change_dest_mode): Reject changes in
12497 REGMODE_NATURAL_SIZE.
12499 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12500 Alan Hayward <alan.hayward@arm.com>
12501 David Sherwood <david.sherwood@arm.com>
12503 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
12504 (expand_debug_source_expr): Likewise.
12505 * combine.c (combine_simplify_rtx): Likewise.
12506 * cse.c (fold_rtx): Likewise.
12507 * fwprop.c (canonicalize_address): Likewise.
12508 * targhooks.c (default_shift_truncation_mask): Likewise.
12510 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12511 Alan Hayward <alan.hayward@arm.com>
12512 David Sherwood <david.sherwood@arm.com>
12514 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
12515 (widened_mode): Likewise.
12516 (expand_unop): Likewise.
12517 * ree.c (transform_ifelse): Likewise.
12518 (merge_def_and_ext): Likewise.
12519 (combine_reaching_defs): Likewise.
12520 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12522 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12523 Alan Hayward <alan.hayward@arm.com>
12524 David Sherwood <david.sherwood@arm.com>
12526 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
12527 * combine.c (gen_lowpart_for_combine): Likewise.
12528 * dwarf2out.c (rtl_for_decl_location): Likewise.
12529 * final.c (alter_subreg): Likewise.
12530 * rtlhooks.c (gen_lowpart_general): Likewise.
12531 (gen_lowpart_if_possible): Likewise.
12533 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12534 Alan Hayward <alan.hayward@arm.com>
12535 David Sherwood <david.sherwood@arm.com>
12537 * calls.c (expand_call): Use subreg_lowpart_offset.
12538 * cse.c (cse_insn): Likewise.
12539 * regcprop.c (copy_value): Likewise.
12540 (copyprop_hardreg_forward_1): Likewise.
12542 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12545 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
12546 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
12547 =Q constraints to +Q and into insn condition add check
12548 that operands[0] and operands[1] are equal.
12549 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
12550 =Q constraints to +Q and into insn condition add check
12551 that operands[0] is equal to either operands[1] or operands[2].
12554 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
12555 instead of handling MINUS_EXPR twice (once for each argument),
12556 canonicalize operand order and handle just once, use rtype where
12557 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
12560 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
12561 any values of __C while still being pattern recognizable as a simple
12562 rotate instruction.
12564 2017-10-13 Richard Biener <rguenther@suse.de>
12566 PR tree-optimization/82451
12568 2017-10-02 Richard Biener <rguenther@suse.de>
12570 PR tree-optimization/82355
12571 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12572 a mapping for the enclosing loop but avoid generating one for
12573 the loop tree root.
12574 (copy_bb_and_scalar_dependences): Remove premature codegen
12575 error on PHIs in blocks duplicated into multiple places.
12576 * graphite-scop-detection.c
12577 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12578 in the region use it as loop and nest to analyze the DR in.
12579 (try_generate_gimple_bb): Likewise.
12580 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12581 (add_loop_constraints): For blocks in a loop not in the region
12582 create a dimension with a single iteration.
12583 * sese.h (gbb_loop_at_index): Remove assert.
12585 * cfgloop.c (loop_preheader_edge): For the loop tree root
12586 return the single successor of the entry block.
12587 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12588 Reset the SCEV hashtable and niters.
12589 * graphite-scop-detection.c
12590 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
12591 assert that we only have POLYNOMIAL_CHREC that vary in loops
12592 contained in the region.
12593 (scop_detection::graphite_can_represent_expr): Adjust.
12594 (scop_detection::stmt_has_simple_data_refs_p): For loops
12595 not in the region set loop to NULL. The nest is now the
12596 entry edge to the region.
12597 (try_generate_gimple_bb): Likewise.
12598 * sese.c (scalar_evolution_in_region): Adjust for
12599 instantiate_scev change.
12600 * tree-data-ref.h (graphite_find_data_references_in_stmt):
12601 Make nest parameter the edge into the region.
12602 (create_data_ref): Likewise.
12603 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
12604 entry edge into a region and adjust instantiate_scev calls.
12605 (create_data_ref): Likewise.
12606 (graphite_find_data_references_in_stmt): Likewise.
12607 (find_data_references_in_stmt): Pass the loop preheader edge
12608 from the nest argument.
12609 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
12610 parameter the edge into the region.
12611 (instantiate_parameters): Use the loop preheader edge as entry.
12612 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
12614 (get_instantiated_value_entry): Make instantiate_below parameter
12615 the edge into the region.
12616 (instantiate_scev_name): Likewise. Adjust dominance checks,
12617 when we cannot use loop-based instantiation instantiate by
12618 walking use-def chains.
12619 (instantiate_scev_poly): Adjust.
12620 (instantiate_scev_binary): Likewise.
12621 (instantiate_scev_convert): Likewise.
12622 (instantiate_scev_not): Likewise.
12623 (instantiate_array_ref): Remove.
12624 (instantiate_scev_3): Likewise.
12625 (instantiate_scev_2): Likewise.
12626 (instantiate_scev_1): Likewise.
12627 (instantiate_scev_r): Do not blindly handle N-operand trees.
12628 Do not instantiate array-refs. Handle all constants and invariants.
12629 (instantiate_scev): Make instantiate_below parameter
12630 the edge into the region.
12631 (resolve_mixers): Use the loop preheader edge for the region
12632 parameter to instantiate_scev_r.
12633 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
12635 2017-10-13 Richard Biener <rguenther@suse.de>
12637 PR tree-optimization/82525
12638 * graphite-isl-ast-to-gimple.c
12639 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
12641 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
12642 Fail code generation when we cannot represent the isl integer.
12643 (binary_op_to_tree): Elide modulo operations that are no-ops
12644 in the type we code generate. Remove now superfluous code
12647 2017-10-13 Richard Biener <rguenther@suse.de>
12649 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
12650 (scop_detection::harmful_loop_in_region): Remove premature
12651 IV type restriction.
12652 (scop_detection::graphite_can_represent_scev): We can handle
12653 pointer IVs just fine.
12655 2017-10-13 Alan Modra <amodra@gmail.com>
12657 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
12658 "Clobbers and Scratch Registers". Add paragraph on
12659 alternative to clobbers for scratch registers and OpenBLAS
12662 2017-10-13 Alan Modra <amodra@gmail.com>
12664 * doc/extend.texi (Clobbers): Correct vax example. Delete old
12665 example of a memory input for a string of known length. Move
12666 commentary out of table. Add a number of new examples
12667 covering array memory inputs.
12669 2017-10-12 Martin Liska <mliska@suse.cz>
12671 PR tree-optimization/82493
12672 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
12673 (test_range_functions): New function.
12674 (sbitmap_c_tests): Likewise.
12675 * selftest-run-tests.c (selftest::run_tests): Run new tests.
12676 * selftest.h (sbitmap_c_tests): New function.
12678 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12680 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12682 * config/rs6000/amo.h: Fix spacing issue.
12684 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12687 * config/i386/i386.md (*ashl<mode>3_mask_1,
12688 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
12689 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
12692 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12694 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
12695 (profile_probability): Set max_probability
12696 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
12698 * profile-count.c (profile_probability::differs_from_p): Do not
12699 rely on max_probaiblity == 10000
12701 2017-10-12 Jeff Law <law@redhat.com>
12703 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
12706 2017-10-12 Martin Sebor <msebor@redhat.com>
12710 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
12711 (handle_alias_pairs): Call it.
12712 * common.opt (-Wattribute-alias): New option.
12713 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
12714 * doc/invoke.texi (-Wattribute-alias): Document.
12716 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
12719 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12721 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12723 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12724 (make_hard_regno_born, make_hard_regno_dead): Update
12725 bb_killed_pseudos and bb_gen_pseudos.
12727 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12729 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
12731 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
12733 * config/alpha/alpha.c (alpha_split_conditional_move):
12734 Use std::swap instead of manually swapping.
12735 (alpha_stdarg_optimize_hook): Ditto.
12736 (alpha_canonicalize_comparison): Ditto.
12738 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12740 * tree-loop-distribution.c (struct builtin_info): New struct.
12741 (struct partition): Refactor fields into struct builtin_info.
12742 (partition_free): Free struct builtin_info.
12743 (build_size_arg_loc, build_addr_arg_loc): Delete.
12744 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
12745 information from struct builtin_info.
12746 (find_single_drs): New function refactored from classify_partition.
12747 Also moved builtin validity checks to this function.
12748 (compute_access_range, alloc_builtin): New functions.
12749 (classify_builtin_st, classify_builtin_ldst): New functions.
12750 (classify_partition): Refactor code into functions find_single_drs,
12751 classify_builtin_st and classify_builtin_ldst.
12752 (distribute_loop): Don't do runtime alias check when distributing
12754 (find_seed_stmts_for_distribution): New function.
12755 (pass_loop_distribution::execute): Refactor code finding seed
12756 stmts into above function. Support distribution for the innermost
12757 two-level loop nest. Adjust dump information.
12759 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12761 * tree-loop-distribution.c: Adjust the general comment.
12762 (NUM_PARTITION_THRESHOLD): New macro.
12763 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
12764 (classify_partition): Skip builtin pattern of loop nest's inner loop.
12765 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
12766 in call to build_partition_graph.
12767 (finalize_partitions): New parameter. Make loop distribution more
12768 conservative by fusing more partitions.
12769 (distribute_loop): Don't do runtime alias check in case of loop nest
12771 (find_seed_stmts_for_distribution): New function.
12772 (prepare_perfect_loop_nest): New function.
12773 (pass_loop_distribution::execute): Refactor code finding seed stmts
12774 and loop nest into above functions. Support loop nest distribution.
12775 Adjust dump information accordingly.
12777 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12779 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12780 and set PTYPE_SEQUENTIAL for merged partition.
12782 2017-10-12 Richard Biener <rguenther@suse.de>
12784 PR tree-optimization/69728
12786 2017-09-19 Richard Biener <rguenther@suse.de>
12788 PR tree-optimization/69728
12789 * graphite-sese-to-poly.c (schedule_error): New global.
12790 (add_loop_schedule): Handle empty domain by failing the
12792 (build_original_schedule): Handle schedule_error.
12794 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12795 domain by returning an unchanged schedule.
12797 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12799 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12800 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12802 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12804 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12807 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12810 * expr.c (store_field): Don't optimize away bitsize == 0 store
12811 from CALL_EXPR with addressable return type.
12813 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12815 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12816 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12817 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12818 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12819 TARGET_ISEL instead of TARGET_ISEL<sel>.
12821 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12823 * config/rs6000/rs6000.c
12824 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12826 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12828 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12830 (reg_or_cint_operand): Fix comment.
12831 (reg_or_zero_operand): New predicate.
12832 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12833 * config/rs6000/rs6000.c (output_isel): Delete.
12834 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12835 instead of reg_or_cint_operand. Output instruction directly (not via
12837 (isel_unsigned_<mode>): Ditto.
12838 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12839 gpc_reg_operand. Add an instruction alternative for this. Output
12840 instruction directly.
12841 (*isel_reversed_unsigned_<mode>): Ditto.
12843 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12845 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12846 (TARGET_CANONICALIZE_COMPARISON): Define.
12848 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12851 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12852 Check whether the dest is REG before adding REG_EQUIV note.
12854 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12857 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12859 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12860 (make_hard_regno_born, make_hard_regno_dead): Update
12861 bb_killed_pseudos and bb_gen_pseudos.
12863 2017-10-11 Nathan Sidwell <nathan@acm.org>
12865 * incpath.h (enum incpath_kind): Name enum, prefix values.
12866 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12867 * incpath.c (heads, tails): Use INC_MAX.
12868 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12869 (merge_include_chains, split_quote_chain,
12870 register_include_chains): Update incpath_kind names.
12871 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12872 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12874 (add_framework_path, darwin_register_objc_includes): Likewise.
12875 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12877 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12879 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12880 Do not use float_operator operator predicate.
12881 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12882 * config/i386/predicates.md (float_operator): Remove predicate.
12884 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12886 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12887 (*jccxf_i387): Ditto.
12888 (*jcc<mode>_i387): Ditto.
12889 (*jccu<mode>_i387): Ditto.
12890 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12891 (*jcc_*_i387 splitters): Remove.
12892 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12893 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12894 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12897 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12899 * profile-count.h (slow_safe_scale_64bit): New function.
12900 (safe_scale_64bit): New inline.
12901 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12902 * profile-count.c: Include wide-int.h
12903 (slow_safe_scale_64bit): New.
12905 2017-10-11 Nathan Sidwell <nathan@acm.org>
12907 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12908 HAS_DECL_ASSEMBLER_NAME_P.
12909 * gimple-expr.c (gimple_decl_printable_name: Check
12910 HAS_DECL_ASSEMBLER_NAME_P too.
12911 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12912 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12913 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12914 * passes.c (rest_of_decl_compilation): Check
12915 HAS_DECL_ASSEMBLER_NAME_P too.
12916 * recog.c (verify_changes): Likewise.
12917 * tree-pretty-print.c (dump_decl_name): Likewise.
12918 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12920 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12921 (SET_DECL_ASSEMBLER_NAME): Use it.
12922 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12923 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12924 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12926 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12928 * config.gcc (i386, x86_64): Add extra objects.
12929 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12930 (ix86_min_insn_size): Declare.
12931 (ix86_issue_rate): Declare.
12932 (ix86_adjust_cost): Declare.
12933 (ia32_multipass_dfa_lookahead): Declare.
12934 (ix86_macro_fusion_p): Declare.
12935 (ix86_macro_fusion_pair_p): Declare.
12936 (ix86_bd_has_dispatch): Declare.
12937 (ix86_bd_do_dispatch): Declare.
12938 (ix86_core2i7_init_hooks): Declare.
12939 (ix86_atom_sched_reorder): Declare.
12940 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12941 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12942 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12943 (rip_relative_addr_p): Rename to ...
12944 (ix86_rip_relative_addr_p): ... this one; export.
12945 (memory_address_length): Update.
12946 (ix86_issue_rate): Move to x86-tune-sched.c.
12947 (ix86_flags_dependent): Move to x86-tune-sched.c.
12948 (ix86_agi_dependent): Move to x86-tune-sched.c.
12949 (exact_dependency_1): Move to x86-tune-sched.c.
12950 (exact_store_load_dependency): Move to x86-tune-sched.c.
12951 (ix86_adjust_cost): Move to x86-tune-sched.c.
12952 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12953 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12954 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12955 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12956 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12957 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12958 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12959 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12960 (min_insn_size): Rename to ...
12961 (ix86_min_insn_size): ... this one; export.
12962 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12963 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12964 (core2i7_first_cycle_multipass_backtrack): Move to
12965 x86-tune-sched-core.c.
12966 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12967 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12968 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12969 (ix86_avoid_jump_mispredicts): Update.
12970 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12971 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12972 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12973 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12974 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12975 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12976 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12977 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12978 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12979 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12980 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12981 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12982 (BIG): Move to ix86-tune-sched-bd.c.
12983 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12984 (enum insn_path): Move to ix86-tune-sched-bd.c.
12985 (get_mem_group): Move to ix86-tune-sched-bd.c.
12986 (is_cmp): Move to ix86-tune-sched-bd.c.
12987 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12988 (is_branch): Move to ix86-tune-sched-bd.c.
12989 (is_prefetch): Move to ix86-tune-sched-bd.c.
12990 (init_window): Move to ix86-tune-sched-bd.c.
12991 (allocate_window): Move to ix86-tune-sched-bd.c.
12992 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12993 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12994 (process_end_window): Move to ix86-tune-sched-bd.c.
12995 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12996 (find_constant): Move to ix86-tune-sched-bd.c.
12997 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12998 (has_immediate): Move to ix86-tune-sched-bd.c.
12999 (get_insn_path): Move to ix86-tune-sched-bd.c.
13000 (get_insn_group): Move to ix86-tune-sched-bd.c.
13001 (count_num_restricted): Move to ix86-tune-sched-bd.c.
13002 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
13003 (add_insn_window): Move to ix86-tune-sched-bd.c.
13004 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
13005 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
13006 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
13007 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
13008 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
13009 (do_dispatch): Move to ix86-tune-sched-bd.c.
13010 (has_dispatch): Move to ix86-tune-sched-bd.c.
13011 * config/i386/t-i386: Add new object files.
13012 * config/i386/x86-tune-costs.h: New file.
13013 * config/i386/x86-tune-sched-atom.c: New file.
13014 * config/i386/x86-tune-sched-bd.c: New file.
13015 * config/i386/x86-tune-sched-core.c: New file.
13016 * config/i386/x86-tune-sched.c: New file.
13018 2017-10-11 Liu Hao <lh_mouse@126.com>
13020 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
13021 the generic version below instead.
13022 (should_colorize): Recognize Windows consoles as terminals
13024 * pretty-print.c [__MINGW32__] (write_all): New function.
13025 [__MINGW32__] (find_esc_head): Likewise.
13026 [__MINGW32__] (find_esc_terminator): Likewise.
13027 [__MINGW32__] (eat_esc_sequence): Likewise.
13028 [__MINGW32__] (mingw_ansi_fputs): New function that handles
13030 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
13033 2017-10-11 Richard Biener <rguenther@suse.de>
13035 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
13036 Properly call analyze_scalar_evolution with the loop of the stmt.
13038 2017-10-11 Richard Biener <rguenther@suse.de>
13040 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
13041 * tree-core.h (tree_base): Add chrec_var union member.
13042 * tree.h (CHREC_VAR): Remove.
13043 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
13044 * tree-chrec.h (build_polynomial_chrec): Adjust.
13045 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
13046 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
13048 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
13050 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
13051 * match.pd: ... here.
13052 ((T) X == (T) Y): Relax condition.
13054 2017-10-11 Bin Cheng <bin.cheng@arm.com>
13056 PR tree-optimization/82472
13057 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
13059 (break_alias_scc_partitions): Update postorder number.
13061 2017-10-11 Martin Liska <mliska@suse.cz>
13064 * opts.c (parse_no_sanitize_attribute): Do not use error_value
13066 * opts.h (parse_no_sanitize_attribute): Remove last argument.
13068 2017-10-11 Martin Liska <mliska@suse.cz>
13070 * print-rtl.c (print_insn): Move declaration of idbuf
13071 to same scope as name.
13073 2017-10-11 Martin Liska <mliska@suse.cz>
13078 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
13079 volatile arguments.
13081 2017-10-11 Martin Liska <mliska@suse.cz>
13084 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
13085 volatile arguments.
13087 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
13089 * config.gcc (default_gnu_indirect_function): Default to yes for
13090 arm*-*-linux* with glibc.
13092 2017-10-11 Richard Biener <rguenther@suse.de>
13094 * tree-scalar-evolution.c (get_scalar_evolution): Handle
13095 default-defs and types we do not want to analyze.
13096 (interpret_loop_phi): Replace unreachable code with an assert.
13097 (compute_scalar_evolution_in_loop): Remove and inline ...
13098 (analyze_scalar_evolution_1): ... here, replacing condition with
13099 what makes the intent clearer. Remove handling of cases
13100 get_scalar_evolution now handles.
13102 2017-10-10 Jim Wilson <wilson@tuliptree.org>
13104 PR rtl-optimization/81434
13105 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
13106 comment for main loop. In sched_group_found if, also add checks for
13107 pass and min_cost_group.
13109 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
13111 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
13112 (rs6000_insn_cost): New function.
13113 * config/rs6000/rs6000.md (cost): New attribute.
13115 2017-10-10 Jakub Jelinek <jakub@redhat.com>
13116 H.J. Lu <hongjiu.lu@intel.com>
13120 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
13121 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
13122 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
13123 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
13124 to OPTION_MASK_ISA_AVX512VL - builtins that have both
13125 OPTION_MASK_ISA_MMX and some other bit set require both
13126 mmx and the ISAs without the mmx bit.
13127 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
13128 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
13129 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
13130 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
13131 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
13132 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
13133 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
13134 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
13135 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
13136 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
13137 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
13138 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
13139 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
13140 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
13141 Add OPTION_MASK_ISA_MMX.
13143 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
13145 * config.gcc (armv7*-*-freebsd*): New target.
13146 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
13148 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
13150 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
13151 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
13154 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
13156 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
13157 a template parameter.
13158 (WIDE_INT_REF_FOR): Update accordingly.
13159 * tree.h (wi::int_traits <const_tree>): Delete.
13160 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
13161 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
13162 (wi::tree_to_wide_ref): New typedef.
13163 (wi::to_wide): New function.
13164 * calls.c (get_size_range): Use wi::to_wide when operating on
13165 trees as wide_ints.
13166 * cgraph.c (cgraph_node::create_thunk): Likewise.
13167 * config/i386/i386.c (ix86_data_alignment): Likewise.
13168 (ix86_local_alignment): Likewise.
13169 * dbxout.c (stabstr_O): Likewise.
13170 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
13171 * expr.c (const_vector_from_tree): Likewise.
13172 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
13173 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
13174 (fold_negate_expr_1, int_const_binop_1, const_binop)
13175 (fold_convert_const_int_from_real, optimize_bit_field_compare)
13176 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
13177 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
13178 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
13179 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
13180 (fold_not_const, round_up_loc): Likewise.
13181 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
13182 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
13183 (alloca_call_type): Likewise.
13184 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
13185 * godump.c (go_output_typedef): Likewise.
13186 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
13187 * internal-fn.c (get_min_precision): Likewise.
13188 * ipa-cp.c (ipcp_store_vr_results): Likewise.
13189 * ipa-polymorphic-call.c
13190 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
13191 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
13192 (ipa_modify_call_arguments): Likewise.
13193 * match.pd: Likewise.
13194 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
13195 * print-tree.c (print_node_brief, print_node): Likewise.
13196 * stmt.c (expand_case): Likewise.
13197 * stor-layout.c (layout_type): Likewise.
13198 * tree-affine.c (tree_to_aff_combination): Likewise.
13199 * tree-cfg.c (group_case_labels_stmt): Likewise.
13200 * tree-data-ref.c (dr_analyze_indices): Likewise.
13201 (prune_runtime_alias_test_list): Likewise.
13202 * tree-dump.c (dequeue_and_dump): Likewise.
13203 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
13204 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
13205 * tree-pretty-print.c (dump_generic_node): Likewise.
13206 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
13207 (simple_iv_with_niters): Likewise.
13208 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
13209 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
13210 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
13211 * tree-ssa-loop-niter.c (split_to_var_and_offset)
13212 (refine_value_range_using_guard, number_of_iterations_ne_max)
13213 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
13214 (get_cst_init_from_scev, record_nonwrapping_iv)
13215 (scev_var_range_cant_overflow): Likewise.
13216 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
13217 * tree-ssa-pre.c (compute_avail): Likewise.
13218 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
13219 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
13220 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
13221 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
13222 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
13223 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
13224 (dump_case_nodes, try_switch_expansion): Likewise.
13225 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
13226 (vect_do_peeling): Likewise.
13227 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
13228 * tree-vect-stmts.c (vectorizable_load): Likewise.
13229 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
13230 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
13231 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
13232 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
13233 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
13234 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
13235 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
13236 (evrp_dom_walker::before_dom_children): Likewise.
13237 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
13238 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
13239 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
13240 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
13241 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
13242 (get_range_pos_neg): Likewise.
13243 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
13244 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
13245 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
13246 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
13247 * config/avr/avr.c (avr_fold_builtin): Likewise.
13248 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
13249 * config/msp430/msp430.c (msp430_attr): Likewise.
13250 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
13251 * config/powerpcspe/powerpcspe-c.c
13252 (altivec_resolve_overloaded_builtin): Likewise.
13253 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
13254 (rs6000_expand_ternop_builtin): Likewise.
13255 * config/rs6000/rs6000-c.c
13256 (altivec_resolve_overloaded_builtin): Likewise.
13257 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
13258 (rs6000_expand_ternop_builtin): Likewise.
13259 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
13261 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13263 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
13264 when copying loop nest with only one inner loop.
13266 2017-10-10 Richard Biener <rguenther@suse.de>
13268 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
13269 blocks if SCEV is active.
13270 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
13272 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
13273 (scev_initialize): Assert we are not yet initialized.
13275 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13277 * tree-loop-distribution.c (generate_loops_for_partition): Remove
13278 inner loop's exit stmt by making it always exit the loop, otherwise
13279 we would generate an infinite empty loop.
13281 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13283 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
13284 renaming variables in new preheader if it's deleted.
13286 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13288 * tree-loop-distribution.c (struct partition): Remove unused field
13289 loops of the structure.
13290 (partition_alloc, partition_free): Ditto.
13291 (build_rdg_partition_for_vertex): Ditto.
13293 2017-10-09 Jeff Law <law@redhat.com>
13295 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
13296 return type to match prototype and documentation.
13298 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13300 * config/rs6000/rs6000.c (processor_costs): Move to ...
13301 * config/rs6000/rs6000.h: ... here.
13302 (rs6000_cost): Declare.
13304 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
13306 * except.c (setjmp_fn): New global variable.
13307 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
13308 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
13309 if DONT_USE_BUILTIN_SETJMP is defined.
13311 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13313 * target.def (insn_cost): New hook.
13314 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
13315 * doc/tm.texi: Regenerate.
13316 * rtlanal.c (insn_cost): Use the new hook.
13318 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13320 * combine.c (combine_validate_cost): Compute the new insn_cost,
13321 not just pattern_cost.
13322 (try_combine): Adjust comment.
13324 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13326 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
13328 * combine.c (uid_insn_cost): Adjust comment.
13329 (combine_validate_cost): Adjust comment. Use pattern_cost instead
13331 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
13332 * dse.c (find_shift_sequence): Ditto.
13333 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
13334 (bb_valid_for_noce_process_p): Use pattern_cost.
13335 * rtl.h (insn_rtx_cost): Delete.
13336 (pattern_cost): New prototype.
13337 (insn_cost): New prototype.
13338 * rtlanal.c (insn_rtx_cost): Rename to...
13339 (pattern_cost): ... this.
13342 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
13344 * config/i386/i386.md (*jcc_2): Remove insn pattern.
13345 (*jcc<mode>_0_r_i387): Ditto.
13346 (*jccxf_r_i387): Ditto.
13347 (*jcc<mode>_r_i387): Ditto.
13348 (*jccu<mode>_r_i387): Ditto.
13349 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
13350 (*jcc): Rename from *jcc_1.
13352 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13354 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
13355 deferred rescans after the lvx/stvx recombination pre-pass.
13357 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13359 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
13360 memory operation instruction support.
13361 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
13362 (rs6000-ibm-aix[789]*): Likewise.
13363 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
13364 Document new functions.
13366 2017-10-09 Richard Biener <rguenther@suse.de>
13368 PR tree-optimization/82397
13369 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
13370 equality only for semantically equal trees.
13372 2017-10-09 Richard Biener <rguenther@suse.de>
13374 PR tree-optimization/82449
13375 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
13376 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
13377 allow constant addresses.
13378 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
13381 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13383 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
13386 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13389 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
13392 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13395 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
13397 2017-10-09 Jakub Jelinek <jakub@redhat.com>
13400 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
13401 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
13403 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
13405 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
13406 (WI_BINARY_PREDICATE_RESULT): Likewise.
13407 (wi::binary_traits::operator_result): New type.
13408 (wi::binary_traits::predicate_result): Likewise.
13409 (generic_wide_int::operator~, unary generic_wide_int::operator-)
13410 (generic_wide_int::operator==, generic_wide_int::operator!=)
13411 (generic_wide_int::operator&, generic_wide_int::and_not)
13412 (generic_wide_int::operator|, generic_wide_int::or_not)
13413 (generic_wide_int::operator^, generic_wide_int::operator+
13414 (binary generic_wide_int::operator-, generic_wide_int::operator*):
13416 (operator~, unary operator-, operator==, operator!=, operator&)
13417 (operator|, operator^, operator+, binary operator-, operator*): New
13419 * expr.c (get_inner_reference): Use wi::bit_and_not.
13420 * fold-const.c (fold_binary_loc): Likewise.
13421 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
13422 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
13423 (bit_value_binop): Likewise.
13424 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
13425 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
13426 (extract_range_from_binary_expr_1): Likewise.
13427 (masked_increment): Likewise.
13428 (simplify_bit_ops_using_ranges): Likewise.
13430 2017-10-09 Martin Jambor <mjambor@suse.cz>
13433 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
13434 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
13435 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
13437 (hsa_fixup_mov_insn_type): New function.
13438 (hsa_op_with_type::get_in_type): Use it.
13439 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
13440 immediates in an assert.
13441 (hsa_op_with_type::extend_int_to_32bit): New method.
13442 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
13443 types. Convert to dest type if necessary.
13444 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
13445 (reg_for_gimple_ssa): Pass false as min32int to
13446 hsa_type_for_scalar_tree_type.
13447 (gen_hsa_addr): Fixup type when creating addresable temporary.
13448 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
13449 (gen_hsa_unary_operation): Extend operands and convert to dest type if
13450 necessary. Call hsa_fixup_mov_insn_type.
13451 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
13452 extend operands and convert to dest type if necessary.
13453 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
13454 to dest type if necessary.
13455 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
13456 if conversion nt necessary and size matches.
13457 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
13458 to dest type if necessary.
13459 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
13460 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
13462 (gen_hsa_clrsb): Likewise.
13463 (gen_hsa_ffs): Likewise.
13464 (gen_hsa_divmod): Extend operands and convert to dest type if
13466 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
13468 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13470 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
13471 Remove empty default arguments. Use a brace block as output
13473 (conditional return): Ditto.
13475 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
13476 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
13478 (group_ending_nop): Ditto.
13479 (doloop_end): Ditto.
13480 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
13481 (splitters for those): Ditto.
13483 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13485 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
13486 a conditional jump (and the compare for it) so that pc_rtx is the
13488 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
13489 for the deleted and renamed ctr<mode>_internal[234] patterns.
13490 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
13491 Delete second conditional return pattern.
13492 (ctr<mode>_internal2): Delete this second bdnz pattern.
13493 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
13494 (ctr<mode>_internal4): Delete this second bdz pattern.
13496 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13498 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
13499 (always_initialized_rtx_for_ssa_name_p): New predicate.
13500 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
13501 (finish_out_of_ssa): Free new field of SA.
13502 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
13503 * tree-ssa-coalesce.c: Include tree-ssa.h.
13504 (get_parm_default_def_partitions): Remove extern keyword.
13505 (get_undefined_value_partitions): New function.
13506 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
13507 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
13508 uninitialized bits.
13509 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
13511 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13513 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
13515 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13517 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
13518 for targets that preffer 128bit.
13520 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13522 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
13524 2017-10-08 Olivier Hainque <hainque@adacore.com>
13526 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
13527 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
13529 (thumb_set_return_address): Likewise.
13531 2017-10-08 Olivier Hainque <hainque@adacore.com>
13533 * common/config/arm/arm-common.c (arm_except_unwind_info):
13534 Handle DWARF2_UNWIND_INFO.
13536 2017-10-07 Michael Collison <michael.collison@arm.com>
13538 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
13541 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
13543 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
13544 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
13545 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
13546 defined, force the creation of a new block for a dispatch label.
13548 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
13550 * invoke.texi (Wsuggest-attribute=cold): Document.
13551 * common.opt (Wsuggest-attribute=cold): New
13552 * ipa-pure-const.c (warn_function_cold): New function.
13553 * predict.c (compute_function_frequency): Use it.
13554 * predict.h (warn_function_cold): Declare.
13556 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
13558 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
13561 2017-10-06 Martin Liska <mliska@suse.cz>
13563 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
13564 keyword for member functions.
13565 (struct sanopt_tree_couple): New struct.
13566 (struct sanopt_tree_couple_hash): New function.
13567 (struct sanopt_ctx): Add new hash_map.
13568 (has_dominating_ubsan_ptr_check): New function.
13569 (record_ubsan_ptr_check_stmt): Likewise.
13570 (maybe_optimize_ubsan_ptr_ifn): Likewise.
13571 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
13572 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
13574 2017-10-06 Sudakshina Das <sudi.das@arm.com>
13577 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
13578 aarch64_simd_valid_immediate on CONST_VECTORs.
13579 (aarch64_reg_or_bic_imm): Likewise.
13581 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
13583 PR rtl-optimization/82396
13584 * haifa-sched.c (ready_sort_real): Disable qsort checking.
13586 2017-10-06 Sebastian Pop <sebpop@gmail.com>
13588 * graphite-dependences.c (scop_get_reads): Move code to...
13589 (scop_get_must_writes): Move code to...
13590 (scop_get_may_writes): Move code to...
13591 (scop_get_reads_and_writes): ... here.
13592 (scop_get_dependences): Call scop_get_reads_and_writes.
13594 2017-10-06 Jakub Jelinek <jakub@redhat.com>
13596 PR tree-optimization/82434
13597 * fold-const.h (can_native_encode_type_p,
13598 can_native_encode_string_p): Remove.
13599 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
13600 don't encode anything, just return what would be otherwise returned.
13601 (native_encode_fixed, native_encode_complex, native_encode_vector):
13603 (native_encode_string): Likewise. Inline by hand
13604 can_native_encode_string_p.
13605 (can_native_encode_type_p): Remove.
13606 (can_native_encode_string_p): Remove.
13607 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
13608 STRING_CSTs using can_native_encode_string_p, test all
13609 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
13610 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
13611 argument from native_encode_expr.
13612 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
13613 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
13616 2017-10-06 Richard Biener <rguenther@suse.de>
13618 PR tree-optimization/82397
13619 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
13620 operand_equal_p but rely on data_ref_compare_tree for detecting
13622 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
13623 to match up with dr_group_sort_cmp.
13625 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13628 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
13630 * config/s390/s390-builtin-types.def: Regenerate.
13632 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13635 * config/s390/s390-builtin-types.def: Regenerate.
13636 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
13637 Change flag from B_VXE to B_VX.
13638 (s390_vec_min_dbl): Remove B_VXE flag.
13640 2017-10-06 Richard Biener <rguenther@suse.de>
13642 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
13643 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
13644 translate_isl_ast_to_gimple::is_valid_rename,
13645 translate_isl_ast_to_gimple::get_rename,
13646 translate_isl_ast_to_gimple::get_def_bb_for_const,
13647 translate_isl_ast_to_gimple::get_new_name,
13648 translate_isl_ast_to_gimple::collect_all_ssa_names,
13649 translate_isl_ast_to_gimple::copy_loop_phi_args,
13650 translate_isl_ast_to_gimple::collect_all_ssa_names,
13651 translate_isl_ast_to_gimple::copy_loop_phi_args,
13652 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
13653 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
13654 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
13655 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
13656 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
13657 translate_isl_ast_to_gimple::copy_cond_phi_args,
13658 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
13659 translate_isl_ast_to_gimple::edge_for_new_close_phis,
13660 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
13661 translate_isl_ast_to_gimple::rename_uses,
13662 translate_isl_ast_to_gimple::rename_all_uses): Remove.
13663 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
13664 (set_rename_for_each_def): Likewise.
13665 (graphite_copy_stmts_from_block): Handle debug stmt resetting
13666 here. Handle rewriting SCEV analyzable uses here.
13667 (copy_bb_and_scalar_dependences): Generate code for PHI
13669 (graphite_regenerate_ast_isl): Adjust.
13670 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
13671 (add_write, add_read): New functions.
13672 (build_cross_bb_scalars_def): Use it and simplify.
13673 (build_cross_bb_scalars_use): Likewise.
13674 (graphite_find_cross_bb_scalar_vars): Inline into...
13675 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
13676 simulating out-of-SSA. Compute liveout and add dependencies.
13677 (build_scops): Force an empty entry block.
13678 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
13680 (sese_build_liveouts): Declare.
13681 (sese_trivially_empty_bb_p): Likewise.
13682 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
13683 compute liveout and debug_liveout.
13684 (sese_bad_liveouts_use): Remove.
13685 (sese_reset_debug_liveouts_bb): Likewise.
13686 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
13687 (sese_build_liveouts): Build liveout and debug_liveout and store
13689 (new_sese_info): Adjust.
13690 (free_sese_info): Likewise.
13691 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
13692 do not build liveout here.
13693 (move_sese_in_condition): Adjust region entry.
13694 (scev_analyzable_p): Match up with chrec_apply requirements.
13695 (sese_trivially_empty_bb_p): New.
13696 * tree-into-ssa.c (get_reaching_def): Properly support generating
13697 default-defs for incremental rewrite of anonymous names.
13699 2017-10-06 Richard Biener <rguenther@suse.de>
13701 * graphite-sese-to-poly.c (extract_affine): For casts increasing
13702 precision do not perform modulo reduction.
13704 2017-10-06 Richard Biener <rguenther@suse.de>
13706 PR tree-optimization/82436
13707 * tree-vect-slp.c (vect_supported_load_permutation_p): More
13708 conservatively choose the vectorization factor when checking
13709 whether we can perform the required load permutation.
13710 (vect_transform_slp_perm_load): Assert when we may not fail.
13712 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
13714 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
13715 message for incompatible -msdata=* and -mcall-* options.
13717 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13719 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
13720 rate for post-reload scheduling.
13722 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13724 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
13726 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13728 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
13729 to improve monte carlo in scimark.
13731 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13733 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
13734 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
13735 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
13736 pentium4_cost, nocona_cost): Set reassociation width to 1.
13737 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
13738 width to 2 for fp operations and 1 otherwise.
13739 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
13741 (atom_cost): Set reassociation width to 2.
13742 (slm_cost, generic_cost): Set fp reassociation width
13743 to 2 and 1 otherwise.
13744 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
13745 (core_cost): Set fp reassociation width to 4 and vector to 2.
13746 (ix86_reassociation_width): Rewrite using cost table; special case
13747 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
13748 and TARGET_AVX128_OPTIMAL.
13749 * config/i386/i386.h (processor_costs): Add
13750 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
13751 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
13752 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
13753 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
13754 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
13755 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
13757 2017-10-05 Nathan Sidwell <nathan@acm.org>
13759 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
13761 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13763 * config/arm/arm.c (arm_test_fpu_data): New.
13764 (arm_run_selftests): Call arm_test_fpu_data.
13766 2017-10-04 Nathan Sidwell <nathan@acm.org>
13768 * toplev.c (toplev::main): Remove excess parens on pretty_printer
13770 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
13772 2017-10-04 Sudakshina Das <sudi.das@arm.com>
13774 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13775 check type for aarch64_simd_valid_immediate.
13776 (aarch64_output_simd_mov_immediate): Update prototype.
13777 (aarch64_simd_valid_immediate): Update prototype.
13778 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13779 support for ORR-immediate.
13780 (and<mode>3): modified pattern to add support for BIC-immediate.
13781 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13782 now checks for valid immediate for BIC and ORR based on new enum
13784 (aarch64_output_simd_mov_immediate): Function now used to output
13785 BIC/ORR imm as well based on new enum argument.
13786 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13788 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13789 (aarch64_reg_or_bic_imm): Likewise.
13791 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13793 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13794 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13796 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13800 PR rtl-optimization/82396
13801 * haifa-sched.c (autopref_multipass_init): Simplify
13803 (autopref_rank_data): Simplify sort order.
13804 * sched-int.h (autopref_multipass_data_): Remove
13805 multi_mem_insn_p, min_offset and max_offset.
13807 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13809 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13811 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13813 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13814 vect_doubleint_cvt.
13816 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13818 * doc/sourcebuild.texi: Document vect_long_mult.
13820 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13822 PR tree-optimization/82413
13823 * fold-const.c (build_range_check): Use widest_int when comparing
13824 the maximum ETYPE value with HIGH.
13826 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13828 PR rtl-optimization/82396
13829 * haifa-sched.c (autopref_multipass_init): Simplify
13831 (autopref_rank_data): Simplify sort order.
13832 * sched-int.h (autopref_multipass_data_): Remove
13833 multi_mem_insn_p, min_offset and max_offset.
13835 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13837 PR tree-optimization/82381
13838 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13839 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13842 PR tree-optimization/82374
13843 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13844 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13845 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13846 current_function_decl to the new decl.
13848 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13850 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13851 helper macro for IEEE float128 hardware built-in functions.
13852 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13854 (TRUNCF128_ODD): Likewise.
13855 (ADDF128_ODD): Likewise.
13856 (SUBF128_ODD): Likewise.
13857 (MULF128_ODD): Likewise.
13858 (DIVF128_ODD): Likewise.
13859 (FMAF128_ODD): Likewise.
13860 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13861 UNSPEC_TRUNC_ROUND_TO_ODD.
13862 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13863 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13864 floating point round to odd instructions.
13865 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13866 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13867 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13868 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13869 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13870 (trunc<mode>sf2_hw): Change the truncate with round to odd
13871 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13872 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13873 to odd hardware instructions.
13874 (sub<mode>3_odd): Likewise.
13875 (mul<mode>3_odd): Likewise.
13876 (div<mode>3_odd): Likewise.
13877 (sqrt<mode>2_odd): Likewise.
13878 (fma<mode>4_odd): Likewise.
13879 (fms<mode>4_odd): Likewise.
13880 (nfma<mode>4_odd): Likewise.
13881 (nfms<mode>4_odd): Likewise.
13882 (trunc<mode>df2_odd): Change the truncate with round to odd
13883 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13885 * doc/extend.texi (PowerPC built-in functions): Update documentation
13886 for existing IEEE float128-bit built-in functions. Add built-in
13887 functions that generate the IEEE 128-bit floating point round to
13890 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13892 PR rtl-optimization/77729
13893 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13894 to (X&(C1&~C2))|C2 transformations.
13896 2017-10-03 Martin Jambor <mjambor@suse.cz>
13898 PR tree-optimization/82363
13899 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13900 mismatch, mark lacc written regardless of racc.
13902 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13904 PR tree-optimization/82381
13905 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13906 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13907 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13908 fallthrough into reassoc_stmt_dominates_stmt_p.
13911 * combine.c (combine_instructions): Don't combine in unreachable
13914 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13917 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13918 function to not use the have_cpu variable. Do not set cpu_index,
13919 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13920 or the default cpu.
13921 (rs6000_valid_attribute_p): Remove duplicate initializations of
13922 old_optimize and func_optimize.
13923 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13924 (rs6000_activate_target_options): Make global.
13925 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13928 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13930 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13931 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13932 Set *poffset to 0 and *psize and *pmax_size to -1 if
13933 *poffset + *psize overflows in HOST_WIDE_INT.
13935 PR tree-optimization/82387
13936 PR tree-optimization/82388
13937 PR tree-optimization/82389
13938 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13939 instead of live_bytes non-NULL.
13941 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13944 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13947 2017-10-02 Richard Biener <rguenther@suse.de>
13949 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13950 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13951 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13953 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13955 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13956 requested precision matches the type's.
13957 * calls.c (alloc_max_size): Calculate the new candidate size as
13958 a widest_int and use wi::to_widest when comparing it with the
13959 current candidate size.
13960 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13961 zero rather than integer_zero_node.
13962 * match.pd: Check for a no-op conversion before using wi::add
13963 rather than after. Use tree_to_uhwi when summing small shift
13964 counts into an unsigned int.
13966 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13967 Alan Hayward <alan.hayward@arm.com>
13968 David Sherwood <david.sherwood@arm.com>
13971 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13972 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13973 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13974 POINTER_AND_FP_REGS.
13976 2017-10-02 Richard Biener <rguenther@suse.de>
13978 PR tree-optimization/82355
13979 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13980 a mapping for the enclosing loop but avoid generating one for
13981 the loop tree root.
13982 (copy_bb_and_scalar_dependences): Remove premature codegen
13983 error on PHIs in blocks duplicated into multiple places.
13984 * graphite-scop-detection.c
13985 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13986 in the region use it as loop and nest to analyze the DR in.
13987 (try_generate_gimple_bb): Likewise.
13988 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13989 (add_loop_constraints): For blocks in a loop not in the region
13990 create a dimension with a single iteration.
13991 * sese.h (gbb_loop_at_index): Remove assert.
13993 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13995 * omp-expand.c (adjust_context_scope): New function.
13996 (expand_parallel_call): Call adjust_context_scope.
13998 2017-10-01 Jeff Law <law@redhat.com>
14000 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
14001 dom_opt_dom_walker class with direct access to private members.
14002 Add comments. Call test_for_singularity.
14003 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
14004 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
14005 m_dummy_cond anymore.
14006 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
14008 (pass_dominator:execute): Build the dummy_cond here and pass it
14009 to the dom_opt_dom_walker ctor.
14010 (test_for_singularity): New function.
14012 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
14013 Maya Rashish <coypu@sdf.org>
14015 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
14016 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
14017 (alpha*-*-netbsd*) Use nbsd_tm_file.
14018 (arm*-*-netbsdelf*) Likewise.
14019 (i[34567]86-*-netbsdelf*) Likewise.
14020 (x86_64-*-netbsd*) Likewise.
14021 (mips*-*-netbsd*) Likewise.
14022 (powerpc-*-netbsd*) Likewise.
14023 (sh*-*-netbsd*) Likewise.
14024 (sparc-*-netbsdelf*) Likewise.
14025 (sparc64-*-netbsd*) Likewise.
14026 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
14028 (vax-*-netbsdelf*) Likewise.
14029 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
14030 (UINT_FAST8_TYPE) Likewise.
14031 (INT_FAST16_TYPE) Check CHAR_FAST16.
14032 (UINT_FAST16_TYPE) Likewise.
14034 2017-09-30 Jakub Jelinek <jakub@redhat.com>
14037 * config/i386/i386.md
14038 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
14039 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
14040 *divmodsi4_zext_2): New define_insn_and_split.
14041 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
14042 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
14043 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
14044 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
14045 New define_insn_and_split.
14046 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
14047 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
14048 operands[1] having DImode when mode is SImode.
14050 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
14051 always SImode for DIV and MOD in REG_EQUAL notes.
14053 2017-09-29 Yury Gribov <tetra2005@gmail.com>
14055 PR middle-end/82319
14056 * match.pd: Fix handling of NaNs in pattern.
14058 2017-09-29 Jeff Law <law@redhat.com>
14060 * sbitmap.c (bitmap_bit_in_range_p): New function.
14061 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
14062 * tree-ssa-dse.c (live_bytes_read): New function.
14063 (dse_classify_store): Ignore reads of dead bytes.
14065 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
14066 typos and whitespace errors.
14067 * config/i386/predicates.md (address_no_seg_operand): Likewise.
14068 * config/s390/s390.c (s390_emit_prologue): Likewise.
14070 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
14073 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
14074 with a symbol for LRA.
14076 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
14078 PR rtl-optimization/82338
14079 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
14081 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
14083 * genmodes.c (calc_wider_mode): Suppress qsort macro.
14084 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
14085 (qsort_chk): Declare.
14086 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
14087 (qsort_chk): New function.
14089 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14091 PR tree-optimization/82337
14092 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
14093 phi definition if the PHI result appears in an abnormal PHI.
14094 (find_basis_for_base_expr): Don't record a basis if the LHS of the
14095 basis appears in an abnormal PHI.
14097 2017-09-29 Richard Biener <rguenther@suse.de>
14099 * graphite-isl-ast-to-gimple.c
14100 (translate_isl_ast_to_gimple::set_codegen_error): New function.
14101 (binary_op_to_tree): Use it.
14102 (get_rename_from_scev): Likewise.
14103 (copy_loop_phi_nodes): Likewise.
14104 (copy_bb_and_scalar_dependences): Likewise.
14105 (translate_pending_phi_nodes): Likewise.
14107 2017-09-29 Jakub Jelinek <jakub@redhat.com>
14110 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
14111 for movabsq $(i32 << shift), r64.
14113 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
14115 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
14116 index when encoding %esp as %rsp to avoid 0x67 prefix.
14118 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
14120 * config/i386/i386.md (*movsf_internal, *movdf_internal):
14121 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
14123 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
14125 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
14126 Extensions with more than 16 double VFP registers.
14127 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
14128 to_clear_mask and all code related to it. Replace the remaining
14129 entry by a sbitmap and adapt code accordingly.
14131 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
14133 * brig-builtins.def: Change pure attributes to const.
14135 2017-09-28 Joseph Myers <joseph@codesourcery.com>
14137 * config.gcc (default_gnu_indirect_function): Default to yes for
14138 sparc*-*-linux* with glibc.
14140 2017-09-28 Joseph Myers <joseph@codesourcery.com>
14142 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
14143 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
14144 when creating .init_array and .fini_array sections with priority
14147 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
14150 * config/aarch64/aarch64.c
14151 (aarch64_builtin_support_vector_misalignment): Always return false
14152 when misalignment is unknown.
14154 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
14156 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
14157 this function to return false if the definition used by the swap
14158 instruction is artificial, or if the memory address from which the
14159 constant value is loaded is not represented by a base address held
14160 in a register or if the base address register is a frame or stack
14161 pointer. Additionally, return false if the base address of the
14162 loaded constant is a SYMBOL_REF but is not considered to be a
14164 (replace_swapped_load_constant): New function.
14165 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
14166 loaded constant vector with a load of a swapped constant vector.
14168 2017-09-27 Carl Love <cel@us.ibm.com>
14170 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
14171 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
14172 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
14175 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
14177 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
14178 first, always call autopref_rank_data otherwise.
14180 2017-09-27 Richard Biener <rguenther@suse.de>
14182 * graphite-scop-detection.c (find_scop_parameters): Move
14183 loop bound handling ...
14184 (gather_bbs::before_dom_children): ... here, avoiding the need
14185 to build scop_info->loop_nest.
14186 (record_loop_in_sese): Remove.
14187 * sese.h (sese_info_t::loop_nest): Remove.
14188 * sese.c (new_sese_info): Do not allocate loop_nest.
14189 (free_sese_info): Do not free loop_nest.
14191 2017-09-27 Jakub Jelinek <jakub@redhat.com>
14194 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
14195 lhs from calls if the lhs has addressable type.
14197 2017-09-27 Richard Biener <rguenther@suse.de>
14199 * graphite.h (scop::max_alias_set): New member.
14200 * graphite-scop-detection.c: Remove references to non-existing
14201 --param in comments.
14202 (build_alias_sets): Record the maximum alias set used for drs.
14203 (build_scops): Support zero as unlimited for
14204 --param graphite-max-arrays-per-scop.
14205 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
14206 and inline into ...
14207 (build_poly_sr_1): ... here. Compute alias set based on the
14208 maximum alias set used for drs rather than
14209 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
14211 2017-09-27 Richard Biener <rguenther@suse.de>
14213 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
14214 --param loop-block-tile-size=0 to disable tiling.
14216 2017-09-27 Richard Biener <rguenther@suse.de>
14218 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
14219 (graphite-max-nb-scop-params): Document special value zero.
14220 * domwalk.h (dom_walker::STOP): New symbolical constant.
14221 (dom_walker::dom_walker): Add optional parameter for bb to
14223 (dom_walker::~dom_walker): Declare.
14224 (dom_walker::before_dom_children): Document STOP return value.
14225 (dom_walker::m_user_bb_to_rpo): New member.
14226 (dom_walker::m_bb_to_rpo): Likewise.
14227 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
14228 mapping here if not provided by the user.
14229 (dom_walker::~dom_walker): Free bb to RPO mapping if not
14230 provided by the user.
14231 (dom_walker::STOP): Define.
14232 (dom_walker::walk): Do not compute bb to RPO mapping here.
14233 Support STOP return value from before_dom_children to stop
14235 * graphite-optimize-isl.c (optimize_isl): If the schedule
14236 is the same still generate code if -fgraphite-identity
14237 or -floop-parallelize-all are given.
14238 * graphite-scop-detection.c: Include cfganal.h.
14239 (gather_bbs::gather_bbs): Get and pass through bb to RPO
14241 (gather_bbs::before_dom_children): Return STOP for BBs
14243 (build_scops): Compute bb to RPO mapping and pass it to
14244 the domwalk. Treat --param graphite-max-nb-scop-params=0
14245 as not limiting the number of params.
14246 * graphite.c (graphite_initialize): Remove limit on the
14247 number of basic-blocks in a function.
14248 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
14249 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
14250 default value of 10.
14252 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14254 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
14255 Adjust code to eliminate needing to do the shift right 32-bits
14256 operation after XSCVDPSPN.
14258 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14260 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
14261 ((X / Y) != 0 -> X >= Y): Likewise.
14263 2017-09-26 Carl Love <cel@us.ibm.com>
14265 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
14266 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
14267 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
14268 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
14269 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
14270 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
14271 definitions and overloading.
14272 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
14273 statement for P9V_BUILTIN_XST_LEN_R.
14274 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
14275 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
14276 define_expand and define_insn for the instructions and builtins.
14277 * doc/extend.texi: Update the built-in documentation file for the new
14278 built-in functions.
14279 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
14280 define_insn for the instructions
14282 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
14285 * gcc/config/netbsd-protos.h: New file.
14286 * gcc/config/netbsd.c: New file.
14287 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
14288 * gcc/config/t-netbsd: New file.
14289 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
14290 (tmake_file) Add t-netbsd.
14291 (extra_objs) Add netbsd.o.
14293 2017-09-26 Janus Weil <janus@gcc.gnu.org>
14297 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
14299 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14301 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
14302 sign extension from a vector register to a GPR by doing a 32-bit
14303 direct move and then an EXTSW.
14304 (extendsi<mode>2 splitter): Likewise.
14305 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
14306 right or vector extract after doing XSCVDPSPN. Use
14307 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
14309 (movdi_from_sf_zero_ext): Likewise.
14310 (reload_gpr_from_vsxsf): Likewise.
14311 (p8_mfvsrd_4_disf): Delete, no longer used.
14312 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
14313 then needing to move the SFmode to a GPR to use the XSCVDPSP
14314 instruction instead of FRSP and XSCVDPSPN.
14315 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
14316 it is adjacent to the other XSCVSPDP insns.
14317 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
14318 SFmode to be in traditional Altivec registers.
14319 (vsx_xscvdpspn): Eliminate useless alternative constraint.
14320 (vsx_xscvspdpn): Likewise.
14321 (vsx_xscvspdpn_scalar): Likewise.
14323 2017-09-26 Martin Jambor <mjambor@suse.cz>
14325 * tree-sra.c (compare_access_positions): Put integral types first,
14326 stabilize sorting of integral types, remove conditions putting
14327 non-full-precision integers last.
14328 (sort_and_splice_var_accesses): Disable scalarization if a
14329 non-integert would be represented by a non-full-precision integer.
14331 2017-09-26 Joseph Myers <joseph@codesourcery.com>
14333 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
14334 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
14335 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
14336 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
14337 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
14338 conditionals inside the function instead of around it. Call
14339 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
14340 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
14342 2017-09-26 Richard Biener <rguenther@suse.de>
14344 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
14346 (scop_detection::build_scop_breadth): ... this. Removed.
14347 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
14348 (scop_detection::harmful_stmt_in_bb): Likewise.
14349 (scop_detection::graphite_can_represent_stmt): Likewise.
14350 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
14351 (scop_detection::can_represent_loop): Remove recursion, fold in ...
14352 (scop_detection::can_represent_loop_1): ... this. Removed.
14353 (scop_detection::harmful_loop_in_region): Simplify after inlining
14354 the above and remove more quadraticness.
14355 (build_scops): Adjust.
14356 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
14359 2017-09-26 Jakub Jelinek <jakub@redhat.com>
14362 * config/i386/i386.c (ix86_print_operand_address_as): Only test
14363 REGNO (base) == SP_REG if base is a REG.
14365 PR middle-end/35691
14366 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
14367 if it is different SSA_NAME.
14368 (optimize_range_tests_cmp_bitwise): New function.
14369 (optimize_range_tests): Call it.
14371 2017-09-26 Richard Biener <rguenther@suse.de>
14373 PR tree-optimization/82321
14374 * graphite.c (canonicalize_loop_closed_ssa): Properly check
14375 for the def being inside the loop.
14377 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14379 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
14381 * config/s390/s390-builtins.def: Fix constraint on op4.
14383 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14385 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
14386 independent expanders.
14387 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
14388 ("vec_ordered", "vec_unordered"): New expanders.
14390 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14392 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
14395 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14397 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
14398 vec_unpacks_lo_v16qi.
14399 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
14401 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14403 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
14404 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
14405 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
14407 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14409 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
14411 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
14413 ("*vec_slb<mode>"): New insn pattern.
14414 ("vec_shr_<mode>"): New expander.
14415 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
14416 and force the shift count operand to V16QImode.
14417 ("vec_srb<mode>"): Set shift count mode to V16QI.
14419 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14421 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
14422 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
14423 ("vec_widen_smult_hi_<mode>"): New expander definitions.
14425 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
14428 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
14430 2017-09-26 Richard Biener <rguenther@suse.de>
14432 PR tree-optimization/82320
14433 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
14436 2017-09-25 Jeff Law <law@redhat.com>
14438 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
14439 prototype for new argument.
14440 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
14441 mostly extracted from rs6000_emit_allocate_stack.
14442 (rs6000_emit_probe_stack_range_stack_clash): New function.
14443 (rs6000_emit_allocate_stack): Call
14444 rs6000_emit_probe_stack_range_stack_clash as needed.
14445 (rs6000_emit_probe_stack_range): Add additional argument
14446 to call to gen_probe_stack_range{si,di}.
14447 (output_probe_stack_range): New.
14448 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
14449 (output_probe_stack_range_stack_clash): New.
14450 (rs6000_emit_prologue): Emit notes into dump file as requested.
14451 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
14452 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
14453 Add additional operand and pass it to output_probe_stack_range.
14455 2017-09-25 Bin Cheng <bin.cheng@arm.com>
14457 PR tree-optimization/82163
14458 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
14459 (checking_verify_loop_closed_ssa): New parameter.
14460 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
14461 (check_loop_closed_ssa_stmt): Delete.
14462 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
14463 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
14464 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
14467 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
14469 * brig-builtins.def: Treat HSAIL barrier builtins as
14470 setjmp/longjump style functions.
14472 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14474 * target.def (constant_alignment): New hook.
14475 * defaults.h (CONSTANT_ALIGNMENT): Delete.
14476 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
14477 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
14478 * doc/tm.texi: Regenerate.
14479 * targhooks.h (default_constant_alignment): Declare.
14480 (constant_alignment_word_strings): Likewise.
14481 * targhooks.c (default_constant_alignment): New function.
14482 (constant_alignment_word_strings): Likewise.
14483 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
14484 instead of CONSTANT_ALIGNMENT.
14485 * varasm.c (align_variable, get_variable_align, build_constant_desc)
14486 (force_const_mem): Likewise.
14487 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
14488 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
14489 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
14490 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14491 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
14493 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
14494 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14495 constant_alignment_word_strings.
14496 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
14497 (CONSTANT_ALIGNMENT): Likewise.
14498 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14499 (arm_constant_alignment): New function.
14500 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
14501 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14502 constant_alignment_word_strings.
14503 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
14504 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14505 constant_alignment_word_strings.
14506 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
14507 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14508 (cris_constant_alignment): New function.
14509 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
14510 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14511 (epiphany_constant_alignment): New function.
14512 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
14513 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14514 constant_alignment_word_strings.
14515 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
14516 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14517 constant_alignment_word_strings.
14518 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
14519 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14520 constant_alignment_word_strings.
14521 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
14522 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
14523 * config/i386/i386.c (ix86_constant_alignment): Make static.
14524 Use the same interface as the target hook.
14525 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14526 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
14527 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14528 constant_alignment_word_strings.
14529 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
14530 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
14531 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14532 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
14533 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14534 constant_alignment_word_strings.
14535 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
14536 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14537 constant_alignment_word_strings.
14538 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
14539 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14540 constant_alignment_word_strings.
14541 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
14542 * config/microblaze/microblaze.c (microblaze_constant_alignment):
14544 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14545 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
14546 * config/mips/mips.c (mips_constant_alignment): New function.
14547 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14548 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
14549 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
14550 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14551 (mmix_constant_alignment): Make static. Use the same interface
14552 as the target hook.
14553 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
14554 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14555 constant_alignment_word_strings.
14556 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
14557 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14558 constant_alignment_word_strings.
14559 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
14560 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14561 constant_alignment_word_strings.
14562 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
14563 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14564 (rs6000_constant_alignment): New function.
14565 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
14566 * config/riscv/riscv.c (riscv_constant_alignment): New function.
14567 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14568 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
14569 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14570 (rs6000_constant_alignment): New function.
14571 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
14572 * config/s390/s390.c (s390_constant_alignment): New function.
14573 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14574 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
14575 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14576 constant_alignment_word_strings.
14577 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
14578 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14579 (sparc_constant_alignment): New function.
14580 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
14581 * config/spu/spu.c (spu_constant_alignment): New function.
14582 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14583 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
14584 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14585 constant_alignment_word_strings.
14586 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
14587 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14588 constant_alignment_word_strings.
14589 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
14590 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14591 constant_alignment_word_strings.
14592 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
14593 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14594 (visium_constant_alignment): New function.
14595 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
14596 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14597 (xtensa_constant_alignment): New function.
14598 * system.h (CONSTANT_ALIGNMENT): Poison.
14600 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
14602 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14603 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
14604 (rs6000_builtin_valid_without_lhs): New helper function.
14605 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14606 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
14608 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14610 * target.h (vec_perm_indices): Use unsigned short rather than
14612 (auto_vec_perm_indices): Likewise.
14613 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
14614 Use unsigned int rather than unsigned char.
14615 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
14617 2017-09-25 Richard Biener <rguenther@suse.de>
14619 * cfgloop.h (sort_sibling_loops): Declare.
14620 * cfgloop.c (sort_sibling_loops_cmp): New helper.
14621 (sort_sibling_loops): New function sorting the sibling loop list
14623 * graphite.c (graphite_transform_loops): Sort sibling loops.
14625 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
14627 * target.def (vec_perm_const_ok): Change sel parameter to
14629 * optabs-query.c (can_vec_perm_p): Update accordingly.
14630 * doc/tm.texi: Regenerate.
14631 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
14632 auto_vec_perm_indices and remove separate nelt field.
14633 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
14634 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
14635 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
14636 (aarch64_expand_vec_perm_const): Update accordingly.
14637 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
14638 to vec_perm_indices.
14639 * config/arm/arm.c (expand_vec_perm_d): Change perm to
14640 auto_vec_perm_indices and remove separate nelt field.
14641 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
14642 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
14643 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
14645 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
14646 to vec_perm_indices.
14647 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
14648 sel to vec_perm_indices.
14649 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
14650 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
14651 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
14653 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
14656 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
14659 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
14660 on the FUNCTION_DECL function context if it has a DIE that is a
14663 2017-09-25 Richard Biener <rguenther@suse.de>
14665 PR tree-optimization/82285
14666 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
14669 2017-09-25 Tom de Vries <tom@codesourcery.com>
14673 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
14676 2017-09-25 Richard Biener <rguenther@suse.de>
14678 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
14679 ISL errors other than isl_error_quota happen. Dump if the
14680 schedule is the same.
14681 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
14682 errors instead of aborting inside ISL.
14684 2017-09-25 Iain Sandoe <iain@codesourcery.com>
14687 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
14688 of libgcc_eh for m64.
14689 * config/i386/darwin64.h: Likewise.
14691 2017-09-25 Richard Biener <rguenther@suse.de>
14693 PR middle-end/82144
14694 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
14695 attribute for incomplete types nor twice for complete ones.
14697 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
14700 * config/i386/i386.c (ix86_print_operand_address_as): Encode
14701 %esp as %rsp to avoid 0x67 prefix if there is no index or base
14704 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
14707 * config/i386/i386.opt (mprefer-avx256): Use
14708 ix86_target_flags variable.
14709 * config/i386/i386.c (ix86_target_string): Move
14710 -mprefer-avx256 to flag2_opts.
14712 2017-09-22 Jakub Jelinek <jakub@redhat.com>
14714 PR middle-end/35691
14715 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
14716 and x != -1 | y != -1 into (x & y) != -1.
14718 2017-09-22 Steve Ellcey <sellcey@cavium.com>
14720 * config.gcc: Add new case statement to set
14721 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
14722 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
14723 s390x-*-linux* case statements. Added aarch64 to the list of
14724 supported architectures.
14726 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14728 PR tree-optimization/82289
14729 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
14730 STMT_VINFO_RELEVANT_P.
14732 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14733 Alan Hayward <alan.hayward@arm.com>
14734 David Sherwood <david.sherwood@arm.com>
14736 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
14737 for VR_RANGE only; don't allow VR_ANTI_RANGE.
14738 (extract_range_from_binary_expr_1): Don't call
14739 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
14741 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14742 Alan Hayward <alan.hayward@arm.com>
14743 David Sherwood <david.sherwood@arm.com>
14745 * target.def (preferred_vector_alignment): New hook.
14746 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
14748 * doc/tm.texi: Regenerate.
14749 * targhooks.h (default_preferred_vector_alignment): Declare.
14750 * targhooks.c (default_preferred_vector_alignment): New function.
14751 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
14753 (DR_TARGET_ALIGNMENT): New macro.
14754 (aligned_access_p): Update commentary.
14755 (vect_known_alignment_in_bytes): New function.
14756 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
14758 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
14759 Calculate the misalignment based on the target alignment rather than
14761 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
14762 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
14763 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
14764 the target alignment, rather than masking the element misalignment
14765 with the number of elements in a vector. Also use the target
14766 alignment when calculating the maximum number of peels.
14767 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
14768 instead of TYPE_ALIGN_UNIT.
14769 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
14770 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
14771 (vect_create_addr_base_for_vector_ref): Update call accordingly.
14772 (vect_create_data_ref_ptr): Likewise.
14773 (vect_setup_realignment): Realign by ANDing with
14774 -DR_TARGET_MISALIGNMENT.
14775 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14776 the number of peels based on DR_TARGET_ALIGNMENT.
14777 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14778 with the guaranteed alignment boundary when deciding whether
14780 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14781 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14782 (ensure_base_align): Remove stmt_info parameter. Get the
14783 target base alignment from DR_TARGET_ALIGNMENT.
14784 (vectorizable_store): Update call accordingly. Interpret
14785 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14787 (vectorizable_load): Likewise.
14789 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14790 Alan Hayward <alan.hayward@arm.com>
14791 David Sherwood <david.sherwood@arm.com>
14793 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14794 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14795 (vect_enhance_data_refs_alignment): Likewise.
14797 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14799 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14800 error. Only quit immediately if parsing is complete.
14801 (BEGIN): Initialize fatal_err and parse_done.
14802 (begin fpu, end fpu): Check number of arguments.
14803 (begin arch, end arch): Likewise.
14804 (begin cpu, end cpu): Likewise.
14805 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14806 (optalias): Likewise.
14808 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14810 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14811 * config/arm/arm-isa.h: Delete. Move definitions to ...
14812 * arm-cpus.in: ... here. Use new feature and fgroup values.
14813 * config/arm/arm.c (arm_option_override): Use lower case for feature
14815 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14816 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14817 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14819 (print_isa_bits_for): New function.
14820 (gen_isa): New function.
14821 (gen_comm_data): Use print_isa_bits_for.
14822 (define feature): New keyword.
14823 (define fgroup): New keyword.
14824 * config/arm/t-arm (TM_H): Remove.
14825 (GTM_H): Add arm-isa.h.
14826 (arm-isa.h): Add rule to generate file.
14827 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14828 case for feature bit names.
14830 2017-09-22 Richard Biener <rguenther@suse.de>
14832 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14834 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14835 print of no dependency loops ...
14836 * graphite.c (graphite_transform_loops): ... here.
14837 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14839 (same_close_phi_node, remove_duplicate_close_phi,
14840 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14841 (canonicalize_loop_closed_ssa): ... here and simplify.
14842 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14843 (optimize_isl): Use dump_printf_loc to tell when we stopped
14844 optimizing because of an ISL timeout.
14846 2017-09-22 Richard Biener <rguenther@suse.de>
14848 PR tree-optimization/82291
14849 * tree-if-conv.c (predicate_mem_writes): Make sure to
14850 remove writes in blocks predicated with false.
14852 2017-09-22 Richard Biener <rguenther@suse.de>
14854 * sese.c: Include cfganal.h.
14855 (if_region_set_false_region): Remove.
14856 (create_if_region_on_edge): Likewise.
14857 (move_sese_in_condition): Re-implement without destroying
14860 2017-09-22 Richard Biener <rguenther@suse.de>
14862 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14863 Verify both BBs contain loop PHI nodes before dispatching to
14864 copy_loop_phi_args.
14865 (graphite_regenerate_ast_isl): Do not recompute dominators,
14866 do not verify three times. Restructure for clarity.
14867 * graphite-scop-detection.c (same_close_phi_node,
14868 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14869 defined_in_loop_p, canonicalize_loop_closed_ssa,
14870 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14871 checking and SSA rewrite, move to ...
14872 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14873 (graphite_initialize): Do not pass in ctx, do not reset the
14874 SCEV cache, compute only dominators.
14875 (graphite_transform_loops): Allocate ISL ctx after
14876 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14877 Maintain post-dominators only around build_scops.
14878 * sese.c (if_region_set_false_region): Make static. Free
14879 and recompute dominators.
14880 (move_sese_in_condition): Assert we don't get called with
14881 post-dominators computed.
14882 * sese.h (if_region_set_false_region): Remove.
14884 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14886 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14887 mode attribute for TARGET_AVX512VL.
14889 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14891 * config/i386/i386.opt (mprefer-avx256): New option.
14892 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14894 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14895 for TARGET_PREFER_AVX256.
14896 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14898 2017-09-21 Jeff Law <law@redhat.com>
14900 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14901 Fix dump output if the only stack space is for pushed registers.
14903 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14905 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14908 2017-09-21 Martin Sebor <msebor@redhat.com>
14911 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14912 code (in C++) or code that triggers warnings.
14914 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14916 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14918 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14920 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14921 * sched-rgn.c: Ditto.
14922 * sel-sched-ir.c: Ditto.
14924 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14926 * toplev.h (set_random_seed): Adjust return type.
14927 * toplev.c (init_local_tick): Move eager initialization of random_seed
14928 to get_random_seed. Adjust comment.
14929 (init_random_seed): Inline to get_random_seed, delete.
14930 (get_random_seed): Initialize random_seed lazily.
14931 (set_random_seed): Do not return previous value.
14932 (print_switch_value): Do not call get_random_seed.
14934 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14936 * cgraph.c (delete_function_version): New, broken out from...
14937 (cgraph_node::delete_function_version): ...here. Rename to
14938 cgraph_node::delete_function_version_by_decl. Update all uses.
14939 (cgraph_node::remove): Call delete_function_version.
14941 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14944 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14945 VAR_DECLs to which addressable non-volatile parameters are mapped
14946 and for id->retvar after the return value assignment. Clear
14947 id->retval and id->retbnd after inlining.
14949 2017-09-21 Richard Biener <rguenther@suse.de>
14951 PR tree-optimization/82276
14952 PR tree-optimization/82244
14953 * tree-vrp.c (build_assert_expr_for): Set
14954 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14956 (remove_range_assertions): Revert earlier change.
14958 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14961 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14963 2017-09-21 Richard Biener <rguenther@suse.de>
14965 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14966 Restore valid IL after code generation errors.
14967 * graphite.c (graphite_transform_loops): Diagnose code
14968 generation issues as MSG_MISSED_OPTIMIZATION and continue
14969 with processing SCOPs.
14971 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14972 Alan Hayward <alan.hayward@arm.com>
14973 David Sherwood <david.sherwood@arm.com>
14975 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14976 rather than choosing between plus_constant and gen_rtx_<CODE>.
14977 * expr.c (emit_push_insn): Likewise.
14978 (expand_expr_real_2): Likewise.
14980 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14981 Alan Hayward <alan.hayward@arm.com>
14982 David Sherwood <david.sherwood@arm.com>
14984 * loop-unroll.c (split_iv): Call copy_rtx on the step.
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 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14991 calling tree_to_uhwi.
14993 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14994 Alan Hayward <alan.hayward@arm.com>
14995 David Sherwood <david.sherwood@arm.com>
14997 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14998 INTEGER_CST rather than a negative test for ADDR_EXPR.
15000 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
15001 Alan Hayward <alan.hayward@arm.com>
15002 David Sherwood <david.sherwood@arm.com>
15004 * tree-vrp.c (extract_range_from_binary_expr_1): Check
15005 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
15007 2017-09-21 Richard Biener <rguenther@suse.de>
15009 PR tree-optimization/71351
15010 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
15011 graphite_create_new_loop_guard): Remove, fold remaining parts
15013 (translate_isl_ast_node_for): ... here and simplify.
15015 2017-09-21 Jakub Jelinek <jakub@redhat.com>
15018 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
15019 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
15020 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
15021 alternative always use QI mode, for -Os imov (=R,R) alternative
15022 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
15025 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15026 Jeff Law <law@redhat.com>
15028 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
15029 (allocate_stack_space): New function, partially extracted from
15030 s390_emit_prologue.
15031 (s390_emit_prologue): Track offset to most recent stack probe.
15032 Code to allocate space moved into allocate_stack_space.
15033 Dump actions when no stack is allocated.
15034 (s390_prologue_plus_offset): New function.
15035 (s390_emit_stack_probe): Likewise.
15037 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
15039 * common.opt (Wa, Wl, Wp, g, gz=): Add
15041 (gno-column-info): Remove.
15042 (gcolumn-info): Drop RejectNegative.
15043 (gno-): New prefix.
15044 (gno-record-gcc-switches): Remove.
15045 (grecord-gcc-switches): Drop RejectNegative.
15046 (gno-split-dwarf): Remove.
15047 (gsplit-dwarf): Drop RejectNegative.
15048 (gno-strict-dwarf): Remove.
15049 (gstrict-dwarf): Drop RejectNegative.
15050 * config/darwin.opt (gfull, gused): Add RejectNegative.
15051 * dwarf2out.c (gen_producer_string): Drop
15052 gno-record-gcc-switches handler.
15053 * optc-gen.awk: Add g to prefixes with negative forms.
15054 * opts-common.c (remapping_prefix_p): New.
15055 (find_opt): Check it.
15056 (generate_canonical_option): Test g prefix.
15057 (option_map): Add -gno- mapping.
15058 (add_misspelling_candidates): Check remapping_prefix_p.
15060 2017-09-20 Jeff Law <law@redhat.com>
15062 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
15063 thinko in stack clash protection support.
15065 * explow.c (compute_stack_clash_protection_loop_data): Use
15066 CONST_INT_P instead of explicit test. Verify object is a
15067 CONST_INT_P before looking at INTVAL.
15068 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
15069 instead of explicit test.
15071 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
15074 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
15075 address instead of to r1 and r11.
15077 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
15079 * config.gcc: Support "knm".
15080 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
15081 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
15083 * config/i386/i386.c (m_KNM): Define.
15084 (processor_target_table): Add "knm".
15086 (ix86_option_override_internal): Add "knm".
15087 (ix86_issue_rate): Add PROCESSOR_KNM.
15088 (ix86_adjust_cost): Ditto.
15089 (ia32_multipass_dfa_lookahead): Ditto.
15090 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
15091 (fold_builtin_cpu): Add M_INTEL_KNM.
15092 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
15093 (processor_type): Add PROCESSOR_KNM.
15094 * config/i386/x86-tune.def: Add m_KNM.
15095 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
15097 2017-09-20 Richard Biener <rguenther@suse.de>
15099 PR tree-optimization/80213
15100 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
15101 are allowed in empty BBs as well.
15102 (canonicalize_loop_closed_ssa): Also look for other complex
15104 (scop_detection::get_sese): Include the loop-closed PHI block
15106 (scop_detection::merge_sese): Remove code adding extra blocks.
15107 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
15108 (build_scops): Assert the final returned scop is invalid.
15110 2017-09-20 Richard Biener <rguenther@suse.de>
15112 PR tree-optimization/82264
15113 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
15115 (vn_phi_lookup): Likewise.
15116 (vn_phi_insert): Likewise.
15118 2017-09-20 Jakub Jelinek <jakub@redhat.com>
15120 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
15121 that fits into uhwi or shwi, add DW_AT_const_value regardless
15122 of early_dwarf without going through RTL, using add_AT_unsigned
15125 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
15126 (DEBUG_LTO_ABBREV_SECTION): Likewise.
15127 (DEBUG_LTO_MACINFO_SECTION): Likewise.
15128 (DEBUG_MACRO_SECTION): Likewise.
15129 (DEBUG_LTO_MACRO_SECTION): Likewise.
15130 (DEBUG_STR_DWO_SECTION): Likewise.
15131 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
15132 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
15133 (DEBUG_LTO_DWO_LINE_SECTION): Define.
15134 (DEBUG_LTO_LINE_STR_SECTION): Define.
15135 (init_sections_and_labels): Initialize debug_line_str_section
15136 variable. Initialize debug_loc_section for -gdwarf-5 to
15137 DEBUG_LOCLISTS_SECTION. Formatting fixes.
15139 2017-09-20 Richard Biener <rguenther@suse.de>
15141 * graphite-sese-to-poly.c (extract_affine): Properly handle
15142 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
15144 2017-09-20 Richard Biener <rguenther@suse.de>
15146 PR tree-optimization/81373
15147 * graphite-scop-detection.c (build_cross_bb_scalars_def):
15148 Force SESE live-out defs to be handled even if they are
15151 2017-09-19 Jeff Law <law@redhat.com>
15153 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
15154 nothing for stack adjustments with REG_STACK_CHECK.
15155 * sched-deps.c (parse_add_or_inc): Reject insns with
15156 REG_STACK_CHECK from dependency breaking.
15157 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
15158 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
15159 * reg-notes.def (STACK_CHECK): New note.
15161 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
15162 (ix86_expand_prologue): Dump stack clash info as needed.
15163 Call ix86_adjust_stack_and_probe_stack_clash as needed.
15165 * function.c (dump_stack_clash_frame_info): New function.
15166 * function.h (dump_stack_clash_frame_info): Prototype.
15167 (enum stack_clash_probes): New enum.
15169 * config/alpha/alpha.c (alpha_expand_prologue): Also check
15170 flag_stack_clash_protection.
15171 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
15172 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
15173 (arm_frame_pointer_required): Likewise.
15174 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
15175 (ia64_expand_prologue): Likewise.
15176 * config/mips/mips.c (mips_expand_prologue): Likewise.
15177 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
15178 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15179 (sparc_flat_expand_prologue): Likewise.
15180 * config/spu/spu.c (spu_expand_prologue): Likewise.
15182 * explow.c: Include "params.h".
15183 (anti_adjust_stack_and_probe_stack_clash): New function.
15184 (get_stack_check_protect): Likewise.
15185 (compute_stack_clash_protection_loop_data): Likewise.
15186 (emit_stack_clash_protection_loop_start): Likewise.
15187 (emit_stack_clash_protection_loop_end): Likewise.
15188 (allocate_dynamic_stack_space): Use get_stack_check_protect.
15189 Use anti_adjust_stack_and_probe_stack_clash.
15190 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
15191 (emit_stack_clash_protection_loop_start): Likewise.
15192 (emit_stack_clash_protection_loop_end): Likewise.
15193 * rtl.h (get_stack_check_protect): Prototype.
15194 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
15195 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
15196 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
15198 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
15200 * doc/tm.texi: Rebuilt.
15201 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
15202 get_stack_check_protect.
15203 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
15204 * config/arm/arm.c (arm_expand_prologue): Likewise.
15205 (arm_frame_pointer_required): Likewise.
15206 * config/i386/i386.c (ix86_expand_prologue): Likewise.
15207 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
15208 * config/mips/mips.c (mips_expand_prologue): Likewise.
15209 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
15210 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
15211 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15212 (sparc_flat_expand_prologue): Likewise.
15214 * common.opt (-fstack-clash-protection): New option.
15215 * flag-types.h (enum stack_check_type): Note difference between
15216 -fstack-check= and -fstack-clash-protection.
15217 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
15218 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
15219 * toplev.c (process_options): Issue warnings/errors for cases
15220 not handled with -fstack-clash-protection.
15221 * doc/invoke.texi (-fstack-clash-protection): Document new option.
15222 (-fstack-check): Note additional problem with -fstack-check=generic.
15223 Note that -fstack-check is primarily for Ada and refer users
15224 to -fstack-clash-protection for stack-clash-protection.
15225 Document new params for stack clash protection.
15227 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
15229 * config/i386/i386.c (ix86_split_long_move): Do not handle
15230 address used for LEA in a special way.
15232 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
15234 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
15236 2017-09-19 Martin Sebor <msebor@redhat.com>
15239 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
15240 of incompatible types.
15242 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
15244 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15245 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
15246 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15247 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
15249 2017-09-19 Richard Biener <rguenther@suse.de>
15251 PR tree-optimization/82244
15252 * tree-vrp.c (remove_range_assertions): Do not propagate
15253 a constant to abnormals but replace the assert with a copy.
15255 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
15257 PR rtl-optimization/57878
15258 PR rtl-optimization/68988
15259 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
15260 avoidance test involving non_reload_pseudos. Move frequency test
15261 below the general fragmentation avoidance test.
15263 2017-09-19 Richard Biener <rguenther@suse.de>
15265 PR tree-optimization/69728
15266 * graphite-sese-to-poly.c (schedule_error): New global.
15267 (add_loop_schedule): Handle empty domain by failing the
15269 (build_original_schedule): Handle schedule_error.
15271 2017-09-19 Richard Biener <rguenther@suse.de>
15273 * graphite-scop-detection.c (scop_detection::can_represent_loop):
15274 Do not iterate to sibling loops but only to siblings of inner
15277 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
15280 * config/m68k/m68k.md (moveq feeding equality comparison): Check
15281 that the registers are different.
15283 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
15285 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
15286 to processor_model and "amdfam17h" to arch_names_table.
15287 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
15289 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15292 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
15294 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15295 Alan Hayward <alan.hayward@arm.com>
15296 David Sherwood <david.sherwood@arm.com>
15298 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
15300 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
15302 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
15303 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
15304 vect_schedule_slp_instance.
15305 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
15306 Update call to vect_slp_analyze_node_operations. Simplify return
15308 (vect_slp_analyze_bb_1): Update call accordingly.
15309 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
15310 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
15311 (vect_schedule_slp): Update call accordingly.
15313 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15314 Alan Hayward <alan.hayward@arm.com>
15315 David Sherwood <david.sherwood@arm.com>
15317 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
15318 with types that aren't in fact scalar.
15320 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15322 * tree-vect-slp.c (vect_record_max_nunits): New function,
15324 (vect_build_slp_tree_1): ...here.
15325 (vect_build_slp_tree_2): Call it for phis too.
15327 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15329 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
15330 to vect_get_vec_def_for_operand when getting the mask operand.
15332 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15333 Alan Hayward <alan.hayward@arm.com>
15334 David Sherwood <david.sherwood@arm.com>
15336 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
15339 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15340 Alan Hayward <alan.hayward@arm.com>
15341 David Sherwood <david.sherwood@arm.com>
15343 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
15344 calculation for vector booleans.
15346 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15347 Alan Hayward <alan.hayward@arm.com>
15348 David Sherwood <david.sherwood@arm.com>
15350 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
15352 (vect_transform_stmt): ...here.
15353 (vect_analyze_stmt): Use it instead of calling
15354 vectorizable_live_operation directly.
15356 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
15358 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
15359 non-SIMT targets in acc vector loops.
15361 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
15363 * configure.ac: Add arc and check if assembler supports gdwarf2.
15364 * configure: Regenerate.
15366 2017-09-18 Richard Biener <rguenther@suse.de>
15368 PR tree-optimization/82220
15369 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
15370 epilogue niters from the min_profitable_iters compute.
15372 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15375 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
15376 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
15377 (ix86_init_pic_reg): Revert 2017-09-01 changes.
15379 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
15382 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
15383 switching to a new text section.
15385 2017-09-18 Richard Biener <rguenther@suse.de>
15387 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
15389 (build_alias_set): Reject aliases with no access function.
15391 2017-09-18 Richard Biener <rguenther@suse.de>
15393 PR tree-optimization/79622
15394 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
15396 (build_cross_bb_scalars_use): Likewise.
15398 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
15400 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
15402 2017-09-18 Alan Modra <amodra@gmail.com>
15405 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
15406 stack_pointer_rtx for count 0. Update comments. Break up
15407 large rtl expression.
15409 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
15411 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
15412 Increase to 20 bytes.
15413 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
15414 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
15415 or avx version of the stub.
15417 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
15420 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
15421 compute the minimum stack alignment. Also update preferred stack
15422 boundary for leaf functions.
15424 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
15426 PR tree-optimization/82228
15427 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
15430 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15432 * common/config/nds32/nds32-common.c
15433 (nds32_option_optimization_table): Refine formatting.
15434 (nds32_option_optimization_table): Use -fsched-pressure and
15435 -fomit-frame-pointer for specific optimization level.
15437 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15439 * config/nds32/nds32.c: Refine formatting and comments.
15440 * config/nds32/nds32.h: Likewise.
15441 * config/nds32/nds32.md: Likewise.
15442 * config/nds32/nds32-cost.c: Likewise.
15443 * config/nds32/nds32-isr.c: Likewise.
15444 * config/nds32/nds32-md-auxiliary.c: Likewise.
15445 * config/nds32/nds32-multiple.md: Likewise.
15446 * config/nds32/nds32-predicates.c: Likewise.
15448 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
15449 Jakub Jelinek <jakub@redhat.com>
15451 Add support for -std=c++2a.
15452 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
15454 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
15456 2017-09-15 Steve Ellcey <sellcey@cavium.com>
15459 * doc/extend.texi (Common Function Attributes): Add
15460 references to ARM, AArch64, and S/390 specific attributes.
15461 (Function Specific Option Pragmas): Add AArch64 and S/390
15462 to list of back ends that support the target pragma.
15464 2017-09-15 Nathan Sidwell <nathan@acm.org>
15466 * doc/standards.texi: Fix C++17 description. Update URLs for
15469 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
15471 * common.opt (Wcast-align=strict): New warning option.
15472 * doc/invoke.texi: Document -Wcast-align=strict.
15474 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
15476 * cgraph.h (cgraph_thunk_info): Add comments.
15477 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
15478 assert for VIRTUAL_* arguments stricter.
15480 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
15482 PR tree-optimization/71026
15483 * match.pd: Move RDIV patterns from fold-const.c
15484 * fold-const.c (distribute_real_division): Removed.
15485 (fold_binary_loc): Remove calls to distribute_real_divison.
15487 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15489 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
15490 c++1z and gnu++1z as deprecated. Change other references to
15491 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
15492 Change -Wc++1z-compat to -Wc++17-compat.
15493 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
15494 * dwarf2out.c (highest_c_language): Handle C++17.
15495 (gen_compile_unit_die): Likewise.
15497 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15499 PR rtl-optimization/82192
15500 * combine.c (make_extraction): Don't look through non-paradoxical
15501 SUBREGs or TRUNCATE if pos + len is or might be bigger than
15504 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15505 Alan Hayward <alan.hayward@arm.com>
15506 David Sherwood <david.sherwood@arm.com>
15508 * target.def (function_arg_offset): New hook.
15509 * targhooks.h (default_function_arg_offset): Declare.
15510 * targhooks.c (default_function_arg_offset): New function.
15511 * function.c (locate_and_pad_parm): Use
15512 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
15513 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
15514 (TARGET_FUNCTION_ARG_OFFSET): ...this.
15515 * doc/tm.texi: Regenerate.
15516 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
15517 * config/spu/spu.c (spu_function_arg_offset): New function.
15518 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
15519 * system.h (FUNCTION_ARG_OFFSET): Poison.
15521 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15522 Alan Hayard <alan.hayward@arm.com>
15523 David Sherwood <david.sherwood@arm.com>
15525 * target.def (truly_noop_truncation): New hook.
15526 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
15527 than TRULY_NOOP_TRUNCATION.
15528 * hooks.h (hook_bool_uint_uint_true): Declare.
15529 * hooks.c (hook_bool_uint_uint_true): New function.
15530 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
15531 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
15532 * doc/tm.texi: Regenerate.
15533 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
15534 rather than TRULY_NOOP_TRUNCATION in comments.
15535 (simplify_comparison): Likewise.
15536 (record_truncated_value): Likewise.
15537 * expmed.c (extract_bit_field_1): Likewise.
15538 (extract_split_bit_field): Likewise.
15539 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
15540 instead of TRULY_NOOP_TRUNCATION.
15541 * function.c (assign_parm_setup_block): Likewise.
15542 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
15543 * rtlhooks.c: Include target.h.
15544 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
15545 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
15546 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
15547 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
15548 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
15549 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
15550 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
15551 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
15552 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
15553 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
15554 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
15555 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
15556 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
15557 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
15558 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
15559 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
15560 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
15561 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
15562 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
15563 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
15564 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
15565 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
15566 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
15567 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
15568 * config/mips/mips.c (mips_truly_noop_truncation): New function.
15569 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15570 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15571 rather than TRULY_NOOP_TRUNCATION in comments.
15572 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
15573 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
15574 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
15575 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
15576 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
15577 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
15578 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
15579 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
15580 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
15581 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
15582 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
15583 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15584 rather than TRULY_NOOP_TRUNCATION in comments.
15585 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
15586 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
15587 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
15588 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
15589 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
15590 TRULY_NOOP_TRUNCATION condition.
15591 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
15592 (TRULY_NOOP_TRUNCATION): Delete.
15593 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
15594 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
15595 * config/spu/spu.c (spu_truly_noop_truncation): New function.
15596 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15597 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
15598 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
15599 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
15600 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15601 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15602 rather than TRULY_NOOP_TRUNCATION in comments.
15603 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
15604 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
15605 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
15606 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
15607 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
15608 * system.h (TRULY_NOOP_TRUNCATION): Poison.
15610 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
15613 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
15614 (*cmp_ior): Likewise.
15615 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
15616 (*ior_scc_scc_cmp): Likewise.
15617 (*and_scc_scc): Likewise.
15618 (*and_scc_scc_cmp): Likewise.
15620 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15621 Alan Hayard <alan.hayward@arm.com>
15622 David Sherwood <david.sherwood@arm.com>
15624 * target.def (can_change_mode_class): New hook.
15625 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
15626 (hard_regno_nregs): Likewise.
15627 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
15628 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
15629 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
15630 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
15631 (LOAD_EXTEND_OP): Update accordingly.
15632 * doc/tm.texi: Regenerate.
15633 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
15634 CANNOT_CHANGE_MODE_CLASS.
15635 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
15636 (REG_CAN_CHANGE_MODE_P): ...this new macro.
15637 * combine.c (simplify_set): Update accordingly.
15638 * emit-rtl.c (validate_subreg): Likewise.
15639 * recog.c (general_operand): Likewise.
15640 * regcprop.c (mode_change_ok): Likewise.
15641 * reload1.c (choose_reload_regs): Likewise.
15642 (inherit_piecemeal_p): Likewise.
15643 * rtlanal.c (simplify_subreg_regno): Likewise.
15644 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
15645 instead of CANNOT_CHANGE_MODE_CLASS.
15646 (reload_cse_simplify_operands): Likewise.
15647 * reload.c (push_reload): Use targetm.can_change_mode_class
15648 instead of CANNOT_CHANGE_MODE_CLASS.
15649 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
15650 REG_CANNOT_CHANGE_MODE_P.
15651 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15652 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
15653 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15654 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15655 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15656 (arm_can_change_mode_class): New function.
15657 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15658 than CANNOT_CHANGE_MODE_CLASS in comments.
15659 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15660 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
15661 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
15662 (ix86_can_change_mode_class): ...this new function, inverting the
15663 sense of the return value.
15664 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15665 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15666 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15667 (ia64_can_change_mode_class): New function.
15668 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15669 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
15670 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
15671 (m32c_can_change_mode_class): ...this new function, inverting the
15672 sense of the return value.
15673 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15674 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15675 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
15676 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
15677 (mips_can_change_mode_class): ...this new function, inverting the
15678 sense of the return value.
15679 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15680 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15681 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15682 (msp430_can_change_mode_class): New function.
15683 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15684 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
15685 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15686 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15687 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15688 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
15689 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15690 (pa_cannot_change_mode_class): Replace with...
15691 (pa_can_change_mode_class): ...this new function, inverting the
15692 sense of the return value.
15693 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15694 than CANNOT_CHANGE_MODE_CLASS in comments.
15695 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15696 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
15697 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15698 (pdp11_cannot_change_mode_class): Replace with...
15699 (pdp11_can_change_mode_class): ...this new function, inverting the
15700 sense of the return value.
15701 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15702 * config/powerpcspe/powerpcspe-protos.h
15703 (rs6000_cannot_change_mode_class_ptr): Delete.
15704 * config/powerpcspe/powerpcspe.c
15705 (rs6000_cannot_change_mode_class_ptr): Delete.
15706 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15707 (rs6000_option_override_internal): Assign to
15708 targetm.can_change_mode_class instead of
15709 rs6000_cannot_change_mode_class_ptr.
15710 (rs6000_cannot_change_mode_class): Replace with...
15711 (rs6000_can_change_mode_class): ...this new function, inverting the
15712 sense of the return value.
15713 (rs6000_debug_cannot_change_mode_class): Replace with...
15714 (rs6000_debug_can_change_mode_class): ...this new function.
15715 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15716 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
15717 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15718 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15719 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
15721 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
15722 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15723 (rs6000_option_override_internal): Assign to
15724 targetm.can_change_mode_class instead of
15725 rs6000_cannot_change_mode_class_ptr.
15726 (rs6000_cannot_change_mode_class): Replace with...
15727 (rs6000_can_change_mode_class): ...this new function, inverting the
15728 sense of the return value.
15729 (rs6000_debug_cannot_change_mode_class): Replace with...
15730 (rs6000_debug_can_change_mode_class): ...this new function.
15731 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15732 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
15733 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
15734 (s390_can_change_mode_class): ...this new function, inverting the
15735 sense of the return value.
15736 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15737 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15738 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
15739 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15740 (sh_cannot_change_mode_class): Replace with...
15741 (sh_can_change_mode_class): ...this new function, inverting the
15742 sense of the return value.
15743 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15744 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15745 (sparc_can_change_mode_class): New function.
15746 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15747 * config/spu/spu.c (spu_can_change_mode_class): New function.
15748 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15749 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15750 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15751 (visium_can_change_mode_class): New function.
15752 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
15754 2017-09-15 Richard Biener <rguenther@suse.de>
15756 PR tree-optimization/82217
15757 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
15758 but not undefined case.
15760 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15763 * postreload.c (reload_cse_simplify_operands): Skip
15764 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
15766 2017-09-15 Richard Biener <rguenther@suse.de>
15768 PR tree-optimization/68823
15769 * graphite-scop-detection.c (build_alias_set): If we have a
15770 possible dependence check whether we can handle them by just
15771 looking at the DRs DR_ACCESS_FNs.
15772 (build_scops): If build_alias_set fails, fail the SCOP.
15774 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15776 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15777 to support float128 built-in functions that require the ISA 3.0
15779 (BU_FLOAT128_3_HW): Likewise.
15780 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15781 built-in functions.
15782 (FMAF128): Likewise.
15784 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15785 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15786 floating point instructions.
15787 (rs6000_invalid_builtin): Likewise.
15788 (rs6000_builtin_mask_names): Likewise.
15789 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15790 (RS6000_BTM_FLOAT128_HW): Likewise.
15791 (RS6000_BTM_COMMON): Likewise.
15792 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15794 * doc/extend.texi (RS/6000 built-in functions): Document the
15795 IEEE 128-bit floating point square root and fused multiply-add
15796 built-in functions.
15798 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15800 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15801 reg (r2) isn't in the set of registers defined in the prologue.
15803 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15804 Alan Hayward <alan.hayward@arm.com>
15805 David Sherwood <david.sherwood@arm.com>
15807 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15808 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15809 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15810 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15811 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15813 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15814 max_vectorization_factor.
15815 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15817 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15818 Alan Hayward <alan.hayward@arm.com>
15819 David Sherwood <david.sherwood@arm.com>
15821 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15822 (vect_worthwhile_without_simd_p): Declare.
15823 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15824 (vectorizable_reduction): Use it.
15825 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15826 (vectorizable_operation): Likewise.
15828 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15829 Alan Hayward <alan.hayward@arm.com>
15830 David Sherwood <david.sherwood@arm.com>
15832 * tree-vectorizer.h (vect_get_num_copies): New function.
15833 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15834 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15835 (vectorizable_induction): Likewise.
15836 (vectorizable_live_operation): Likewise.
15837 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15838 (vectorizable_bswap): Likewise.
15839 (vectorizable_call): Likewise.
15840 (vectorizable_conversion): Likewise.
15841 (vectorizable_assignment): Likewise.
15842 (vectorizable_shift): Likewise.
15843 (vectorizable_operation): Likewise.
15844 (vectorizable_store): Likewise.
15845 (vectorizable_load): Likewise.
15846 (vectorizable_condition): Likewise.
15847 (vectorizable_comparison): Likewise.
15848 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15850 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15851 Alan Hayward <alan.hayward@arm.com>
15852 David Sherwood <david.sherwood@arm.com>
15854 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15855 of vect_init_vector.
15857 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15858 Alan Hayward <alan.hayward@arm.com>
15859 David Sherwood <david.sherwood@arm.com>
15861 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15862 an inline wrapper that provides a location.
15863 (gimple_build_vector): Likewise.
15864 * gimple-fold.c (gimple_build_vector_from_val): New function.
15865 (gimple_build_vector): Likewise.
15866 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15867 functions to build the initial value. Always return a gimple value.
15868 (get_initial_defs_for_reduction): Likewise. Only compute
15870 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15871 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15872 (vectorizable_induction): Use gimple_build_vector rather than
15875 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15876 Alan Hayward <alan.hayward@arm.com>
15877 David Sherwood <david.sherwood@arm.com>
15879 * target.h (vec_perm_indices): New typedef.
15880 (auto_vec_perm_indices): Likewise.
15881 * optabs-query.h: Include target.h
15882 (can_vec_perm_p): Take a vec_perm_indices *.
15883 * optabs-query.c (can_vec_perm_p): Likewise.
15884 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15885 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15886 * tree-vect-generic.c (lower_vec_perm): Likewise.
15887 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15888 (vect_grouped_load_supported): Likewise.
15889 (vect_shift_permute_load_chain): Likewise.
15890 (vect_permute_store_chain): Use auto_vec_perm_indices.
15891 (vect_permute_load_chain): Likewise.
15892 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15893 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15894 Update uses of can_vec_perm_p.
15895 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15896 mode with a number of elements. Take a vec_perm_indices *.
15897 (vect_create_epilog_for_reduction): Update accordingly.
15898 Use auto_vec_perm_indices.
15899 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15900 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15901 (vect_transform_slp_perm_load): Likewise.
15902 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15903 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15904 (vect_gen_perm_mask_checked): Likewise.
15905 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15906 (vect_gen_perm_mask_checked): Likewise.
15907 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15908 (vectorizable_store): Likewise.
15909 (vectorizable_load): Likewise.
15910 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15911 (vectorizable_bswap): Likewise.
15913 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15914 Alan Hayward <alan.hayward@arm.com>
15915 David Sherwood <david.sherwood@arm.com>
15917 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15918 * tree.c (build_vector): Likewise.
15919 (build_vector_from_ctor): Update accordingly.
15920 (build_vector_from_val): Likewise.
15921 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15922 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15923 * tree-vect-generic.c (add_rshift): Likewise.
15924 (expand_vector_divmod): Likewise.
15925 (optimize_vector_constructor): Likewise.
15926 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15927 (vect_transform_slp_perm_load): Likewise.
15928 (vect_schedule_slp_instance): Likewise.
15929 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15930 (vectorizable_call): Likewise.
15931 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15932 * expmed.c (make_tree): Likewise.
15933 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15934 a vector passed to build_vector.
15935 (fold_convert_const): Likewise.
15936 (exact_inverse): Likewise.
15937 (fold_ternary_loc): Likewise.
15938 (fold_relational_const): Likewise.
15939 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15940 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15941 (const_unop): Likewise. Store the reduction accumulator in a
15942 variable rather than an array.
15943 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15944 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15945 the new vector, rather than constructing it after the input arrays.
15946 (native_interpret_vector): Use auto_vec<tree> when building
15947 a vector passed to build_vector. Add elements in order.
15948 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15949 auto_vec<tree> when building a vector passed to build_vector.
15950 (vect_create_epilog_for_reduction): Likewise.
15951 (vectorizable_induction): Likewise.
15952 (get_initial_def_for_reduction): Likewise. Fix indentation of
15954 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15956 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15957 passed to build_vector.
15959 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15960 Alan Hayward <alan.hayward@arm.com>
15961 David Sherwood <david.sherwood@arm.com>
15963 * tree-core.h (tree_base::u): Add an "nelts" field.
15964 (tree_vector): Use VECTOR_CST_NELTS as the length.
15965 * tree.c (tree_size): Likewise.
15966 (make_vector): Initialize VECTOR_CST_NELTS.
15967 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15968 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15969 TYPE_VECTOR_SUBPARTS.
15970 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15971 as the number of units, setting it from VECTOR_CST_NELTS.
15972 (const_vector_from_tree): Likewise.
15973 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15974 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15975 (fold_negate_expr_1): Likewise.
15976 (fold_convert_const): Likewise.
15977 (const_binop): Likewise. Differentiate the number of output and
15979 (const_unop): Likewise.
15980 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15981 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15982 in cases that did the opposite.
15984 2017-09-14 Richard Biener <rguenther@suse.de>
15986 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15989 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15991 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15993 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15996 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15997 if and where to split a bb, except for splitting before debug insn
15998 sequences followed by non-label real insn. Delete debug insns
15999 in between basic blocks.
16001 * combine.c (make_compound_operation_int): Formatting fixes.
16003 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
16004 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16005 * config/netbsd.h (LINK_EH_SPEC): Likewise.
16006 * config/sol2.h (LINK_EH_SPEC): Likewise.
16007 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16008 * config/s390/linux.h (LINK_SPEC): Likewise.
16009 * config/freebsd.h (LINK_EH_SPEC): Likewise.
16010 * config/openbsd.h (LINK_EH_SPEC): Likewise.
16011 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16012 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
16013 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
16014 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16015 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
16016 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
16018 2017-09-13 Jakub Jelinek <jakub@redhat.com>
16020 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
16022 (ENDFILE_LINUX_SPEC): Likewise.
16023 (LINK_EH_SPEC): Likewise.
16024 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
16025 (LINK_OS_LINUX_SPEC32): Likewise.
16026 (LINK_OS_LINUX_SPEC64): Likewise.
16027 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
16028 (LINK_OS_LINUX_SPEC): Likewise.
16030 2017-09-13 Martin Liska <mliska@suse.cz>
16032 PR middle-end/82154
16033 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
16034 CASE_HIGH is NULL_TREE.
16036 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
16037 Alan Hayward <alan.hayward@arm.com>
16038 David Sherwood <david.sherwood@arm.com>
16040 * target.def (secondary_memory_needed): New hook.
16041 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
16042 instead of SECONDARY_MEMORY_NEEDED.
16043 (secondary_memory_needed_mode): Likewise.
16044 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
16045 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
16046 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
16047 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
16048 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
16049 * doc/tm.texi: Regenerate.
16050 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
16051 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
16052 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16053 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
16054 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
16055 * config/i386/i386.c (inline_secondary_memory_needed): Put the
16056 mode argument first and change the reg_class arguments to reg_class_t.
16057 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
16058 Make static. Update the call to inline_secondary_memory_needed.
16059 (ix86_register_move_cost): Update the call to
16060 inline_secondary_memory_needed.
16061 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16062 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
16064 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
16065 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16067 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
16068 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
16069 * config/mips/mips.c (mips_secondary_memory_needed): Make static
16070 and match hook interface. Add comment from mips.h.
16071 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16072 * config/mmix/mmix.md (truncdfsf2): Refer to
16073 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16075 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
16076 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
16077 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16078 (pa_secondary_memory_needed): New function.
16079 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
16080 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
16081 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16082 (pdp11_secondary_memory_needed): Make static and match hook interface.
16083 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
16084 * config/powerpcspe/powerpcspe-protos.h
16085 (rs6000_secondary_memory_needed_ptr): Delete.
16086 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
16088 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16089 (rs6000_option_override_internal): Assign to
16090 targetm.secondary_memory_needed rather than
16091 rs6000_secondary_memory_needed_ptr.
16092 (rs6000_secondary_memory_needed): Match hook interface.
16093 (rs6000_debug_secondary_memory_needed): Likewise.
16094 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
16095 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
16096 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
16097 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16098 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
16099 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
16101 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
16102 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16103 (rs6000_option_override_internal): Assign to
16104 targetm.secondary_memory_needed rather than
16105 rs6000_secondary_memory_needed_ptr.
16106 (rs6000_secondary_memory_needed): Match hook interface.
16107 (rs6000_debug_secondary_memory_needed): Likewise.
16108 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
16109 * config/s390/s390.c (s390_secondary_memory_needed): New function.
16110 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16111 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
16112 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16113 (sparc_secondary_memory_needed): New function.
16114 * lra-constraints.c (check_and_process_move): Refer to
16115 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16117 (curr_insn_transform): Likewise.
16118 (process_alt_operands): Use targetm.secondary_memory_needed
16119 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16120 (check_secondary_memory_needed_p): Likewise.
16121 (choose_split_class): Likewise.
16122 * reload.c: Unconditionally include code that was previously
16123 conditional on SECONDARY_MEMORY_NEEDED.
16124 (push_secondary_reload): Use targetm.secondary_memory_needed
16125 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16126 (push_reload): Likewise.
16127 * reload1.c: Unconditionally include code that was previously
16128 conditional on SECONDARY_MEMORY_NEEDED.
16129 (choose_reload_regs): Use targetm.secondary_memory_needed
16130 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16131 (gen_reload): Likewise.
16132 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
16134 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
16135 Alan Hayward <alan.hayward@arm.com>
16136 David Sherwood <david.sherwood@arm.com>
16138 * target.def (secondary_memory_needed_mode): New hook:
16139 * targhooks.c (default_secondary_memory_needed_mode): Declare.
16140 * targhooks.h (default_secondary_memory_needed_mode): New function.
16141 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
16142 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
16143 * doc/tm.texi: Regenerate.
16144 * lra-constraints.c (check_and_process_move): Use
16145 targetm.secondary_memory_needed_mode instead of
16146 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
16147 (curr_insn_transform): Likewise.
16148 * reload.c (get_secondary_mem): Likewise.
16149 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16150 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
16152 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16153 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16154 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
16155 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16156 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
16158 * config/powerpcspe/powerpcspe-protos.h
16159 (rs6000_secondary_memory_needed_mode): Delete.
16160 * config/powerpcspe/powerpcspe.c
16161 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16162 (rs6000_secondary_memory_needed_mode): Make static.
16163 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16164 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
16166 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16168 (rs6000_secondary_memory_needed_mode): Make static.
16169 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16170 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
16171 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16172 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16173 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16175 (sparc_secondary_memory_needed_mode): New function.
16176 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
16178 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
16180 * config/aarch64/constraints.md (Umq): New constraint.
16181 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
16183 (mov<mode>): Update condition.
16185 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16187 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
16188 when bitposition is the same.
16190 2017-09-13 Richard Biener <rguenther@suse.de>
16192 * dwarf2out.c (output_die_symbol): Remove.
16193 (output_die): Do not output a DIEs symbol.
16195 2017-09-13 Richard Biener <rguenther@suse.de>
16197 PR middle-end/82128
16198 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
16199 default-def to avoid breaking iterator update with the weird
16200 interaction with cgraph_update_edges_for_call_stmt_node.
16202 2017-09-13 Richard Biener <rguenther@suse.de>
16204 * tree-cfg.c (verify_gimple_assign_binary): Add verification
16205 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
16206 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
16207 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
16209 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
16211 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
16212 Disable pc relative literal load irrespective of
16213 TARGET_FIX_ERR_A53_84341 for default.
16215 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
16217 * config/sparc/sparc.c (output_return): Output the source location of
16218 the insn in the delay slot, if any.
16219 (output_sibcall): Likewise.
16221 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
16224 * common.opt (-static-pie): New alias.
16225 (shared): Negate static-pie.
16226 (-no-pie): Update help text.
16228 (static-pie): New option.
16229 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
16230 -static-pie support.
16231 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
16232 (LINK_EH_SPEC): Likewise.
16233 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16234 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16235 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16236 * gcc.c (LINK_COMMAND_SPEC): Likewise.
16237 (init_gcc_specs): Likewise.
16238 (init_spec): Likewise.
16239 (display_help): Update help message for -pie.
16240 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
16243 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
16245 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
16246 (movdi_aarch64): Likewise.
16247 (movti_aarch64): Likewise.
16249 2017-09-12 Simon Wright <simon@pushface.org>
16252 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
16253 calculation of the minor version, always output as 0.
16255 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16258 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16259 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
16260 on it early, rather than manual conversion late. For
16261 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
16262 instead of performing manual conversion.
16264 2017-09-12 Carl Love <cel@us.ibm.com>
16266 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
16267 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
16269 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16270 VMULOSW): Add definitions.
16271 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16272 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16273 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16274 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16275 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16277 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16279 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
16281 (movti_aarch64): Likewise.
16282 (movdf_aarch64): Likewise.
16283 (movtf_aarch64): Likewise.
16284 (load_pairdi): Likewise.
16285 (store_pairdi): Likewise.
16286 (load_pairdf): Likewise.
16287 (store_pairdf): Likewise.
16288 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
16289 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
16290 (ldr_got_small_<mode>): Likewise.
16291 (ldr_got_small_28k_<mode>): Likewise.
16292 (ldr_got_tiny): Likewise.
16293 * config/aarch64/iterators.md (ldst_sz): New.
16294 (ldpstp_sz): Likewise.
16295 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
16297 (thunderx_load): Split load_8 to load_16.
16298 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
16300 (thunderx2t99_storepair_basic): Split store_8 to store_16.
16301 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
16302 (xgene1_store_pair): Split store_8 to store_16.
16303 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
16304 (falkor_st_0_st_sd): Split store_8 to store_16.
16306 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16308 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
16309 and store1/2/3/4 to store_4/8/12/16.
16310 * config/aarch64/aarch64.md: Update for rename.
16311 * config/arm/arm.md: Likewise.: Likewise.
16312 * config/arm/arm.c: Likewise.
16313 * config/arm/thumb1.md: Likewise.
16314 * config/arm/thumb2.md: Likewise.
16315 * config/arm/vfp.md: Likewise.
16316 * config/arm/arm-generic.md: Likewise.
16317 * config/arm/arm1020e.md: Likewise.
16318 * config/arm/arm1026ejs.md: Likewise.
16319 * config/arm/arm1136jfs.md: Likewise.
16320 * config/arm/arm926ejs.md: Likewise.
16321 * config/arm/cortex-a15.md: Likewise.
16322 * config/arm/cortex-a17.md: Likewise.
16323 * config/arm/cortex-a5.md: Likewise.
16324 * config/arm/cortex-a53.md: Likewise.
16325 * config/arm/cortex-a57.md: Likewise.
16326 * config/arm/cortex-a7.md: Likewise.
16327 * config/arm/cortex-a8.md: Likewise.
16328 * config/arm/cortex-a9.md: Likewise.
16329 * config/arm/cortex-m4.md: Likewise.
16330 * config/arm/cortex-m7.md: Likewise.
16331 * config/arm/cortex-r4.md: Likewise.
16332 * config/arm/exynos-m1.md: Likewise.
16333 * config/arm/fa526.md: Likewise.
16334 * config/arm/fa606te.md: Likewise.
16335 * config/arm/fa626te.md: Likewise.
16336 * config/arm/fa726te.md: Likewise.
16337 * config/arm/fmp626.md: Likewise.
16338 * config/arm/iwmmxt.md: Likewise.
16339 * config/arm/ldmstm.md: Likewise.
16340 * config/arm/marvell-pj4.md: Likewise.
16341 * config/arm/xgene1.md: Likewise.
16342 * config/aarch64/thunderx.md: Likewise.
16343 * config/aarch64/thunderx2t99.md: Likewise.
16344 * config/aarch64/falkor.md: Likewise.
16346 2017-09-12 Martin Liska <mliska@suse.cz>
16348 * attribs.c (private_lookup_attribute): New function.
16349 * attribs.h (private_lookup_attribute): Declared here.
16350 (lookup_attribute): Called from this place.
16352 2017-09-12 Richard Biener <rguenther@suse.de>
16354 PR tree-optimization/82157
16355 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
16356 stmts with side-effects.
16358 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16359 Alan Hayward <alan.hayward@arm.com>
16360 David Sherwood <david.sherwood@arm.com>
16362 * target.def (hard_regno_nregs): New hook.
16363 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
16364 * targhooks.h (default_hard_regno_nregs): Declare.
16365 * targhooks.c (default_hard_regno_nregs): New function.
16366 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
16367 (TARGET_HARD_REGNO_NREGS): ...this hook.
16368 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
16369 (CLASS_MAX_NREGS): Likewise.
16370 * doc/tm.texi: Regenerate.
16371 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
16372 instead of HARD_REGNO_NREGS.
16373 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
16374 HARD_REGNO_NREGS in the comment.
16375 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
16376 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
16377 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
16378 Return an unsigned int.
16379 (TARGET_HARD_REGNO_NREGS): Redefine.
16380 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
16381 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
16382 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16383 (arc_hard_regno_nregs): New function.
16384 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
16385 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
16386 (arm_hard_regno_nregs): New function.
16387 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
16388 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
16389 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
16390 (TARGET_HARD_REGNO_NREGS): Redefine.
16391 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
16392 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
16393 (HARD_REGNO_NREGS): Delete.
16394 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
16395 (cr16_hard_regno_nregs): New function.
16396 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
16397 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
16398 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
16399 (cris_hard_regno_nregs): New function.
16400 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
16401 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
16402 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16403 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
16404 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
16405 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
16406 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
16407 (frv_hard_regno_nregs): Make static. Take and return an
16409 (frv_class_max_nregs): Remove outdated copy of documentation.
16410 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
16411 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
16412 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
16413 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
16414 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
16415 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
16416 (TARGET_HARD_REGNO_NREGS): Redefine.
16417 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
16418 (CLASS_MAX_NREGS): Update comment.
16419 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
16420 (ia64_hard_regno_nregs): New function.
16421 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
16422 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
16423 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
16424 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
16425 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
16427 (m32c_hard_regno_nregs): Likewise. Make static.
16428 (TARGET_HARD_REGNO_NREGS): Redefine.
16429 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
16430 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
16431 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
16432 (m68k_hard_regno_nregs): New function.
16433 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
16434 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
16435 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
16436 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
16437 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
16438 Take and return an unsigned int.
16439 (TARGET_HARD_REGNO_NREGS): Redefine.
16440 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
16441 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16442 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
16443 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
16444 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
16445 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
16446 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
16447 (msp430_hard_regno_nregs): Make static. Take and return an
16449 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
16450 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
16451 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
16452 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
16453 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
16454 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
16455 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
16456 (TARGET_HARD_REGNO_NREGS): Redefine.
16457 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
16458 (PA_HARD_REGNO_NREGS): ...this.
16459 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
16460 (PA_HARD_REGNO_NREGS): ...this.
16461 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16462 (pa_hard_regno_nregs): New function.
16463 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
16464 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
16465 (pdp11_hard_regno_nregs): New function.
16466 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
16467 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
16468 (rs6000_hard_regno_nregs_hook): New function.
16469 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
16470 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
16471 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
16472 Take and return an unsigned int. Move earlier in file.
16473 (TARGET_HARD_REGNO_NREGS): Redefine.
16474 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
16475 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
16476 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
16477 (rl78_hard_regno_nregs): Make static. Take and return an
16479 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
16480 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
16481 (rs6000_hard_regno_nregs_hook): New function.
16482 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
16483 * config/rx/rx.c (rx_hard_regno_nregs): New function.
16484 (TARGET_HARD_REGNO_NREGS): Redefine.
16485 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
16486 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
16487 instead of HARD_REGNO_NREGS.
16488 (s390_hard_regno_nregs): New function.
16489 (s390_hard_regno_mode_ok): Add comment from s390.h.
16490 (TARGET_HARD_REGNO_NREGS): Redefine.
16491 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
16492 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
16493 (sh_hard_regno_nregs): New function.
16494 (sh_pass_in_reg_p): Use it.
16495 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
16496 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16497 (sparc_hard_regno_nregs): New function.
16498 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
16499 * config/spu/spu.c (spu_hard_regno_nregs): New function.
16500 (spu_function_arg_advance): Use it, supplying a valid register number.
16501 (TARGET_HARD_REGNO_NREGS): Redefine.
16502 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
16503 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
16504 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
16505 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
16506 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
16507 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
16508 (CLASS_MAX_NREGS): Remove copy of old documentation.
16509 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
16510 (visium_hard_regno_nregs): New function.
16511 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
16512 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
16513 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16514 xtensa_hard_regno_nregs): New function.
16515 * system.h (HARD_REGNO_NREGS): Poison.
16517 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16519 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
16520 hard_regno_nregs instead of HARD_REGNO_NREGS.
16521 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
16522 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
16523 (c6x_expand_epilogue): Likewise.
16524 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
16525 (frv_read_iacc_argument): Likewise.
16526 * config/sh/sh.c: Include regs.h.
16527 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16528 (regs_used): Likewise.
16529 (output_stack_adjust): Likewise.
16530 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
16531 * expmed.c: Include regs.h.
16532 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16533 * ree.c: Include regs.h.
16534 (combine_reaching_defs): Use hard_regno_nregs instead of
16536 (add_removable_extension): Likewise.
16538 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16540 * regs.h (hard_regno_nregs): Turn into a function.
16541 (end_hard_regno): Update accordingly.
16542 * caller-save.c (setup_save_areas): Likewise.
16543 (save_call_clobbered_regs): Likewise.
16544 (replace_reg_with_saved_mem): Likewise.
16545 (insert_restore): Likewise.
16546 (insert_save): Likewise.
16547 * combine.c (can_change_dest_mode): Likewise.
16548 (move_deaths): Likewise.
16549 (distribute_notes): Likewise.
16550 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
16551 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
16552 (rs6000_split_multireg_move): Likewise.
16553 (rs6000_register_move_cost): Likewise.
16554 (rs6000_memory_move_cost): Likewise.
16555 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
16556 (rs6000_split_multireg_move): Likewise.
16557 (rs6000_register_move_cost): Likewise.
16558 (rs6000_memory_move_cost): Likewise.
16559 * cselib.c (cselib_reset_table): Likewise.
16560 (cselib_lookup_1): Likewise.
16561 * emit-rtl.c (set_mode_and_regno): Likewise.
16562 * function.c (aggregate_value_p): Likewise.
16563 * ira-color.c (setup_profitable_hard_regs): Likewise.
16564 (check_hard_reg_p): Likewise.
16565 (calculate_saved_nregs): Likewise.
16566 (assign_hard_reg): Likewise.
16567 (improve_allocation): Likewise.
16568 (calculate_spill_cost): Likewise.
16569 * ira-emit.c (modify_move_list): Likewise.
16570 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
16571 (ira_hard_reg_in_set_p): Likewise.
16572 * ira.c (setup_reg_mode_hard_regset): Likewise.
16573 (clarify_prohibited_class_mode_regs): Likewise.
16574 (check_allocation): Likewise.
16575 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16576 (lra_setup_reg_renumber): Likewise.
16577 (setup_try_hard_regno_pseudos): Likewise.
16578 (spill_for): Likewise.
16579 (assign_hard_regno): Likewise.
16580 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
16581 * lra-constraints.c (in_class_p): Likewise.
16582 (lra_constraint_offset): Likewise.
16583 (simplify_operand_subreg): Likewise.
16584 (lra_constraints): Likewise.
16585 (split_reg): Likewise.
16586 (split_if_necessary): Likewise.
16587 (invariant_p): Likewise.
16588 (inherit_in_ebb): Likewise.
16589 * lra-lives.c (process_bb_lives): Likewise.
16590 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
16591 (get_hard_regs): Likewise.
16592 (do_remat): Likewise.
16593 * lra-spills.c (assign_spill_hard_regs): Likewise.
16594 * mode-switching.c (create_pre_exit): Likewise.
16595 * postreload.c (reload_combine_recognize_pattern): Likewise.
16596 * recog.c (peep2_find_free_register): Likewise.
16597 * regcprop.c (kill_value_regno): Likewise.
16598 (set_value_regno): Likewise.
16599 (copy_value): Likewise.
16600 (maybe_mode_change): Likewise.
16601 (find_oldest_value_reg): Likewise.
16602 (copyprop_hardreg_forward_1): Likewise.
16603 * regrename.c (check_new_reg_p): Likewise.
16604 (regrename_do_replace): Likewise.
16605 * reload.c (push_reload): Likewise.
16606 (combine_reloads): Likewise.
16607 (find_dummy_reload): Likewise.
16608 (operands_match_p): Likewise.
16609 (find_reloads): Likewise.
16610 (find_equiv_reg): Likewise.
16611 (reload_adjust_reg_for_mode): Likewise.
16612 * reload1.c (count_pseudo): Likewise.
16613 (count_spilled_pseudo): Likewise.
16614 (find_reg): Likewise.
16615 (clear_reload_reg_in_use): Likewise.
16616 (free_for_value_p): Likewise.
16617 (allocate_reload_reg): Likewise.
16618 (choose_reload_regs): Likewise.
16619 (reload_adjust_reg_for_temp): Likewise.
16620 (emit_reload_insns): Likewise.
16621 (delete_output_reload): Likewise.
16622 * rtlanal.c (subreg_get_info): Likewise.
16623 * sched-deps.c (sched_analyze_reg): Likewise.
16624 * sel-sched.c (init_regs_for_mode): Likewise.
16625 (mark_unavailable_hard_regs): Likewise.
16626 (choose_best_reg_1): Likewise.
16627 (verify_target_availability): Likewise.
16628 * valtrack.c (dead_debug_insert_temp): Likewise.
16629 * var-tracking.c (track_loc_p): Likewise.
16630 (emit_note_insn_var_location): Likewise.
16631 * varasm.c (make_decl_rtl): Likewise.
16632 * reginfo.c (choose_hard_reg_mode): Likewise.
16633 (init_reg_modes_target): Refer directly to
16634 this_target_regs->x_hard_regno_nregs.
16636 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16638 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
16639 instead of hard_regno_nregs.
16641 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16643 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
16644 end_hard_regno instead of hard_regno_nregs.
16645 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
16646 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
16647 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
16648 * ira-color.c (improve_allocation): Likewise.
16649 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16650 * lra-lives.c (mark_regno_live): Likewise.
16651 (mark_regno_dead): Likewise.
16652 * lra-remat.c (operand_to_remat): Likewise.
16653 * lra.c (collect_non_operand_hard_regs): Likewise.
16654 * postreload.c (reload_combine_note_store): Likewise.
16655 (move2add_valid_value_p): Likewise.
16656 * reload.c (regno_clobbered_p): Likewise.
16658 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16660 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
16662 * config/v850/v850.c (v850_reorg): Likewise.
16663 * reload.c (refers_to_regno_for_reload_p): Likewise.
16664 (find_equiv_reg): Likewise.
16665 * reload1.c (reload_reg_reaches_end_p): Likewise.
16667 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16669 * caller-save.c (add_used_regs): Use REG_NREGS instead of
16671 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
16672 * config/arm/arm.c (output_move_neon): Likewise.
16673 (arm_attr_length_move_neon): Likewise.
16674 (neon_split_vcombine): Likewise.
16675 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
16676 (c6x_mark_reg_written): Likewise.
16677 (c6x_dwarf_register_span): Likewise.
16678 * config/i386/i386.c (ix86_save_reg): Likewise.
16679 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
16680 (rws_access_reg): Likewise.
16681 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
16682 * mode-switching.c (create_pre_exit): Likewise.
16683 * ree.c (combine_reaching_defs): Likewise.
16684 (add_removable_extension): Likewise.
16685 * regcprop.c (find_oldest_value_reg): Likewise.
16686 (copyprop_hardreg_forward_1): Likewise.
16687 * reload.c (reload_inner_reg_of_subreg): Likewise.
16688 (push_reload): Likewise.
16689 (combine_reloads): Likewise.
16690 (find_dummy_reload): Likewise.
16691 (reload_adjust_reg_for_mode): Likewise.
16692 * reload1.c (find_reload_regs): Likewise.
16693 (forget_old_reloads_1): Likewise.
16694 (reload_reg_free_for_value_p): Likewise.
16695 (reload_adjust_reg_for_temp): Likewise.
16696 (emit_reload_insns): Likewise.
16697 (delete_output_reload): Likewise.
16698 * sel-sched.c (choose_best_reg_1): Likewise.
16699 (choose_best_pseudo_reg): Likewise.
16701 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16702 Alan Hayward <alan.hayward@arm.com>
16703 David Sherwood <david.sherwood@arm.com>
16705 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
16706 * target.def (slow_unaligned_access): New hook.
16707 * targhooks.h (default_slow_unaligned_access): Declare.
16708 * targhooks.c (default_slow_unaligned_access): New function.
16709 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
16710 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
16711 * doc/tm.texi: Regenerate.
16712 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
16713 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
16714 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
16716 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
16717 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
16719 (rs6000_slow_unaligned_access): New function.
16720 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16721 (expand_block_compare): Likewise.
16722 (expand_strn_compare): Likewise.
16723 (rs6000_rtx_costs): Likewise.
16724 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
16725 (riscv_slow_unaligned_access): Likewise.
16726 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
16727 (riscv_slow_unaligned_access_p): ...this and make static.
16728 (riscv_option_override): Update accordingly.
16729 (riscv_slow_unaligned_access): New function.
16730 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16731 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
16732 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16733 (rs6000_slow_unaligned_access): New function.
16734 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16735 (rs6000_rtx_costs): Likewise.
16736 * config/rs6000/rs6000-string.c (expand_block_compare)
16737 (expand_strn_compare): Use targetm.slow_unaligned_access instead
16738 of SLOW_UNALIGNED_ACCESS.
16739 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
16740 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
16741 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
16742 of SLOW_UNALIGNED_ACCESS.
16743 * expmed.c (simple_mem_bitfield_p): Likewise.
16744 * expr.c (alignment_for_piecewise_move): Likewise.
16745 (emit_group_load_1): Likewise.
16746 (emit_group_store): Likewise.
16747 (copy_blkmode_from_reg): Likewise.
16748 (emit_push_insn): Likewise.
16749 (expand_assignment): Likewise.
16750 (store_field): Likewise.
16751 (expand_expr_real_1): Likewise.
16752 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16753 * lra-constraints.c (simplify_operand_subreg): Likewise.
16754 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16755 * gimple-ssa-store-merging.c: Likewise in block comment at start
16757 * tree-ssa-strlen.c: Include target.h.
16758 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
16759 of SLOW_UNALIGNED_ACCESS.
16760 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
16762 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16764 PR rtl-optimization/82185
16765 * expmed.c (emit_store_flag_int): Only test tem if it has been
16768 2017-09-12 Richard Biener <rguenther@suse.de>
16770 PR middle-end/82149
16771 * match.pd ((FTYPE) N CMP CST): Fix typo.
16773 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16775 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16777 (mips_near_type_p): Add 'short_call' attribute as a synonym
16779 * doc/extend.texi (short_call): Document new function attribute.
16781 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16784 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16785 assertion check that in the condition.
16786 (get_atomic_generic_size): Likewise. Before testing if parameter
16787 has pointer type, if it has array type, call for C++
16788 default_conversion to perform array-to-pointer conversion.
16790 2017-09-12 Richard Biener <rguenther@suse.de>
16792 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16793 for operations we cannot scalarize.
16795 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16797 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16798 vectors heap vectors. Clean up comments.
16799 Make visited_bbs a reference.
16800 (profitable_jump_thread_path): Make GC
16801 vectors heap vectors. Clean up comments.
16803 (convert_and_register_jump_thread_path): Make GC vectors heap
16805 (check_subpath_and_update_thread_path): Same. Clean up comments.
16806 Make visited_bbs a reference.
16807 (handle_phi): Abstract common code to to
16808 register_jump_thread_path_if_profitable.
16809 Rename VAR_BB to DEF_BB.
16811 Make GC vectors heap vectors.
16812 Make visited_bbs a reference.
16813 (handle_assignment): Same.
16814 (register_jump_thread_path_if_profitable): New.
16815 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16817 Make GC vectors heap vectors. Clean up comments.
16818 Make visited_bbs a reference.
16819 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16820 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16823 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16826 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16827 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16829 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16831 Revert r251800 and r251799.
16833 2017-09-11 Martin Jambor <mjambor@suse.cz>
16836 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16837 arguments in advance.
16838 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16839 use it to find predecessor edges.
16840 (naive_outof_ssa): Collect vector of predecessors.
16842 2017-09-08 Jason Merrill <jason@redhat.com>
16844 PR c++/70029 - ICE with ref-qualifier and -flto
16845 * langhooks.h (struct lang_hooks_for_types): Add
16846 copy_lang_qualifiers.
16847 * attribs.c (build_type_attribute_qual_variant): Use it.
16848 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16850 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16851 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16853 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16856 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16857 (*mulsi3_sp64): New instruction.
16858 (mulsi3): New expander.
16860 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16862 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16864 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16866 * sancov.c: Include memmodel.h.
16868 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16871 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16874 2017-09-07 Carl Love <cel@us.ibm.com>
16876 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16877 the sldi instruction.
16879 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16881 * sancov.c: Include tm_p.h.
16883 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16886 * output.h (switch_to_other_text_partition): New declaration.
16887 * varasm.c (switch_to_other_text_partition): New function.
16888 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16889 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16890 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16891 to the other text partition before emitting LCL label and switch back
16892 after emitting the word after it.
16894 2017-09-07 Richard Biener <rguenther@suse.de>
16896 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16897 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16898 critical edges here, after loop init.
16899 (pass_data_pre): Remove PROP_no_crit_edges flags.
16900 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16901 for valueization of call args to avoid leaking VN_TOP.
16902 (visit_use): Assert we do not visit default defs.
16903 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16904 Use error_mark_node to more easily detect leaking VN_TOP.
16905 All default-defs are varying, not VN_TOP. Mark them visited.
16906 (run_scc_vn): Make code match comment.
16908 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16910 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16911 OPTION_MASK_FLOAT128_KEYWORD.
16912 (POWERPC_MASKS): Likewise.
16913 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16914 support for the -mfloat128-type option, and make -mfloat128
16915 default on PowerPC Linux systems. Define or undefine
16916 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16917 Define __float128 to be __ieee128 if IEEE 128-bit support is
16918 enabled, or undefine it.
16919 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16920 Delete defining __FLOAT128_TYPE__.
16921 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16922 -mfloat128-type option and make -mfloat128 default on PowerPC
16924 (TARGET_FLOAT128_TYPE): Likewise.
16925 (-mfloat128-type): Likewise.
16926 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16927 Delete the -mfloat128-type option and make -mfloat128 default on
16928 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16929 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16930 128-bit floating point is enabled. Change tests from using
16931 -mfloat128-type to -mfloat128.
16932 (rs6000_mangle_type): Use the correct mangling for the __float128
16933 type even if normal long double is restricted to 64-bits.
16934 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16936 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16937 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16938 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16940 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16941 documentation for -mfloat128.
16943 2017-09-06 Olivier Hainque <hainque@adacore.com>
16945 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16947 2017-09-06 Wish Wu <wishwu007@gmail.com>
16948 Jakub Jelinek <jakub@redhat.com>
16950 * asan.c (initialize_sanitizer_builtins): Add
16951 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16952 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16953 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16954 BT_FN_VOID_UINT64_PTR variables.
16955 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16956 (BT_FN_VOID_UINT16_UINT16): Likewise.
16957 (BT_FN_VOID_UINT32_UINT32): Likewise.
16958 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16959 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16960 (BT_FN_VOID_UINT64_PTR): Likewise.
16961 * common.opt (flag_sanitize_coverage): New variable.
16962 (fsanitize-coverage=trace-pc): Remove.
16963 (fsanitize-coverage=): Add.
16964 * flag-types.h (enum sanitize_coverage_code): New enum.
16965 * fold-const.c (fold_range_test): Disable non-short-circuit
16966 optimization if flag_sanitize_coverage.
16967 (fold_truth_andor): Likewise.
16968 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16969 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16970 (coverage_sanitizer_opts): New array.
16971 (get_closest_sanitizer_option): Add OPTS argument, handle also
16972 OPT_fsanitize_coverage_.
16973 (parse_sanitizer_options): Adjusted to also handle
16974 OPT_fsanitize_coverage_.
16975 (common_handle_option): Add OPT_fsanitize_coverage_.
16976 * sancov.c (instrument_comparison, instrument_switch): New function.
16977 (sancov_pass): Add trace-cmp support.
16978 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16979 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16980 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16981 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16982 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16983 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16984 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16985 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16986 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16987 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16989 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16991 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16992 error. Only quit immediately if parsing is complete.
16993 (BEGIN): Initialize fatal_err and parse_done.
16994 (begin fpu, end fpu): Check number of arguments.
16995 (begin arch, end arch): Likewise.
16996 (begin cpu, end cpu): Likewise.
16997 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16998 (optalias): Likewise.
17000 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
17002 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
17003 * config/arm/arm-isa.h: Delete. Move definitions to ...
17004 * arm-cpus.in: ... here. Use new feature and fgroup values.
17005 * config/arm/arm.c (arm_option_override): Use lower case for feature
17007 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
17008 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
17009 * config/arm/parsecpu.awk (END): Add new command 'isa'.
17011 (print_isa_bits_for): New function.
17012 (gen_isa): New function.
17013 (gen_comm_data): Use print_isa_bits_for.
17014 (define feature): New keyword.
17015 (define fgroup): New keyword.
17016 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
17017 (arm-isa.h): Add rule to generate file.
17018 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
17019 case for feature bit names.
17021 2017-09-06 Richard Biener <rguenther@suse.de>
17023 * tree-ssa-pre.c (NECESSARY): Remove.
17024 (create_expression_by_pieces): Do not touch pass-local flags.
17025 (insert_into_preds_of_block): Likewise.
17026 (do_pre_regular_insertion): Likewise.
17027 (eliminate_insert): Likewise.
17028 (eliminate_dom_walker::before_dom_children): Likewise.
17029 (fini_eliminate): Do not look at inserted_exprs.
17030 (mark_operand_necessary): Remove.
17031 (remove_dead_inserted_code): Replace with simple work-list
17032 algorithm based on inserted_exprs and SSA uses.
17033 (pass_pre::execute): Re-order fini_eliminate and
17034 remove_dead_inserted_code.
17036 2017-09-06 Olivier Hainque <hainque@adacore.com>
17038 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
17039 for VxWorks 7. Adjust surrounding comments.
17041 2017-09-06 Richard Biener <rguenther@suse.de>
17043 * gimple-ssa-strength-reduction.c
17044 (find_candidates_dom_walker::before_dom_children): Also allow
17047 2017-09-06 Richard Biener <rguenther@suse.de>
17049 PR tree-optimization/82108
17050 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
17051 for gap in the non-permutation SLP case.
17053 2017-09-06 Martin Jambor <mjambor@suse.cz>
17055 PR tree-optimization/82078
17056 * tree-sra.c (sort_and_splice_var_accesses): Move call to
17057 add_access_to_work_queue...
17058 (build_accesses_from_assign): ...here.
17059 (propagate_all_subaccesses): Make sure racc is the group
17060 representative, if there is one.
17062 2017-09-06 Jakub Jelinek <jakub@redhat.com>
17064 PR middle-end/82095
17065 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
17068 2017-09-06 Richard Biener <rguenther@suse.de>
17070 * gimple-ssa-strength-reduction.c
17071 (find_candidates_dom_walker::before_doom_children): Use a
17072 type and not a mode check.
17074 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
17077 * config/arm/predicates.md (arm_general_adddi_operand): Create new
17079 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
17081 2017-09-05 Jeff Law <law@redhat.com>
17083 PR tree-optimization/64910
17084 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
17085 cases where we have 3 or more operands.
17087 2017-09-05 Jakub Jelinek <jakub@redhat.com>
17089 PR middle-end/81768
17090 * omp-low.c (lower_omp_for): Recompute tree invariant if
17091 gimple_omp_for_initial/final is ADDR_EXPR.
17093 PR middle-end/81768
17094 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
17095 into gimple val before gimplification fo the COND_EXPR.
17097 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
17099 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
17100 REGION_COPY argument.
17101 (thread_through_all_blocks): Remove unused argument to
17102 duplicate_thread_path.
17104 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17105 Alan Hayward <alan.hayward@arm.com>
17106 David Sherwood <david.sherwood@arm.com>
17108 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
17109 Take a scalar_mode rather than a machine_mode.
17110 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17111 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
17112 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17113 (aarch64_gen_adjusted_ldpstp): Likewise.
17114 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
17116 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17117 Alan Hayward <alan.hayward@arm.com>
17118 David Sherwood <david.sherwood@arm.com>
17120 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
17121 Take a scalar_int_mode instead of a machine_mode.
17122 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
17123 (aarch64_output_scalar_simd_mov_immediate): Likewise.
17124 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
17125 (aarch64_simd_attr_length_rglist): Delete.
17126 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
17127 a scalar_int_mode instead of a machine_mode.
17128 (aarch64_add_offset): Likewise.
17129 (aarch64_internal_mov_immediate): Likewise
17130 (aarch64_add_constant_internal): Likewise.
17131 (aarch64_add_constant): Likewise.
17132 (aarch64_movw_imm): Likewise.
17133 (aarch64_rtx_arith_op_extract_p): Likewise.
17134 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
17135 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
17136 Remove assert that the mode isn't a vector.
17137 (aarch64_output_scalar_simd_mov_immediate): Likewise.
17138 (aarch64_expand_mov_immediate): Update calls after above changes.
17139 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
17140 (aarch64_and_bitmask_imm): Check for scalar integer modes.
17141 (aarch64_move_imm): Likewise.
17142 (aarch64_can_const_movi_rtx_p): Likewise.
17143 (aarch64_strip_extend): Likewise.
17144 (aarch64_extr_rtx_p): Likewise.
17145 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
17146 a CONST_INT when the mode parameter is VOIDmode.
17147 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
17149 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17151 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
17152 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17153 (bitwise_type_for_mode): Update accordingly.
17155 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17157 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
17158 * stor-layout.c (mode_for_size_tree): Likewise.
17159 (mode_for_array): Update accordingly.
17160 (layout_decl): Likewise.
17161 (compute_record_mode): Likewise. Only set the mode once.
17163 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17165 * target.def (get_mask_mode): Change return type to opt_mode.
17167 * doc/tm.texi: Regenerate.
17168 * targhooks.h (default_get_mask_mode): Return an opt_mode.
17169 * targhooks.c (default_get_mask_mode): Likewise.
17170 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17171 * optabs-query.c (can_vec_mask_load_store_p): Update use of
17172 targetm.get_mask_mode.
17173 * tree.c (build_truth_vector_type): Likewise.
17175 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17177 * machmode.h (mode_for_vector): Return an opt_mode.
17178 * stor-layout.c (mode_for_vector): Likewise.
17179 (mode_for_int_vector): Update accordingly.
17180 (layout_type): Likewise.
17181 * config/i386/i386.c (emit_memmov): Likewise.
17182 (ix86_expand_set_or_movmem): Likewise.
17183 (ix86_expand_vector_init): Likewise.
17184 (ix86_get_mask_mode): Likewise.
17185 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
17187 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
17188 * expmed.c (extract_bit_field_1): Likewise.
17189 * expr.c (expand_expr_real_2): Likewise.
17190 * optabs-query.c (can_vec_perm_p): Likewise.
17191 (can_vec_mask_load_store_p): Likewise.
17192 * optabs.c (expand_vec_perm): Likewise.
17193 * targhooks.c (default_get_mask_mode): Likewise.
17194 * tree-vect-stmts.c (vectorizable_store): Likewise.
17195 (vectorizable_load): Likewise.
17196 (get_vectype_for_scalar_type_and_size): Likewise.
17198 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17200 * machmode.h (mode_for_int_vector): New function.
17201 * stor-layout.c (mode_for_int_vector): Likewise.
17202 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
17203 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
17204 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17205 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17206 (s390_expand_vcond): Likewise.
17208 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17210 * machmode.h (opt_machine_mode): New type.
17211 (opt_mode<T>): Allow construction from anything that can be
17213 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
17214 (mode_for_size): Return an opt_machine_mode.
17215 * stor-layout.c (mode_for_size): Likewise.
17216 (mode_for_size_tree): Update call accordingly.
17217 (bitwise_mode_for_mode): Likewise.
17218 (make_fract_type): Likewise.
17219 (make_accum_type): Likewise.
17220 * caller-save.c (replace_reg_with_saved_mem): Update call
17222 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17223 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17224 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17225 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17226 * expmed.c (extract_bit_field_1): Likewise.
17227 * reload.c (get_secondary_mem): Likewise.
17228 * varasm.c (assemble_integer): Likewise.
17229 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
17232 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17234 * machmode.h (decimal_float_mode_for_size): New function.
17235 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
17236 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
17237 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
17238 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
17239 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
17240 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
17242 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17244 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
17245 (get_builtin_sync_mode): Likewise.
17246 (expand_ifn_atomic_compare_exchange): Likewise.
17247 (expand_builtin_atomic_clear): Likewise.
17248 (expand_builtin_atomic_test_and_set): Likewise.
17249 (fold_builtin_atomic_always_lock_free): Likewise.
17250 * calls.c (compute_argument_addresses): Likewise.
17251 (emit_library_call_value_1): Likewise.
17252 (store_one_arg): Likewise.
17253 * combine.c (combine_instructions): Likewise.
17254 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
17255 * config/arm/arm.c (arm_function_value): Likewise.
17256 (aapcs_allocate_return_reg): Likewise.
17257 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
17258 * config/i386/i386.c (construct_container): Likewise.
17259 (ix86_gimplify_va_arg): Likewise.
17260 (ix86_expand_sse_cmp): Likewise.
17261 (emit_memmov): Likewise.
17262 (emit_memset): Likewise.
17263 (expand_small_movmem_or_setmem): Likewise.
17264 (ix86_expand_pextr): Likewise.
17265 (ix86_expand_pinsr): Likewise.
17266 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
17267 * config/microblaze/microblaze.c (microblaze_block_move_straight):
17269 * config/mips/mips.c (mips_function_value_1) Likewise.
17270 (mips_block_move_straight): Likewise.
17271 (mips_expand_ins_as_unaligned_store): Likewise.
17272 * config/powerpcspe/powerpcspe.c
17273 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17274 (rs6000_darwin64_record_arg_flush): Likewise.
17275 * config/rs6000/rs6000.c
17276 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17277 (rs6000_darwin64_record_arg_flush): Likewise.
17278 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
17279 (sparc_function_value_1): Likewise.
17280 * config/spu/spu.c (adjust_operand): Likewise.
17281 (spu_emit_branch_or_set): Likewise.
17282 (arith_immediate_p): Likewise.
17283 * emit-rtl.c (gen_lowpart_common): Likewise.
17284 * expr.c (expand_expr_real_1): Likewise.
17285 * function.c (assign_parm_setup_block): Likewise.
17286 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
17287 * reload1.c (alter_reg): Likewise.
17288 * stor-layout.c (mode_for_vector): Likewise.
17289 (layout_type): Likewise.
17291 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17293 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
17294 (spu_convert_move): Likewise.
17295 * lower-subreg.c (resolve_simple_move): Likewise.
17297 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17300 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
17301 define_insn to a define_expand.
17302 (altivec_vsum2sws_direct): New define_insn.
17303 (altivec_vsumsws): Convert from a define_insn to a define_expand.
17305 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17307 * config/arm/arm.c (arm_option_params_internal): Improve setting of
17310 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
17315 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
17316 realign stack if stack alignment needed is less than incoming
17319 2017-09-05 Marek Polacek <polacek@redhat.com>
17322 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
17325 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17327 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
17329 2017-09-05 Richard Biener <rguenther@suse.de>
17331 PR tree-optimization/82084
17332 * fold-const.c (can_native_encode_string_p): Handle wide characters.
17334 2017-09-05 Richard Biener <rguenther@suse.de>
17336 PR tree-optimization/82102
17337 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
17339 2017-09-05 Martin Liska <mliska@suse.cz>
17341 PR tree-optimization/82032
17342 * tree-cfg.c (generate_range_test): New function.
17343 * tree-cfg.h (generate_range_test): Declared here.
17344 * tree-cfgcleanup.c (convert_single_case_switch): New function.
17345 (cleanup_control_expr_graph): Use it.
17346 * tree-switch-conversion.c (try_switch_expansion): Remove
17348 (emit_case_nodes): Use generate_range_test.
17350 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17353 * config/i386/i386.md (*<btsc><mode>_mask): Add
17354 TARGET_USE_BT to insn constraint.
17355 (*btr<mode>_mask): Ditto.
17357 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
17359 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
17360 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
17362 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
17365 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
17366 TARGET_NEON and TARGET_IWMMXT.
17367 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
17368 TARGET_NEON and TARGET_IWMMXT.
17369 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
17371 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17373 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
17374 (ix86_rewrite_tls_address): Ditto.
17375 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
17376 (ix86_rewrite_tls_address_1): Ditto.
17377 (ix86_rewrite_tls_address): Ditto.
17378 * config/i386/predicates.md (tls_address_pattern): New predicate.
17379 * config/i386/i386.md (TLS address splitter): New splitter.
17381 2017-09-04 Richard Biener <rguenther@suse.de>
17383 PR tree-optimization/82084
17384 * fold-const.h (can_native_encode_string_p): Declare.
17385 * fold-const.c (can_native_encode_string_p): Factor out from ...
17386 (native_encode_string): ... here.
17387 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
17388 vectorizing stores from constants we later cannot handle.
17390 2017-09-04 Marek Polacek <polacek@redhat.com>
17393 * doc/invoke.texi: Update -Wtautological-compare documentation.
17395 2017-09-04 Jeff Law <law@redhat.com>
17397 PR tree-optimization/64910
17398 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
17399 swap the first and last operand if the last is a constant.
17401 2017-09-04 Marek Polacek <polacek@redhat.com>
17404 * convert.c (do_narrow): When sanitizing signed integer overflows,
17405 bail out for signed types.
17406 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
17408 2017-09-04 Richard Biener <rguenther@suse.de>
17410 PR tree-optimization/82060
17411 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17412 Move devirtualization after stmt folding and before EH/AB/noreturn
17413 cleanup to get the stmt refs canonicalized. Use a bool instead
17414 of gimple_modified_p since that doesn't work for NOPs. Schedule
17415 NOPs generated by folding for removal.
17417 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17418 Alan Hayward <alan.hayward@arm.com>
17419 David Sherwood <david.sherwood@arm.com>
17421 * coretypes.h (pad_direction): New enum.
17422 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
17423 (FUNCTION_ARG_PADDING): Likewise.
17424 * target.def (function_arg_padding): New hook.
17425 * targhooks.h (default_function_arg_padding): Declare.
17426 * targhooks.c (default_function_arg_padding): New function.
17427 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
17428 (TARGET_FUNCTION_ARG_PADDING): ...this.
17429 * doc/tm.texi: Regenerate.
17430 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
17431 instead of direction.
17432 (compute_argument_addresses): Likewise.
17433 (load_register_parameters): Likewise.
17434 (emit_library_call_value_1): Likewise.
17435 (store_one_arg): Use targetm.calls.function_arg_padding instead
17436 of FUNCTION_ARG_PADDING.
17437 (must_pass_in_stack_var_size_or_pad): Likewise.
17438 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
17439 (emit_group_store): Likewise.
17440 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
17441 instead of FUNCTION_ARG_PADDING.
17442 (emit_push_insn): Likewise, and propagate enum change throughout
17444 * function.h (direction): Delete.
17445 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
17447 * function.c (assign_parm_find_stack_rtl): Likewise.
17448 (assign_parm_setup_block_p): Likewise.
17449 (assign_parm_setup_block): Likewise.
17450 (gimplify_parameters): Likewise.
17451 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
17452 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
17454 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
17455 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17456 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
17457 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
17458 (aarch64_function_arg_padding): ...this new function.
17459 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
17460 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17461 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
17462 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17463 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
17464 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17465 (arm_pad_arg_upward): Replace with...
17466 (arm_function_arg_padding): ...this new function.
17467 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
17469 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
17470 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
17471 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17472 (ia64_hpux_function_arg_padding): Replace with...
17473 (ia64_function_arg_padding): ...this new function. Use pad_direction
17474 instead of direction. Check for TARGET_HPUX.
17475 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
17476 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17477 (iq2000_function_arg_padding): New function.
17478 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
17479 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
17480 (mips_function_arg_padding): ...this new function.
17481 (mips_pad_reg_upward): Update accordingly.
17482 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17483 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
17484 targetm.calls.function_arg_padding.
17485 (FUNCTION_ARG_PADDING): Delete.
17486 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17487 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
17488 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17489 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
17490 (nios2_block_reg_padding): Return pad_direction instead of direction.
17491 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
17492 instead of direction.
17493 (nios2_function_arg_padding): Likewise. Make static.
17494 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17495 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
17496 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
17497 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
17498 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17499 (pa_function_arg_padding): Make static. Return pad_direction instead
17501 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
17502 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17503 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
17504 instead of direction. Use targetm.calls.function_arg_padding.
17505 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
17506 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17507 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
17508 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
17509 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
17511 (function_arg_padding): Rename to...
17512 (rs6000_function_arg_padding): ...this. Make static. Return
17513 pad_direction instead of direction.
17514 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17515 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
17516 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17517 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
17518 instead of direction. Use targetm.calls.function_arg_padding.
17519 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
17520 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17521 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
17522 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
17523 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17524 (function_arg_padding): Rename to...
17525 (rs6000_function_arg_padding): ...this. Make static. Return
17526 pad_direction instead of direction.
17527 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17528 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
17529 * config/s390/s390.c (s390_function_arg_padding): New function.
17530 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17531 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
17532 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
17533 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17534 (function_arg_padding): Rename to...
17535 (sparc_function_arg_padding): ...this. Make static. Return
17536 pad_direction instead of direction.
17537 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
17538 * config/spu/spu.c (spu_function_arg_padding): New function.
17539 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17540 * system.h (FUNCTION_ARG_PADDING): Poison.
17542 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17543 Alan Hayward <alan.hayward@arm.com>
17544 David Sherwood <david.sherwood@arm.com>
17546 * target.def (modes_tieable_p): New hook.
17547 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
17548 (TARGET_MODES_TIEABLE_P): ...this.
17549 * doc/tm.texi.in: Regenerate.
17550 * hooks.h (hook_bool_mode_mode_true): Declare.
17551 * hooks.c (hook_bool_mode_mode_true): New function.
17552 * combine.c (subst): Use targetm.modes_tieable_p instead of
17554 * dse.c (find_shift_sequence): Likewise.
17555 * expmed.c (extract_low_bits): Likewise.
17556 * lower-subreg.c: Include target.h.
17557 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
17559 * rtlanal.c (rtx_cost): Likewise.
17560 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
17561 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
17562 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
17563 (TARGET_MODES_TIEABLE_P): Redefine.
17564 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
17565 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
17566 (TARGET_MODES_TIEABLE_P): Redefine.
17567 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
17568 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
17569 (arc_modes_tieable_p): New function.
17570 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
17571 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
17572 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
17573 (arm_modes_tieable_p): Make static.
17574 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
17575 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
17576 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
17577 (TARGET_MODES_TIEABLE_P): Redefine.
17578 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
17579 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
17580 (TARGET_MODES_TIEABLE_P): Redefine.
17581 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
17582 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
17583 (cr16_modes_tieable_p): New function.
17584 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
17585 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
17586 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
17587 (TRULY_NOOP_TRUNCATION): Update comment.
17588 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
17589 (TRULY_NOOP_TRUNCATION): Update comment.
17590 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
17591 (frv_modes_tieable_p): New function.
17592 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
17593 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
17594 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
17595 (TARGET_MODES_TIEABLE_P): Redefine.
17596 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
17597 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
17598 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
17599 (TARGET_MODES_TIEABLE_P): Redefine.
17600 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
17601 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
17602 (ia64_modes_tieable_p): New function.
17603 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
17604 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
17605 (iq2000_modes_tieable_p): New function.
17606 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
17607 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
17608 (lm32_modes_tieable_p): New function.
17609 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
17610 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
17611 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
17612 (TARGET_MODES_TIEABLE_P): Redefine.
17613 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
17614 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
17615 (m32r_modes_tieable_p): New function.
17616 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
17617 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
17618 (m68k_modes_tieable_p): New function.
17619 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
17620 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
17621 (mcore_modes_tieable_p): New function.
17622 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
17623 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
17625 (TARGET_MODES_TIEABLE_P): Redefine.
17626 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
17627 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
17628 * config/mips/mips.c (mips_modes_tieable_p): Make static.
17629 (TARGET_MODES_TIEABLE_P): Redefine.
17630 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
17631 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
17632 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
17633 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
17634 (mn10300_modes_tieable_p): ...this and make static.
17635 (TARGET_MODES_TIEABLE_P): Redefine.
17636 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
17637 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
17638 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
17639 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
17640 (msp430_modes_tieable_p): Make static.
17641 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
17642 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
17643 (TARGET_MODES_TIEABLE_P): Redefine.
17644 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
17645 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
17646 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
17647 (TARGET_MODES_TIEABLE_P): Redefine.
17648 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
17649 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
17650 * config/pa/pa.c (pa_modes_tieable_p): Make static.
17651 (TARGET_MODES_TIEABLE_P): Redefine.
17652 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
17653 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
17654 (pdp11_modes_tieable_p): New function.
17655 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
17656 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
17657 (rs6000_modes_tieable_p): New function.
17658 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17659 * config/powerpcspe/powerpcspe.md: Update comment.
17660 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
17661 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
17662 (TARGET_MODES_TIEABLE_P): Redefine.
17663 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
17664 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
17665 (rl78_modes_tieable_p): New function.
17666 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
17667 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
17668 (rs6000_modes_tieable_p): New function.
17669 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17670 * config/rs6000/rs6000.md: Update comment.
17671 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
17672 * config/rx/rx.c (rx_modes_tieable_p): New function.
17673 (TARGET_MODES_TIEABLE_P): Redefine.
17674 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
17675 * config/s390/s390.c (s390_modes_tieable_p): New function.
17676 (TARGET_MODES_TIEABLE_P): Redefine.
17677 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
17678 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
17679 (sh_modes_tieable_p): New function.
17680 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
17681 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
17682 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
17683 (sparc_modes_tieable_p): Make static.
17684 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
17685 * config/spu/spu.c (spu_modes_tieable_p): New function.
17686 (TARGET_MODES_TIEABLE_P): Redefine.
17687 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
17688 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
17689 (TARGET_MODES_TIEABLE_P): Redefine.
17690 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
17691 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
17692 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
17693 * config/v850/v850.c (v850_modes_tieable_p): New function.
17694 (TARGET_MODES_TIEABLE_P): Redefine.
17695 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
17696 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
17697 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
17698 (visium_modes_tieable_p): New function.
17699 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
17700 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
17701 (xtensa_modes_tieable_p): New function.
17702 * system.h (MODES_TIEABLE_P): Poison.
17704 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17705 Alan Hayward <alan.hayward@arm.com>
17706 David Sherwood <david.sherwood@arm.com>
17708 * target.def (hard_regno_mode_ok): New hook.
17709 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
17710 (TARGET_HARD_REGNO_MODE_OK): ...this.
17711 * doc/tm.texi.in: Regenerate.
17712 * hooks.h (hook_bool_uint_mode_true): Declare.
17713 * hooks.c (hook_bool_uint_mode_true): New function.
17714 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
17715 HARD_REGNO_MODE_OK.
17716 * genpreds.c (write_insn_preds_c): Add an include of target.h.
17717 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
17718 instead of HARD_REGNO_MODE_OK.
17719 * caller-save.c: Include target.h.
17720 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
17721 HARD_REGNO_MODE_OK.
17722 * combine.c (can_combine_p): Likewise.
17723 (combinable_i3pat): Likewise.
17724 (can_change_dest_mode): Likewise.
17725 * expr.c (init_expr_target): Likewise.
17726 (convert_move): Likewise.
17727 (convert_modes): Likewise.
17728 * ira.c (setup_prohibited_class_mode_regs): Likewise.
17729 (setup_prohibited_mode_move_regs): Likewise.
17730 * ira.h (target_ira): Likewise.
17731 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17732 * lra-constraints.c (process_alt_operands): Likewise.
17733 (split_reg): Likewise.
17734 * recog.c (peep2_find_free_register): Likewise.
17735 * ree.c (combine_reaching_defs): Likewise.
17736 * regcprop.c (maybe_mode_change): Likewise.
17737 * reginfo.c (init_reg_sets_1): Likewise.
17738 (choose_hard_reg_mode): Likewise.
17739 (simplifiable_subregs): Likewise.
17740 * regrename.c (check_new_reg_p): Likewise.
17741 * reload.c (find_valid_class): Likewise.
17742 (find_valid_class_1): Likewise.
17743 (reload_inner_reg_of_subreg): Likewise.
17744 (push_reload): Likewise.
17745 (combine_reloads): Likewise.
17746 (find_dummy_reload): Likewise.
17747 (find_reloads): Likewise.
17748 * reload1.c (find_reg): Likewise.
17749 (set_reload_reg): Likewise.
17750 (allocate_reload_reg): Likewise.
17751 (choose_reload_regs): Likewise.
17752 (reload_adjust_reg_for_temp): Likewise.
17753 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
17754 (simplify_subreg_regno): Likewise.
17755 * sel-sched.c (init_regs_for_mode): Likewise.
17756 * varasm.c (make_decl_rtl): Likewise.
17757 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
17758 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
17759 HARD_REGNO_MODE_OK.
17760 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
17761 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
17762 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17763 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
17764 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
17765 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17766 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
17767 (arc_mode_class): Delete.
17768 (HARD_REGNO_MODE_OK): Delete.
17769 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17770 (arc_hard_regno_mode_ok): Rename old array to...
17771 (arc_hard_regno_mode_ok_modes): ...this.
17772 (arc_conditional_register_usage): Update accordingly.
17773 (arc_mode_class): Make static.
17774 (arc_hard_regno_mode_ok): New function.
17775 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17776 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17777 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17778 (arm_hard_regno_mode_ok): Make static.
17779 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17780 HARD_REGNO_MODE_OK.
17781 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17782 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17783 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17785 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17786 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17787 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17788 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17789 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17790 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17791 * config/bfin/predicates.md (valid_reg_operand): Use
17792 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17793 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17794 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17795 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17796 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17797 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17798 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17799 (cr16_hard_regno_mode_ok): Make static and return a bool.
17800 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17801 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17802 (cris_hard_regno_mode_ok): New function.
17803 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17804 (epiphany_mode_class): Delete.
17805 (HARD_REGNO_MODE_OK): Delete.
17806 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17807 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17808 (hard_regno_mode_ok): Rename to...
17809 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17810 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17811 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17812 HARD_REGNO_MODE_OK.
17813 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17814 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17815 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17816 (frv_hard_regno_mode_ok): Make static and return a bool.
17817 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17818 HARD_REGNO_MODE_OK.
17819 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17820 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17821 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17822 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17824 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17825 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17826 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17827 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17829 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17830 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17831 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17832 (ia64_hard_regno_mode_ok): New function.
17833 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17834 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17835 (iq2000_hard_regno_mode_ok): New function.
17836 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17837 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17838 (lm32_hard_regno_mode_ok): New function.
17839 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17840 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17841 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17842 instead of HARD_REGNO_MODE_OK.
17843 (m32c_hard_regno_ok): Rename to...
17844 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17845 (m32c_cannot_change_mode_class): Update accordingly.
17846 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17847 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17848 (m32r_mode_class): Delete.
17849 (HARD_REGNO_MODE_OK): Delete.
17850 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17851 (m32r_hard_regno_mode_ok): Rename to...
17852 (m32r_hard_regno_modes): ...this.
17853 (m32r_mode_class): Make static.
17854 (m32r_hard_regno_mode_ok): New function.
17855 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17856 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17857 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17858 (m68k_hard_regno_mode_ok): Make static.
17859 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17860 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17861 (mcore_hard_regno_mode_ok): New function.
17862 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17863 (HARD_REGNO_MODE_OK): Delete.
17864 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17866 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17867 (microblaze_hard_regno_mode_ok): New function.
17868 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17869 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17870 (mips_hard_regno_mode_ok): Delete.
17871 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17872 (mips_hard_regno_mode_ok_p): ...this and make static.
17873 (mips_hard_regno_mode_ok_p): Rename to...
17874 (mips_hard_regno_mode_ok_uncached): ...this.
17875 (mips_hard_regno_mode_ok): New function.
17876 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17877 of HARD_REGNO_MODE_OK.
17878 (mips_option_override): Update after above name changes.
17879 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17880 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17881 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17882 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17883 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17884 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17885 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17886 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17887 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17888 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17889 (msp430_hard_regno_mode_ok): Make static and return a bool.
17890 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17891 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17892 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17894 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17895 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17896 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17897 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17898 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17899 (PA_HARD_REGNO_MODE_OK): ...this
17900 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17901 (PA_HARD_REGNO_MODE_OK): ...this.
17902 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17903 (pa_hard_regno_mode_ok): New function.
17904 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17905 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17906 (pdp11_hard_regno_mode_ok): New function.
17907 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17908 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17910 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17912 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17913 (rs6000_hard_regno_mode_ok): Rename to...
17914 (rs6000_hard_regno_mode_ok_uncached): ...this.
17915 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17916 (rs6000_hard_regno_mode_ok): New function.
17917 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17918 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17919 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17920 (riscv_hard_regno_mode_ok): ...this and make static.
17921 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17922 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17923 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17924 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17925 (rl78_hard_regno_mode_ok): Make static and return bool.
17926 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17927 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17929 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17930 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17931 (rs6000_hard_regno_mode_ok): Rename to...
17932 (rs6000_hard_regno_mode_ok_uncached): ...this.
17933 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17934 (rs6000_hard_regno_mode_ok): New function.
17935 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17936 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17937 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17938 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17939 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17940 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17941 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17942 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17943 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17944 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17945 (sh_hard_regno_mode_ok): Make static.
17946 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17947 instead of HARD_REGNO_MODE_OK.
17948 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17949 (sparc_mode_class): Delete.
17950 (HARD_REGNO_MODE_OK): Delete.
17951 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17952 (hard_regno_mode_classes): Make static.
17953 (sparc_mode_class): Likewise.
17954 (sparc_hard_regno_mode_ok): New function.
17955 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17956 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17957 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17959 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17960 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17961 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17962 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17963 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17964 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17965 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17966 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17967 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17968 (visium_hard_regno_mode_ok): New function.
17969 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17970 instead of HARD_REGNO_MODE_OK.
17971 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17972 (HARD_REGNO_MODE_OK): Delete.
17973 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17974 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17975 (xtensa_option_override): Update accordingly.
17976 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17977 (xtensa_hard_regno_mode_ok): New function.
17978 * system.h (HARD_REGNO_MODE_OK): Poison.
17980 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17981 Alan Hayward <alan.hayward@arm.com>
17982 David Sherwood <david.sherwood@arm.com>
17984 * target.def (hard_regno_call_part_clobbered): New hook.
17985 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17986 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17987 * doc/tm.texi: Regenerate.
17988 * hooks.h (hook_bool_uint_mode_false): Declare.
17989 * hooks.c (hook_bool_uint_mode_false): New function.
17990 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17991 * cselib.c (cselib_process_insn): Use
17992 targetm.hard_regno_call_part_clobbered instead of
17993 HARD_REGNO_CALL_PART_CLOBBERED.
17994 * ira-conflicts.c (ira_build_conflicts): Likewise.
17995 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17996 * lra-constraints.c (need_for_call_save_p): Likewise.
17997 * lra-lives.c: Include target.h.
17998 (check_pseudos_live_through_calls): Use
17999 targetm.hard_regno_call_part_clobbered instead of
18000 HARD_REGNO_CALL_PART_CLOBBERED.
18001 * regcprop.c: Include target.h.
18002 (copyprop_hardreg_forward_1): Use
18003 targetm.hard_regno_call_part_clobbered instead of
18004 HARD_REGNO_CALL_PART_CLOBBERED.
18005 * reginfo.c (choose_hard_reg_mode): Likewise.
18006 * regrename.c (check_new_reg_p): Likewise.
18007 * reload.c (find_equiv_reg): Likewise.
18008 * reload1.c (emit_reload_insns): Likewise.
18009 * sched-deps.c (deps_analyze_insn): Likewise.
18010 * sel-sched.c (init_regs_for_mode): Likewise.
18011 (mark_unavailable_hard_regs): Likewise.
18012 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
18013 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18014 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
18016 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18017 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18018 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
18020 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
18022 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18023 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18024 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
18026 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18027 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18028 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
18030 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18031 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
18033 * config/powerpcspe/powerpcspe.c
18034 (rs6000_hard_regno_call_part_clobbered): New function.
18035 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18036 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18037 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
18039 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18040 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18041 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
18043 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
18044 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
18045 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
18047 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
18048 Alan Hayward <alan.hayward@arm.com>
18049 David Sherwood <david.sherwood@arm.com>
18051 * rtl.h (subreg_memory_offset): Declare.
18052 * emit-rtl.c (subreg_memory_offset): New function.
18053 * expmed.c (store_bit_field_1): Use it.
18054 * expr.c (undefined_operand_subword_p): Likewise.
18055 * simplify-rtx.c (simplify_subreg): Likewise.
18057 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
18059 PR rtl-optimization/57448
18062 * optabs.c (expand_atomic_load): Place compiler memory barriers if
18063 using atomic_load pattern.
18064 (expand_atomic_store): Likewise.
18066 2017-09-04 Jakub Jelinek <jakub@redhat.com>
18069 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
18070 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
18071 handling. Use replace_call_with_value with NULL instead of
18072 gsi_replace, unlink_stmt_vdef and release_defs.
18074 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
18077 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
18079 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
18082 * rtl.h (emit_library_call_value_1): Declare.
18083 (emit_library_call): Replace declaration with a series of overloads.
18084 Remove the parameter count argument.
18085 (emit_library_call_value): Likewise.
18086 * calls.c (emit_library_call_value_1): Make global. Replace varargs
18087 with an "rtx_mode_t *".
18088 (emit_library_call_value): Delete.
18089 (emit_library_call): Likewise.
18090 * asan.c (asan_emit_stack_protection): Update calls accordingly.
18091 (asan_emit_allocas_unpoison): Likewise.
18092 * builtins.c (expand_builtin_powi): Likewise.
18093 (expand_asan_emit_allocas_unpoison): Likewise.
18094 * cfgexpand.c (expand_main_function): Likewise.
18095 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
18096 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
18097 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
18098 * config/arm/arm.c (arm_trampoline_init): Likewise.
18099 (arm_call_tls_get_addr): Likewise.
18100 (arm_expand_divmod_libfunc): Likewise.
18101 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
18102 (smulsi3_highpart): Likewise.
18103 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
18104 (c6x_expand_compare): Likewise.
18105 (c6x_expand_movmem): Likewise.
18106 * config/frv/frv.c (frv_trampoline_init): Likewise.
18107 * config/i386/i386.c (ix86_trampoline_init): Likewise.
18108 (ix86_expand_divmod_libfunc): Likewise.
18109 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
18110 (ia64_expand_compare): Likewise.
18111 (ia64_profile_hook): Likewise.
18112 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
18113 (nonlocal_goto): Likewise.
18114 (restore_stack_nonlocal): Likewise.
18115 * config/m32r/m32r.c (block_move_call): Likewise.
18116 (m32r_trampoline_init): Likewise.
18117 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
18118 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
18119 (m68k_call_m68k_read_tp): Likewise.
18120 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
18121 (microblaze_expand_divide): Likewise.
18122 * config/mips/mips.h (mips_args): Likewise.
18123 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
18124 (MIPS_ICACHE_SYNC): Likewise.
18125 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
18126 (nios2_trampoline_init): Likewise.
18127 * config/pa/pa.c (hppa_tls_call): Likewise.
18128 (pa_trampoline_init): Likewise.
18129 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
18130 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
18131 (expand_strn_compare): Likewise.
18132 (rs6000_generate_compare): Likewise.
18133 (rs6000_expand_float128_convert): Likewise.
18134 (output_profile_hook): Likewise.
18135 (rs6000_trampoline_init): Likewise.
18136 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
18137 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
18138 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
18139 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
18140 (rs6000_generate_compare): Likewise.
18141 (rs6000_expand_float128_convert): Likewise.
18142 (output_profile_hook): Likewise.
18143 (rs6000_trampoline_init): Likewise.
18144 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
18145 * config/sh/sh.c (sh_trampoline_init): Likewise.
18146 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
18147 (sparc_emit_float_lib_cmp): Likewise.
18148 (sparc32_initialize_trampoline): Likewise.
18149 (sparc64_initialize_trampoline): Likewise.
18150 (sparc_profile_hook): Likewise.
18151 * config/spu/spu.c (ea_load_store): Likewise.
18152 * config/spu/spu.md (floatunssidf2): Likewise.
18153 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
18154 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
18155 * config/visium/visium.c (expand_block_move_4): Likewise.
18156 (expand_block_move_2): Likewise.
18157 (expand_block_move_1): Likewise.
18158 (expand_block_set_4): Likewise.
18159 (expand_block_set_2): Likewise.
18160 (expand_block_set_1): Likewise.
18161 (visium_trampoline_init): Likewise.
18162 (visium_profile_hook): Likewise.
18163 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
18164 (xtensa_setup_frame_addresses): Likewise.
18165 (xtensa_trampoline_init): Likewise.
18166 * except.c (sjlj_emit_function_enter): Likewise.
18167 (sjlj_emit_function_exit): Likewise.
18168 * explow.c (allocate_dynamic_stack_space): Likewise.
18169 (probe_stack_range): Likewise.
18170 * expr.c (convert_mode_scalar): Likewise.
18171 * optabs.c (expand_binop): Likewise.
18172 (expand_twoval_binop_libfunc): Likewise.
18173 (expand_unop): Likewise.
18174 (prepare_cmp_insn): Likewise.
18175 (prepare_float_lib_cmp): Likewise.
18176 (expand_float): Likewise.
18177 (expand_fix): Likewise.
18178 (expand_fixed_convert): Likewise.
18179 (maybe_emit_sync_lock_test_and_set): Likewise.
18180 (expand_atomic_compare_and_swap): Likewise.
18181 (expand_mem_thread_fence): Likewise.
18182 (expand_atomic_fetch_op): Likewise.
18184 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
18186 * doc/generic.texi (OpenACC): Adjust URL.
18187 * doc/invoke.texi (C Dialect Options): Ditto.
18189 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
18191 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
18192 predicate for operand 1. Add (m,<S>) constraint.
18193 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
18194 Prevent memory operand 1 with register operand 2.
18196 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
18198 PR rtl-optimization/82024
18199 * combine.c (try_combine): If the combination result is a PARALLEL,
18200 and we only need to retain the SET in there that would be placed
18201 at I2, check that we can place that at I3 instead, before doing so.
18203 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18206 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
18208 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
18209 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
18212 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
18213 Jeff Law <law@redhat.com>
18215 * varasm.c (bss_initializer_p): Do not put constants into .bss
18216 (categorize_decl_for_section): Handle bss_initializer_p returning
18217 false when DECL_INITIAL is NULL.
18219 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18222 * config/s390/s390.c (s390_can_inline_p): New function.
18224 2017-09-01 Jeff Law <law@redhat.com>
18226 PR tree-optimization/82052
18227 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18228 Always initialize the returned slot after a hash table miss
18229 when INSERT is true.
18231 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
18233 * config/s390/s390.md (mem_signal_fence): Remove.
18234 * doc/md.texi (mem_signal_fence): Remove.
18235 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
18237 * target-insns.def (mem_signal_fence): Remove.
18239 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18242 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
18245 * asan.c (create_odr_indicator): Strip name encoding from assembler
18246 name before appending it after __odr_asan_.
18248 2017-09-01 Martin Liska <mliska@suse.cz>
18250 PR tree-optimization/82059
18251 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
18252 frequency only when an edge is redirected.
18254 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18256 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
18257 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
18258 (arc_conditional_register_usage): Remove ARC600 lp_count
18260 (arc_file_start): Emit Tag_ARC_CPU_variation.
18261 (arc_can_use_doloop_p): New conditions to use ZOLs.
18262 (hwloop_fail): New function.
18263 (hwloop_optimize): Likewise.
18264 (hwloop_pattern_reg): Likewise.
18265 (arc_doloop_hooks): New struct, to be used with reorg_loops.
18266 (arc_reorg_loops): New function, calls reorg_loops.
18267 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
18268 (arc600_corereg_hazard): Remove ZOL checking, case handled by
18270 (arc_loop_hazard): Remove function, functionality moved into
18272 (arc_hazard): Remove arc_loop_hazard call.
18273 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
18274 into hwloop_optimize.
18275 (arc_label_align): Remove ZOL handling.
18276 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
18277 * config/arc/arc.md (doloop_begin): Remove pattern.
18278 (doloop_begin_i): Likewise.
18279 (doloop_end_i): Likewise.
18280 (doloop_fallback): Likewise.
18281 (doloop_fallback_m): Likewise.
18282 (doloop_end): Reimplement expand.
18283 (arc_lp): New pattern for LP instruction.
18284 (loop_end): New pattern.
18285 (loop_fail): Likewise.
18286 (decrement_and_branch_until_zero): Likewise.
18287 * config/arc/arc.opt (mlpc-width): New option.
18288 * doc/invoke.texi (mlpc-width): Document option.
18290 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18292 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
18293 (arc_ccfsm_advance): Fix checking for delay slots.
18294 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
18296 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18298 * config/arc/arc.md (movqi_insn): Add stores to save constant long
18300 (movhi_insn): Update store instruction constraint which are saving
18301 6-bit short immediates.
18302 (movsi_insn): Consider also short scaled load operations.
18303 (zero_extendhisi2_i): Use Usd constraint instead of T.
18304 (extendhisi2_i): Add q constraint.
18305 (arc_clzsi2): Add type and length attributes.
18306 (arc_ctzsi2): Likewise.
18307 * config/arc/constraints.md (Usc): Update constraint, the
18308 assembler can parse two relocations for a single instruction.
18310 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18312 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
18313 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
18315 2017-08-31 Olivier Hainque <hainque@adacore.com>
18317 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
18318 match as powerpc-wrs-vxworks*.
18320 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
18322 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
18323 register constraint for by-element operand.
18324 (aarch64_mls_elt_merge<mode>): Likewise.
18326 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18328 * config/arc/arc.c (arc_can_follow_jump): Check for short
18331 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18333 * config.gcc: Use g.opt for arc.
18334 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
18335 functionality moved to ...
18336 (legitimate_scaled_address_p): New function, ...here.
18337 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
18338 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
18339 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
18341 (arc_override_options): Handle G option.
18342 (arc_output_pic_addr_const): Correct function definition.
18343 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
18344 (arc_decl_anon_ns_mem_p): Delete.
18345 (arc_in_small_data_p): Overhaul this function to take into
18346 consideration the value given via G option.
18347 (arc_rewrite_small_data_1): Renamed and corrected old
18348 arc_rewrite_small_data function.
18349 (arc_rewrite_small_data): New function.
18350 (small_data_pattern): Don't use pic_offset_table_rtx.
18351 * config/arc/arc.h (CC1_SPEC): Recognize G option.
18352 * config/arc/simdext.md (movmisalignv2hi): Use
18353 prepare_move_operands function.
18355 (movmisalign*): Likewise.
18356 * doc/invoke.texi (ARC options): Document -G option.
18358 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18360 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
18362 * config/arc/arc.c (arc_print_operand): Output scalled address for
18363 sdata whenever is possible.
18364 (arc_in_small_data_p): Allow sdata for 64bit datum when double
18365 load/stores are available.
18366 (compact_sda_memory_operand): Check for the alignment required by
18367 code density instructions.
18368 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
18370 * config/arc/constraints.md (Usd): Update constraint.
18371 (Us0): New constraint.
18372 (Usc): Update constraint.
18374 2017-08-31 Richard Biener <rguenther@suse.de>
18376 PR middle-end/82054
18377 * dwarf2out.c (dwarf2out_early_global_decl): Process each
18378 function only once.
18380 2017-08-31 Tamar Christina <tamar.christina@arm.com>
18382 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
18383 Resize type_signature.
18385 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18386 Alan Hayward <alan.hayward@arm.com>
18387 David Sherwood <david.sherwood@arm.com>
18389 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
18390 subregs whose inner modes can be stored in GPRs.
18391 (aarch64_classify_index): Likewise.
18393 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18394 Alan Hayward <alan.hayward@arm.com>
18395 David Sherwood <david.sherwood@arm.com>
18397 * config/aarch64/iterators.md (V_cmp_result): Rename to...
18398 (V_INT_EQUIV): ...this.
18399 (v_cmp_result): Rename to...
18400 (v_int_equiv): ...this.
18401 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
18402 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
18403 (copysign<mode>3): Likewise.
18404 (aarch64_simd_bsl<mode>_internal): Likewise.
18405 (aarch64_simd_bsl<mode>): Likewise.
18406 (vec_cmp<mode><mode>): Likewise.
18407 (vcond<mode><mode>): Likewise.
18408 (vcond<v_cmp_mixed><mode>): Likewise.
18409 (vcondu<mode><v_cmp_mixed>): Likewise.
18410 (aarch64_cm<optab><mode>): Likewise.
18411 (aarch64_cmtst<mode>): Likewise.
18412 (aarch64_fac<optab><mode>): Likewise.
18413 (vec_perm_const<mode>): Likewise.
18414 (vcond_mask_<mode><v_cmp_result>): Rename to...
18415 (vcond_mask_<mode><v_int_equiv>): ...this.
18416 (vec_cmp<mode><v_cmp_result>): Rename to...
18417 (vec_cmp<mode><v_int_equiv>): ...this.
18419 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18420 Alan Hayward <alan.hayward@arm.com>
18421 David Sherwood <david.sherwood@arm.com>
18423 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
18425 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
18426 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
18427 (UNSPEC_LD4_DREG): New unspecs.
18428 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
18429 (aarch64_ld2<mode>_dreg_be): Replace with...
18430 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
18432 (aarch64_ld3<mode>_dreg_le)
18433 (aarch64_ld3<mode>_dreg_be): Replace with...
18434 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
18436 (aarch64_ld4<mode>_dreg_le)
18437 (aarch64_ld4<mode>_dreg_be): Replace with...
18438 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
18441 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18443 PR tree-optimization/81987
18444 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
18445 insert an initializer in a location not dominated by the stride
18448 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18450 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
18451 on the entire header of the finally block in the fallthru case.
18453 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18455 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
18457 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
18459 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
18460 rs6000_emit_move_from_cr and call renamed function.
18461 (rs6000_emit_prologue): Call renamed functions.
18462 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
18463 movesi_from_cr, remove volatile CRs.
18465 2017-08-30 Jon Beniston <jon@beniston.com>
18466 Richard Biener <rguenther@suse.de>
18468 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
18469 of VECTOR_MODE_P check.
18470 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
18471 element vector types.
18473 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18475 * df.h (df_read_modify_subreg_p): Remove in favor of...
18476 * rtl.h (read_modify_subreg_p): ...this new function. Take a
18477 const_rtx instead of an rtx.
18478 * cprop.c (local_cprop_find_used_regs): Update accordingly.
18479 * df-problems.c (df_word_lr_mark_ref): Likewise.
18480 * ira-lives.c (mark_pseudo_reg_live): Likewise.
18481 (mark_pseudo_reg_dead): Likewise.
18482 (mark_ref_dead): Likewise.
18483 * reginfo.c (init_subregs_of_mode): Likewise.
18484 * sched-deps.c (sched_analyze_1): Likewise.
18485 * df-scan.c (df_def_record_1): Likewise.
18486 (df_uses_record): Likewise.
18487 (df_read_modify_subreg_p): Remove in favor of...
18488 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
18489 const_rtx instead of an rtx.
18491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18492 Alan Hayward <alan.hayward@arm.com>
18493 David Sherwood <david.sherwood@arm.com>
18495 * rtl.h (partial_subreg_p): New function.
18496 * caller-save.c (save_call_clobbered_regs): Use it.
18497 * calls.c (expand_call): Likewise.
18498 * combine.c (combinable_i3pat): Likewise.
18499 (simplify_set): Likewise.
18500 (make_extraction): Likewise.
18501 (make_compound_operation_int): Likewise.
18502 (gen_lowpart_or_truncate): Likewise.
18503 (force_to_mode): Likewise.
18504 (make_field_assignment): Likewise.
18505 (reg_truncated_to_mode): Likewise.
18506 (record_truncated_value): Likewise.
18507 (move_deaths): Likewise.
18508 * cse.c (record_jump_cond): Likewise.
18509 (cse_insn): Likewise.
18510 * cselib.c (cselib_lookup_1): Likewise.
18511 * expmed.c (extract_bit_field_using_extv): Likewise.
18512 * function.c (assign_parm_setup_reg): Likewise.
18513 * ifcvt.c (noce_convert_multiple_sets): Likewise.
18514 * ira-build.c (create_insn_allocnos): Likewise.
18515 * lra-coalesce.c (merge_pseudos): Likewise.
18516 * lra-constraints.c (match_reload): Likewise.
18517 (simplify_operand_subreg): Likewise.
18518 (curr_insn_transform): Likewise.
18519 * lra-lives.c (process_bb_lives): Likewise.
18520 * lra.c (new_insn_reg): Likewise.
18521 (lra_substitute_pseudo): Likewise.
18522 * regcprop.c (mode_change_ok): Likewise.
18523 (maybe_mode_change): Likewise.
18524 (copyprop_hardreg_forward_1): Likewise.
18525 * reload.c (push_reload): Likewise.
18526 (find_reloads): Likewise.
18527 (find_reloads_subreg_address): Likewise.
18528 * reload1.c (alter_reg): Likewise.
18529 (eliminate_regs_1): Likewise.
18530 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18532 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
18534 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
18535 back to if statements, including unpack.
18537 2017-08-30 Martin Liska <mliska@suse.cz>
18539 PR inline-asm/82001
18540 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
18542 (func_checker::compare_asm_inputs_outputs): ... this function.
18543 (func_checker::compare_gimple_asm): Use the function to compare
18544 also ASM constrains.
18545 * ipa-icf-gimple.h: Rename the function.
18547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18548 Alan Hayward <alan.hayward@arm.com>
18549 David Sherwood <david.sherwood@arm.com>
18551 * coretypes.h (complex_mode): New type.
18552 * gdbhooks.py (build_pretty_printer): Handle it.
18553 * machmode.h (complex_mode): New class.
18554 (complex_mode::includes_p): New function.
18555 (is_complex_int_mode): Likewise.
18556 (is_complex_float_mode): Likewise.
18557 * genmodes.c (get_mode_class): Handle complex mode classes.
18558 * function.c (expand_function_end): Use is_complex_int_mode.
18560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18561 Alan Hayward <alan.hayward@arm.com>
18562 David Sherwood <david.sherwood@arm.com>
18564 * coretypes.h (scalar_mode_pod): New typedef.
18565 * gdbhooks.py (build_pretty_printer): Handle it.
18566 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
18567 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
18568 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
18569 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
18570 as_a <scalar_mode>.
18572 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18573 Alan Hayward <alan.hayward@arm.com>
18574 David Sherwood <david.sherwood@arm.com>
18576 * machmode.h (mode_for_vector): Take a scalar_mode instead
18578 * stor-layout.c (mode_for_vector): Likewise.
18579 * explow.c (promote_mode): Use as_a <scalar_mode>.
18580 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
18582 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18583 Alan Hayward <alan.hayward@arm.com>
18584 David Sherwood <david.sherwood@arm.com>
18586 * target.def (preferred_simd_mode): Take a scalar_mode
18587 instead of a machine_mode.
18588 * targhooks.h (default_preferred_simd_mode): Likewise.
18589 * targhooks.c (default_preferred_simd_mode): Likewise.
18590 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
18591 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
18592 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
18593 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
18594 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
18595 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
18596 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
18597 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
18599 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
18600 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
18601 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
18602 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
18603 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
18604 * doc/tm.texi: Regenerate.
18605 * optabs-query.c (can_vec_mask_load_store_p): Return false for
18608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18609 Alan Hayward <alan.hayward@arm.com>
18610 David Sherwood <david.sherwood@arm.com>
18612 * target.def (scalar_mode_supported_p): Take a scalar_mode
18613 instead of a machine_mode.
18614 * targhooks.h (default_scalar_mode_supported_p): Likewise.
18615 * targhooks.c (default_scalar_mode_supported_p): Likewise.
18616 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
18617 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18618 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
18619 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
18620 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
18621 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
18622 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
18623 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
18624 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
18625 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
18626 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
18627 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
18629 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
18630 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
18631 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18632 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18633 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
18635 * doc/tm.texi: Regenerate.
18637 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18638 Alan Hayward <alan.hayward@arm.com>
18639 David Sherwood <david.sherwood@arm.com>
18641 * coretypes.h (opt_scalar_mode): New typedef.
18642 * gdbhooks.py (build_pretty_printers): Handle it.
18643 * machmode.h (mode_iterator::get_2xwider): Add overload for
18645 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
18647 * expr.c (convert_mode_scalar): Likewise.
18648 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18649 * optabs.c (expand_float): Likewise.
18650 (expand_fix): Likewise.
18651 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18653 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18654 Alan Hayward <alan.hayward@arm.com>
18655 David Sherwood <david.sherwood@arm.com>
18657 * optabs.c (expand_float): Explicitly check for scalars before
18658 using a branching expansion.
18659 (expand_fix): Likewise.
18661 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18662 Alan Hayward <alan.hayward@arm.com>
18663 David Sherwood <david.sherwood@arm.com>
18665 * expr.c (convert_mode): Split scalar handling out into...
18666 (convert_mode_scalar): ...this new function. Treat the modes
18669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18670 Alan Hayward <alan.hayward@arm.com>
18671 David Sherwood <david.sherwood@arm.com>
18673 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
18675 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
18677 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18678 Alan Hayward <alan.hayward@arm.com>
18679 David Sherwood <david.sherwood@arm.com>
18681 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
18682 rather than a machine_mode.
18683 (fixed_from_string): Likewise.
18684 (fixed_convert): Likewise.
18685 (fixed_convert_from_int): Likewise.
18686 (fixed_convert_from_real): Likewise.
18687 (real_convert_from_fixed): Likewise.
18688 * fixed-value.c (fixed_from_double_int): Likewise.
18689 (fixed_from_string): Likewise.
18690 (fixed_convert): Likewise.
18691 (fixed_convert_from_int): Likewise.
18692 (fixed_convert_from_real): Likewise.
18693 (real_convert_from_fixed): Likewise.
18694 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
18696 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18697 Alan Hayward <alan.hayward@arm.com>
18698 David Sherwood <david.sherwood@arm.com>
18700 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
18701 of separate mode class checks. Do not allow vector modes here.
18702 (immed_wide_int_const): Use as_a <scalar_mode>.
18703 * explow.c (trunc_int_for_mode): Likewise.
18704 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
18705 (wi::shwi): Likewise.
18706 (wi::min_value): Likewise.
18707 (wi::max_value): Likewise.
18708 * dwarf2out.c (loc_descriptor): Likewise.
18709 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
18710 for CONST_WIDE_INT.
18712 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18713 Alan Hayward <alan.hayward@arm.com>
18714 David Sherwood <david.sherwood@arm.com>
18716 * tree.h (SCALAR_TYPE_MODE): New macro.
18717 * expr.c (expand_expr_addr_expr_1): Use it.
18718 (expand_expr_real_2): Likewise.
18719 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
18720 (fold_convert_const_fixed_from_int): Likewise.
18721 (fold_convert_const_fixed_from_real): Likewise.
18722 (native_encode_fixed): Likewise
18723 (native_encode_complex): Likewise
18724 (native_encode_vector): Likewise.
18725 (native_interpret_fixed): Likewise.
18726 (native_interpret_real): Likewise.
18727 (native_interpret_complex): Likewise.
18728 (native_interpret_vector): Likewise.
18729 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
18730 (simd_clone_adjust_argument_types): Likewise.
18731 (simd_clone_init_simd_arrays): Likewise.
18732 (simd_clone_adjust): Likewise.
18733 * stor-layout.c (layout_type): Likewise.
18734 * tree.c (build_minus_one_cst): Likewise.
18735 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18736 * tree-inline.c (estimate_move_cost): Likewise.
18737 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
18738 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
18739 (vectorizable_reduction): Likewise.
18740 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
18741 (vect_recog_mixed_size_cond_pattern): Likewise.
18742 (check_bool_pattern): Likewise.
18743 (adjust_bool_pattern): Likewise.
18744 (search_type_for_mask_1): Likewise.
18745 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
18746 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18747 (vectorizable_load): Likewise.
18748 (vectorizable_store): Likewise.
18749 * ubsan.c (ubsan_encode_value): Likewise.
18750 * varasm.c (output_constant): Likewise.
18752 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18753 Alan Hayward <alan.hayward@arm.com>
18754 David Sherwood <david.sherwood@arm.com>
18756 * coretypes.h (scalar_mode): New class.
18757 * machmode.h (scalar_mode): Likewise.
18758 (scalar_mode::includes_p): New function.
18759 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
18760 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
18761 * genmodes.c (get_mode_class): Handle remaining scalar modes.
18762 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
18763 * expmed.c (store_bit_field_1): Likewise.
18764 (extract_bit_field_1): Likewise.
18765 * expr.c (write_complex_part): Likewise.
18766 (read_complex_part): Likewise.
18767 (emit_move_complex_push): Likewise.
18768 (expand_expr_real_2): Likewise.
18769 * function.c (assign_parm_setup_reg): Likewise.
18770 (assign_parms_unsplit_complex): Likewise.
18771 * optabs.c (expand_binop): Likewise.
18772 * rtlanal.c (subreg_get_info): Likewise.
18773 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18774 * varasm.c (output_constant_pool_2): Likewise.
18776 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18777 Alan Hayward <alan.hayward@arm.com>
18778 David Sherwood <david.sherwood@arm.com>
18780 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18782 (expmed_mult_highpart_optab): Likewise.
18783 (expmed_mult_highpart): Likewise.
18785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18786 Alan Hayward <alan.hayward@arm.com>
18787 David Sherwood <david.sherwood@arm.com>
18789 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18790 instead of a machine_mode.
18791 (builtin_memset_read_str): Likewise.
18792 * builtins.c (c_readstr): Likewise.
18793 (builtin_memcpy_read_str): Likewise.
18794 (builtin_strncpy_read_str): Likewise.
18795 (builtin_memset_read_str): Likewise.
18796 (builtin_memset_gen_str): Likewise.
18797 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18798 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18799 instead of a machine_mode.
18800 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18802 (make_extraction): Likewise.
18803 (try_widen_shift_mode): Take and return scalar_int_modes instead
18805 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18806 a scalar_int_mode instead of a machine_mode.
18807 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18808 (avr_addr_space_pointer_mode): Likewise.
18809 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18810 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18811 (msp430_unwind_word_mode): Likewise.
18812 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18813 (spu_addr_space_pointer_mode): Likewise.
18814 (spu_addr_space_address_mode): Likewise.
18815 (spu_libgcc_cmp_return_mode): Likewise.
18816 (spu_libgcc_shift_count_mode): Likewise.
18817 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18818 (rl78_addr_space_pointer_mode): Likewise.
18819 (fl78_unwind_word_mode): Likewise.
18820 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18822 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18823 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18824 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18825 (mips_valid_pointer_mode): Likewise.
18826 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18827 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18828 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18830 (ft32_addr_space_address_mode): Likewise.
18831 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18832 scalar_int_mode instead of a machine_mode.
18833 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18835 (m32c_addr_space_address_mode): Likewise.
18836 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18837 (rs6000_eh_return_filter_mode): Likewise.
18838 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18839 (rs6000_eh_return_filter_mode): Likewise.
18840 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18841 (s390_libgcc_shift_count_mode): Likewise.
18842 (s390_unwind_word_mode): Likewise.
18843 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18845 * target.def (mode_rep_extended): Likewise.
18846 (valid_pointer_mode): Likewise.
18847 (addr_space.valid_pointer_mode): Likewise.
18848 (eh_return_filter_mode): Return a scalar_int_mode rather than
18850 (libgcc_cmp_return_mode): Likewise.
18851 (libgcc_shift_count_mode): Likewise.
18852 (unwind_word_mode): Likewise.
18853 (addr_space.pointer_mode): Likewise.
18854 (addr_space.address_mode): Likewise.
18855 * doc/tm.texi: Regenerate.
18856 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18858 (do_jump): Use scalar_int_mode for local variables.
18859 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18860 rather than a machine_mode.
18861 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18862 (scompare_loc_descriptor_wide): Likewise.
18863 (scompare_loc_descriptor_narrow): Likewise.
18864 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18866 * except.c (sjlj_emit_dispatch_table): Likewise.
18867 (expand_builtin_eh_copy_values): Likewise.
18868 * explow.c (convert_memory_address_addr_space_1): Likewise.
18869 Take a scalar_int_mode rather than a machine_mode.
18870 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18871 than a machine_mode.
18872 (memory_address_addr_space): Use scalar_int_mode for local variables.
18873 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18874 rather than a machine_mode.
18875 * expmed.c (mask_rtx): Likewise.
18876 (init_expmed_one_conv): Likewise.
18877 (expand_mult_highpart_adjust): Likewise.
18878 (extract_high_half): Likewise.
18879 (expmed_mult_highpart_optab): Likewise.
18880 (expmed_mult_highpart): Likewise.
18881 (expand_smod_pow2): Likewise.
18882 (expand_sdiv_pow2): Likewise.
18883 (emit_store_flag_int): Likewise.
18884 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18886 (extract_low_bits): Likewise.
18887 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18889 * expr.c (pieces_addr::adjust): Likewise.
18890 (can_store_by_pieces): Likewise.
18891 (store_by_pieces): Likewise.
18892 (clear_by_pieces_1): Likewise.
18893 (expand_expr_addr_expr_1): Likewise.
18894 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18895 (expand_expr_real_1): Likewise.
18896 (try_casesi): Likewise.
18897 * final.c (shorten_branches): Likewise.
18898 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18899 type of "mode" to machine_mode.
18900 * internal-fn.c (expand_arith_overflow_result_store): Take a
18901 scalar_int_mode rather than a machine_mode.
18902 (expand_mul_overflow): Use scalar_int_mode for local variables.
18903 * loop-doloop.c (doloop_modify): Likewise.
18904 (doloop_optimize): Likewise.
18905 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18906 than a machine_mode.
18907 (expand_doubleword_shift_condmove): Likewise.
18908 (expand_doubleword_shift): Likewise.
18909 (expand_doubleword_clz): Likewise.
18910 (expand_doubleword_popcount): Likewise.
18911 (expand_doubleword_parity): Likewise.
18912 (expand_absneg_bit): Use scalar_int_mode for local variables.
18913 (prepare_float_lib_cmp): Likewise.
18914 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18915 rather than a machine_mode.
18916 (convert_memory_address_addr_space): Likewise.
18917 (get_mode_bounds): Likewise.
18918 (get_address_mode): Return a scalar_int_mode rather than a
18920 * rtlanal.c (get_address_mode): Likewise.
18921 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18922 than a machine_mode.
18923 * targhooks.c (default_mode_rep_extended): Likewise.
18924 (default_valid_pointer_mode): Likewise.
18925 (default_addr_space_valid_pointer_mode): Likewise.
18926 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18927 than a machine_mode.
18928 (default_libgcc_cmp_return_mode): Likewise.
18929 (default_libgcc_shift_count_mode): Likewise.
18930 (default_unwind_word_mode): Likewise.
18931 (default_addr_space_pointer_mode): Likewise.
18932 (default_addr_space_address_mode): Likewise.
18933 * targhooks.h (default_eh_return_filter_mode): Likewise.
18934 (default_libgcc_cmp_return_mode): Likewise.
18935 (default_libgcc_shift_count_mode): Likewise.
18936 (default_unwind_word_mode): Likewise.
18937 (default_addr_space_pointer_mode): Likewise.
18938 (default_addr_space_address_mode): Likewise.
18939 (default_mode_rep_extended): Take a scalar_int_mode rather than
18941 (default_valid_pointer_mode): Likewise.
18942 (default_addr_space_valid_pointer_mode): Likewise.
18943 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18945 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18946 rather than a machine_mode.
18947 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18948 for local variables.
18949 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18950 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18951 than a machine_mode.
18953 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18954 Alan Hayward <alan.hayward@arm.com>
18955 David Sherwood <david.sherwood@arm.com>
18957 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18958 the mode argument to scalar_int_mode.
18959 (do_jump_by_parts_zero_rtx): Likewise.
18960 (do_jump_by_parts_equality_rtx): Likewise.
18961 (do_jump_by_parts_greater): Take a mode argument.
18962 (do_jump_by_parts_equality): Likewise.
18963 (do_jump_1): Update calls accordingly.
18965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18966 Alan Hayward <alan.hayward@arm.com>
18967 David Sherwood <david.sherwood@arm.com>
18969 * is-a.h (safe_dyn_cast): New function.
18970 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18971 (jump_table_for_label): Likewise.
18972 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18973 instead of an rtx_insn *.
18974 (shorten_branches): Use dyn_cast instead of LABEL_P and
18975 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18976 rtx_jump_table_data::get_data_mode.
18977 (final_scan_insn): Likewise.
18979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18980 Alan Hayward <alan.hayward@arm.com>
18981 David Sherwood <david.sherwood@arm.com>
18983 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18984 trying to combine a full-register integer set with a subreg
18987 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18988 Alan Hayward <alan.hayward@arm.com>
18989 David Sherwood <david.sherwood@arm.com>
18991 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18992 that is always either address_mode or pointer_mode.
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 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18999 when the two are known to be equal.
19001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19002 Alan Hayward <alan.hayward@arm.com>
19003 David Sherwood <david.sherwood@arm.com>
19005 * simplify-rtx.c (simplify_const_unary_operation): Use
19006 is_a <scalar_int_mode> instead of checking for a nonzero
19007 precision. Forcibly convert op_mode to a scalar_int_mode
19008 in that case. More clearly differentiate the operand and
19009 result modes and use the former when deciding what the value
19010 of a count-bits operation should be. Use is_int_mode instead
19011 of checking for a MODE_INT. Remove redundant check for whether
19012 this mode has a zero precision.
19014 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19015 Alan Hayward <alan.hayward@arm.com>
19016 David Sherwood <david.sherwood@arm.com>
19018 * optabs.c (widen_leading): Change the type of the mode argument
19019 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
19020 (widen_bswap): Likewise.
19021 (expand_parity): Likewise.
19022 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
19023 (expand_ffs): Likewise.
19024 (epand_unop): Check for scalar integer modes before calling the
19027 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19028 Alan Hayward <alan.hayward@arm.com>
19029 David Sherwood <david.sherwood@arm.com>
19031 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
19033 (expand_expr_real_1): Update call accordingly.
19035 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19036 Alan Hayward <alan.hayward@arm.com>
19037 David Sherwood <david.sherwood@arm.com>
19039 * expmed.c (store_bit_field_using_insv): Add op0_mode and
19040 value_mode arguments. Use scalar_int_mode internally.
19041 (store_bit_field_1): Rename the new integer mode from imode
19042 to op0_mode and use it instead of GET_MODE (op0). Update calls
19043 to store_split_bit_field, store_bit_field_using_insv and
19044 store_fixed_bit_field.
19045 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
19046 Use scalar_int_mode internally. Use a bit count rather than a mode
19047 when calculating the largest bit size for get_best_mode.
19048 Update calls to store_split_bit_field and store_fixed_bit_field_1.
19049 (store_fixed_bit_field_1): Add mode and value_mode arguments.
19050 Remove assertion that OP0 has a scalar integer mode.
19051 (store_split_bit_field): Add op0_mode and value_mode arguments.
19052 Update calls to extract_fixed_bit_field.
19053 (extract_bit_field_using_extv): Add an op0_mode argument.
19054 Use scalar_int_mode internally.
19055 (extract_bit_field_1): Rename the new integer mode from imode to
19056 op0_mode and use it instead of GET_MODE (op0). Update calls to
19057 extract_split_bit_field, extract_bit_field_using_extv and
19058 extract_fixed_bit_field.
19059 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
19060 to extract_split_bit_field and extract_fixed_bit_field_1.
19061 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
19062 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
19063 on the target mode.
19064 (extract_split_bit_field): Add an op0_mode argument. Update call
19065 to extract_fixed_bit_field.
19067 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19068 Alan Hayward <alan.hayward@arm.com>
19069 David Sherwood <david.sherwood@arm.com>
19071 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
19072 * explow.c (hard_function_value): Likewise.
19073 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
19074 convert_to_mode call outside the loop.
19075 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
19076 for the mode iterator. Require the mode specified by max_pieces
19078 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
19080 (copy_blkmode_to_reg): Likewise.
19081 (set_storage_via_setmem): Likewise.
19082 * optabs.c (prepare_cmp_insn): Likewise.
19083 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19084 * stor-layout.c (finish_bitfield_representative): Likewise.
19086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19087 Alan Hayward <alan.hayward@arm.com>
19088 David Sherwood <david.sherwood@arm.com>
19090 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
19091 * expr.c (convert_move): Use them.
19092 (convert_modes): Likewise.
19093 (store_expr_with_bounds): Likewise.
19095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19096 Alan Hayward <alan.hayward@arm.com>
19097 David Sherwood <david.sherwood@arm.com>
19099 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
19100 parameter for the mode of "x". Remove the "known_x", "known_mode"
19101 and "known_ret" arguments. Change the type of the mode argument
19102 to scalar_int_mode.
19103 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
19104 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
19105 (reg_num_sign_bit_copies_for_combine): Likewise.
19106 * rtlanal.c (nonzero_bits1): Likewise.
19107 (num_sign_bit_copies1): Likewise.
19108 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
19109 (reg_num_sign_bit_copies_general): Likewise.
19110 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
19111 (reg_nonzero_bits_general): Likewise.
19113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19114 Alan Hayward <alan.hayward@arm.com>
19115 David Sherwood <david.sherwood@arm.com>
19117 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
19118 than in subroutines. Return 1 for non-integer modes.
19119 (cached_num_sign_bit_copies): Change the type of the mode parameter
19120 to scalar_int_mode.
19121 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
19122 classes. Handle CONST_INT_P first and then check whether X also
19123 has a scalar integer mode. Check the same thing for inner registers
19124 of a SUBREG and for values that are being extended or truncated.
19126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19127 Alan Hayward <alan.hayward@arm.com>
19128 David Sherwood <david.sherwood@arm.com>
19130 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
19131 in subroutines. Return the mode mask for non-integer modes.
19132 (cached_nonzero_bits): Change the type of the mode parameter
19133 to scalar_int_mode.
19134 (nonzero_bits1): Likewise. Remove early exit for other mode
19135 classes. Handle CONST_INT_P first and then check whether X
19136 also has a scalar integer mode.
19138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19139 Alan Hayward <alan.hayward@arm.com>
19140 David Sherwood <david.sherwood@arm.com>
19142 * expr.c (widest_int_mode_for_size): Make the comment match the code.
19143 Return a scalar_int_mode and assert that the size is greater than
19145 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
19146 (op_by_pieces_d::op_by_pieces_d): Likewise.
19147 (op_by_pieces_d::run): Likewise.
19148 (can_store_by_pieces): Likewise.
19150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19151 Alan Hayward <alan.hayward@arm.com>
19152 David Sherwood <david.sherwood@arm.com>
19154 * combine.c (extract_left_shift): Add a mode argument and update
19156 (make_compound_operation_int): Change the type of the mode parameter
19157 to scalar_int_mode and update the call to extract_left_shift.
19159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19160 Alan Hayward <alan.hayward@arm.com>
19161 David Sherwood <david.sherwood@arm.com>
19163 * combine.c (simplify_and_const_int): Change the type of the mode
19164 parameter to scalar_int_mode.
19165 (simplify_and_const_int_1): Likewise. Update recursive call.
19167 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19168 Alan Hayward <alan.hayward@arm.com>
19169 David Sherwood <david.sherwood@arm.com>
19171 * combine.c (simplify_compare_const): Check that the mode is a
19172 scalar_int_mode (rather than VOIDmode) before testing its
19174 (simplify_comparison): Move COMPARISON_P handling out of the
19175 loop and restrict the latter part of the loop to scalar_int_modes.
19176 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
19177 and when considering SUBREG_REGs. Use is_int_mode instead of
19178 checking GET_MODE_CLASS against MODE_INT.
19180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19181 Alan Hayward <alan.hayward@arm.com>
19182 David Sherwood <david.sherwood@arm.com>
19184 * combine.c (try_widen_shift_mode): Move check for equal modes to...
19185 (simplify_shift_const_1): ...here. Use scalar_int_mode for
19186 shift_unit_mode and for modes involved in scalar shifts.
19188 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19189 Alan Hayward <alan.hayward@arm.com>
19190 David Sherwood <david.sherwood@arm.com>
19192 * combine.c (force_int_to_mode): New function, split out from...
19193 (force_to_mode): ...here. Keep xmode up-to-date and use it
19194 instead of GET_MODE (x).
19196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19197 Alan Hayward <alan.hayward@arm.com>
19198 David Sherwood <david.sherwood@arm.com>
19200 * optabs-query.h (extraction_insn::struct_mode): Change type to
19201 opt_scalar_int_mode and update comment.
19202 (extraction_insn::field_mode): Change type to scalar_int_mode.
19203 (extraction_insn::pos_mode): Likewise.
19204 * combine.c (make_extraction): Update accordingly.
19205 * optabs-query.c (get_traditional_extraction_insn): Likewise.
19206 (get_optab_extraction_insn): Likewise.
19207 * recog.c (simplify_while_replacing): Likewise.
19208 * expmed.c (narrow_bit_field_mem): Change the type of the mode
19209 parameter to opt_scalar_int_mode.
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 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
19216 to a scalar_int_mode instead of a machine_mode.
19217 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
19218 (get_best_mode): Return a boolean and use a pointer argument to store
19219 the selected mode. Replace the limit mode parameter with a bit limit.
19220 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
19221 for the values returned by bit_field_mode_iterator::next_mode.
19222 (store_bit_field): Update call to get_best_mode.
19223 (store_fixed_bit_field): Likewise.
19224 (extract_fixed_bit_field): Likewise.
19225 * expr.c (optimize_bitfield_assignment_op): Likewise.
19226 * fold-const.c (optimize_bit_field_compare): Likewise.
19227 (fold_truth_andor_1): Likewise.
19228 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
19229 Update for new type of m_mode.
19230 (get_best_mode): As above.
19232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19233 Alan Hayward <alan.hayward@arm.com>
19234 David Sherwood <david.sherwood@arm.com>
19236 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
19237 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
19238 (store_bit_field): Check is_a <scalar_int_mode> before calling
19239 strict_volatile_bitfield_p.
19240 (extract_bit_field): Likewise.
19242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19243 Alan Hayward <alan.hayward@arm.com>
19244 David Sherwood <david.sherwood@arm.com>
19246 * target.def (cstore_mode): Return a scalar_int_mode.
19247 * doc/tm.texi: Regenerate.
19248 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
19249 * targhooks.h (default_cstore_mode): Likewise.
19250 * targhooks.c (default_cstore_mode): Likewise, using a forced
19252 * expmed.c (emit_cstore): Expect the target of the cstore to be
19255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19256 Alan Hayward <alan.hayward@arm.com>
19257 David Sherwood <david.sherwood@arm.com>
19259 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
19261 (niter_desc): Likewise mode.
19262 (iv_analyze): Add a mode parameter.
19264 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
19265 and change its type to scalar_int_mode.
19266 * loop-iv.c: Update commentary at head of file.
19267 (iv_constant): Pass the mode paraeter before the rtx it describes
19268 and change its type to scalar_int_mode. Remove VOIDmode handling.
19269 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
19270 (iv_extend): Likewise.
19271 (shorten_into_mode): Likewise.
19272 (iv_add): Use scalar_int_mode.
19273 (iv_mult): Likewise.
19274 (iv_shift): Likewise.
19275 (canonicalize_iv_subregs): Likewise.
19276 (get_biv_step_1): Pass the outer_mode parameter before the rtx
19277 it describes and change its mode to scalar_int_mode. Also change
19278 the type of the returned inner_mode to scalar_int_mode.
19279 (get_biv_step): Likewise, turning outer_mode from a pointer
19280 into a direct parameter. Update call to get_biv_step_1.
19281 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
19282 iv_constant and get_biv_step.
19283 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
19284 and change its type to scalar_int_mode. Don't initialise iv->mode
19285 to VOIDmode and remove later checks for its still being VOIDmode.
19286 Update calls to iv_analyze_op and iv_analyze_expr. Check
19287 is_a <scalar_int_mode> when changing the mode under consideration.
19288 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
19289 Update call to iv_analyze_expr.
19290 (iv_analyze_op): Add a mode parameter. Reject subregs whose
19291 inner register is not also a scalar_int_mode. Update call to
19293 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
19294 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
19295 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
19296 separate mode class checks. Update calls to iv_analyze. Remove
19297 fix-up of VOIDmodes after iv_analyze_biv.
19298 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
19299 don't have a scalar_int_mode. Update call to biv_p.
19301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19302 Alan Hayward <alan.hayward@arm.com>
19303 David Sherwood <david.sherwood@arm.com>
19305 * cfgexpand.c (convert_debug_memory_address): Use
19306 as_a <scalar_int_mode>.
19307 * combine.c (expand_compound_operation): Likewise.
19308 (make_extraction): Likewise.
19309 (change_zero_ext): Likewise.
19310 (simplify_comparison): Likewise.
19311 * cse.c (cse_insn): Likewise.
19312 * dwarf2out.c (minmax_loc_descriptor): Likewise.
19313 (mem_loc_descriptor): Likewise.
19314 (loc_descriptor): Likewise.
19315 * expmed.c (init_expmed_one_mode): Likewise.
19316 (synth_mult): Likewise.
19317 (emit_store_flag_1): Likewise.
19318 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
19319 of a comparison with size.
19320 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
19321 (reduce_to_bit_field_precision): Likewise.
19322 * function.c (expand_function_end): Likewise.
19323 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
19324 * loop-doloop.c (doloop_modify): Likewise.
19325 * optabs.c (expand_binop): Likewise.
19326 (expand_unop): Likewise.
19327 (expand_copysign_absneg): Likewise.
19328 (prepare_cmp_insn): Likewise.
19329 (maybe_legitimize_operand): Likewise.
19330 * recog.c (const_scalar_int_operand): Likewise.
19331 * rtlanal.c (get_address_mode): Likewise.
19332 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19333 (simplify_cond_clz_ctz): Likewise.
19334 * tree-nested.c (get_nl_goto_field): Likewise.
19335 * tree.c (build_vector_type_for_mode): Likewise.
19336 * var-tracking.c (use_narrower_mode): Likewise.
19338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19339 Alan Hayward <alan.hayward@arm.com>
19340 David Sherwood <david.sherwood@arm.com>
19342 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
19343 * builtins.c (expand_builtin_signbit): Use it.
19344 * cfgexpand.c (expand_debug_expr): Likewise.
19345 * dojump.c (do_jump): Likewise.
19346 (do_compare_and_jump): Likewise.
19347 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
19348 * expmed.c (make_tree): Likewise.
19349 * expr.c (expand_expr_real_2): Likewise.
19350 (expand_expr_real_1): Likewise.
19351 (try_casesi): Likewise.
19352 * fold-const-call.c (fold_const_call_ss): Likewise.
19353 * fold-const.c (unextend): Likewise.
19354 (extract_muldiv_1): Likewise.
19355 (fold_single_bit_test): Likewise.
19356 (native_encode_int): Likewise.
19357 (native_encode_string): Likewise.
19358 (native_interpret_int): Likewise.
19359 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
19360 * internal-fn.c (expand_addsub_overflow): Likewise.
19361 (expand_neg_overflow): Likewise.
19362 (expand_mul_overflow): Likewise.
19363 (expand_arith_overflow): Likewise.
19364 * match.pd: Likewise.
19365 * stor-layout.c (layout_type): Likewise.
19366 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
19367 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
19368 * tree-ssanames.c (get_range_info): Likewise.
19369 * tree-switch-conversion.c (array_value_type) Likewise.
19370 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
19371 (vect_recog_divmod_pattern): Likewise.
19372 (vect_recog_mixed_size_cond_pattern): Likewise.
19373 * tree-vrp.c (extract_range_basic): Likewise.
19374 (simplify_float_conversion_using_ranges): Likewise.
19375 * tree.c (int_fits_type_p): Likewise.
19376 * ubsan.c (instrument_bool_enum_load): Likewise.
19377 * varasm.c (mergeable_string_section): Likewise.
19378 (narrowing_initializer_constant_valid_p): Likewise.
19379 (output_constant): Likewise.
19381 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19382 Alan Hayward <alan.hayward@arm.com>
19383 David Sherwood <david.sherwood@arm.com>
19385 * machmode.h (NARROWEST_INT_MODE): New macro.
19386 * expr.c (alignment_for_piecewise_move): Use it instead of
19387 GET_CLASS_NARROWEST_MODE (MODE_INT).
19388 (push_block): Likewise.
19389 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
19391 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19394 Alan Hayward <alan.hayward@arm.com>
19395 David Sherwood <david.sherwood@arm.com>
19397 * postreload.c (move2add_valid_value_p): Change the type of the
19398 mode parameter to scalar_int_mode.
19399 (move2add_use_add2_insn): Add a mode parameter and use it instead
19401 (move2add_use_add3_insn): Likewise.
19402 (reload_cse_move2add): Update accordingly.
19404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19405 Alan Hayward <alan.hayward@arm.com>
19406 David Sherwood <david.sherwood@arm.com>
19408 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
19410 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
19411 * optabs.c (expand_unop): Likewise.
19413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19414 Alan Hayward <alan.hayward@arm.com>
19415 David Sherwood <david.sherwood@arm.com>
19417 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
19418 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
19419 (popcount_loc_descriptor): Likewise.
19420 (bswap_loc_descriptor): Likewise.
19421 (rotate_loc_descriptor): Likewise.
19422 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
19423 calling the functions above.
19425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19426 Alan Hayward <alan.hayward@arm.com>
19427 David Sherwood <david.sherwood@arm.com>
19429 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
19431 (try_combine): Likewise.
19432 (simplify_if_then_else): Likewise.
19433 * cse.c (cse_insn): Likewise.
19434 * dwarf2out.c (mem_loc_descriptor): Likewise.
19435 * emit-rtl.c (gen_lowpart_common): Likewise.
19436 * simplify-rtx.c (simplify_truncation): Likewise.
19437 (simplify_binary_operation_1): Likewise.
19438 (simplify_const_relational_operation): Likewise.
19439 (simplify_ternary_operation): Likewise.
19440 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
19442 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19443 Alan Hayward <alan.hayward@arm.com>
19444 David Sherwood <david.sherwood@arm.com>
19446 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
19447 * reload.c (push_reload): Likewise.
19448 (find_reloads): Likewise.
19450 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19451 Alan Hayward <alan.hayward@arm.com>
19452 David Sherwood <david.sherwood@arm.com>
19454 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
19455 (make_compound_operation_int): Likewise.
19456 (change_zero_ext): Likewise.
19457 * expr.c (convert_move): Likewise.
19458 (convert_modes): Likewise.
19459 * fwprop.c (forward_propagate_subreg): Likewise.
19460 * loop-iv.c (get_biv_step_1): Likewise.
19461 * optabs.c (widen_operand): Likewise.
19462 * postreload.c (move2add_valid_value_p): Likewise.
19463 * recog.c (simplify_while_replacing): Likewise.
19464 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19465 (simplify_binary_operation_1): Likewise. Remove redundant
19466 mode equality check.
19468 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19469 Alan Hayward <alan.hayward@arm.com>
19470 David Sherwood <david.sherwood@arm.com>
19472 * combine.c (combine_simplify_rtx): Add checks for
19473 is_a <scalar_int_mode>.
19474 (simplify_if_then_else): Likewise.
19475 (make_field_assignment): Likewise.
19476 (simplify_comparison): Likewise.
19477 * ifcvt.c (noce_try_bitop): Likewise.
19478 * loop-invariant.c (canonicalize_address_mult): Likewise.
19479 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19481 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19482 Alan Hayward <alan.hayward@arm.com>
19483 David Sherwood <david.sherwood@arm.com>
19485 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
19486 is_a <scalar_int_mode> instead of != BLKmode.
19488 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19489 Alan Hayward <alan.hayward@arm.com>
19490 David Sherwood <david.sherwood@arm.com>
19492 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
19493 instead of != VOIDmode.
19494 * combine.c (if_then_else_cond): Likewise.
19495 (change_zero_ext): Likewise.
19496 * dwarf2out.c (mem_loc_descriptor): Likewise.
19497 (loc_descriptor): Likewise.
19498 * rtlanal.c (canonicalize_condition): Likewise.
19499 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
19501 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19502 Alan Hayward <alan.hayward@arm.com>
19503 David Sherwood <david.sherwood@arm.com>
19505 * simplify-rtx.c (simplify_binary_operation_1): Use
19506 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
19508 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19509 Alan Hayward <alan.hayward@arm.com>
19510 David Sherwood <david.sherwood@arm.com>
19512 * wide-int.h (int_traits<unsigned char>) New class.
19513 (int_traits<unsigned short>) Likewise.
19514 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
19515 Use GET_MODE_UNIT_PRECISION and remove redundant test for
19517 * combine.c (set_nonzero_bits_and_sign_copies): Use
19518 is_a <scalar_int_mode>.
19519 (find_split_point): Likewise.
19520 (combine_simplify_rtx): Likewise.
19521 (simplify_logical): Likewise.
19522 (expand_compound_operation): Likewise.
19523 (expand_field_assignment): Likewise.
19524 (make_compound_operation): Likewise.
19525 (extended_count): Likewise.
19526 (change_zero_ext): Likewise.
19527 (simplify_comparison): Likewise.
19528 * dwarf2out.c (scompare_loc_descriptor): Likewise.
19529 (ucompare_loc_descriptor): Likewise.
19530 (minmax_loc_descriptor): Likewise.
19531 (mem_loc_descriptor): Likewise.
19532 (loc_descriptor): Likewise.
19533 * expmed.c (init_expmed_one_mode): Likewise.
19534 * lra-constraints.c (lra_constraint_offset): Likewise.
19535 * optabs.c (prepare_libcall_arg): Likewise.
19536 * postreload.c (move2add_note_store): Likewise.
19537 * reload.c (operands_match_p): Likewise.
19538 * rtl.h (load_extend_op): Likewise.
19539 * rtlhooks.c (gen_lowpart_general): Likewise.
19540 * simplify-rtx.c (simplify_truncation): Likewise.
19541 (simplify_unary_operation_1): Likewise.
19542 (simplify_binary_operation_1): Likewise.
19543 (simplify_const_binary_operation): Likewise.
19544 (simplify_const_relational_operation): Likewise.
19545 (simplify_subreg): Likewise.
19546 * stor-layout.c (bitwise_mode_for_mode): Likewise.
19547 * var-tracking.c (adjust_mems): Likewise.
19548 (prepare_call_arguments): Likewise.
19550 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19551 Alan Hayward <alan.hayward@arm.com>
19552 David Sherwood <david.sherwood@arm.com>
19554 * machmode.h (is_int_mode): New fuction.
19555 * combine.c (find_split_point): Use it.
19556 (combine_simplify_rtx): Likewise.
19557 (simplify_if_then_else): Likewise.
19558 (simplify_set): Likewise.
19559 (simplify_shift_const_1): Likewise.
19560 (simplify_comparison): Likewise.
19561 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
19562 * cse.c (notreg_cost): Likewise.
19563 (cse_insn): Likewise.
19564 * cselib.c (cselib_lookup_1): Likewise.
19565 * dojump.c (do_jump_1): Likewise.
19566 (do_compare_rtx_and_jump): Likewise.
19567 * dse.c (get_call_args): Likewise.
19568 * dwarf2out.c (rtl_for_decl_init): Likewise.
19569 (native_encode_initializer): Likewise.
19570 * expmed.c (emit_store_flag_1): Likewise.
19571 (emit_store_flag): Likewise.
19572 * expr.c (convert_modes): Likewise.
19573 (store_field): Likewise.
19574 (expand_expr_real_1): Likewise.
19575 * fold-const.c (fold_read_from_constant_string): Likewise.
19576 * gimple-ssa-sprintf.c (get_format_string): Likewise.
19577 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
19578 * optabs.c (expand_binop): Likewise.
19579 (expand_unop): Likewise.
19580 (expand_abs_nojump): Likewise.
19581 (expand_one_cmpl_abs_nojump): Likewise.
19582 * simplify-rtx.c (mode_signbit_p): Likewise.
19583 (val_signbit_p): Likewise.
19584 (val_signbit_known_set_p): Likewise.
19585 (val_signbit_known_clear_p): Likewise.
19586 (simplify_relational_operation_1): Likewise.
19587 * tree.c (vector_type_mode): Likewise.
19589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19590 Alan Hayward <alan.hayward@arm.com>
19591 David Sherwood <david.sherwood@arm.com>
19593 * machmode.h (smallest_mode_for_size): Fix formatting.
19594 (smallest_int_mode_for_size): New function.
19595 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
19596 instead of smallest_mode_for_size.
19597 * combine.c (make_extraction): Likewise.
19598 * config/arc/arc.c (arc_expand_movmem): Likewise.
19599 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
19600 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
19601 * config/s390/s390.c (s390_expand_insv): Likewise.
19602 * config/sparc/sparc.c (assign_int_registers): Likewise.
19603 * config/spu/spu.c (spu_function_value): Likewise.
19604 (spu_function_arg): Likewise.
19605 * coverage.c (get_gcov_type): Likewise.
19606 (get_gcov_unsigned_t): Likewise.
19607 * dse.c (find_shift_sequence): Likewise.
19608 * expmed.c (store_bit_field_1): Likewise.
19609 * expr.c (convert_move): Likewise.
19610 (store_field): Likewise.
19611 * internal-fn.c (expand_arith_overflow): Likewise.
19612 * optabs-query.c (get_best_extraction_insn): Likewise.
19613 * optabs.c (expand_twoval_binop_libfunc): Likewise.
19614 * stor-layout.c (layout_type): Likewise.
19615 (initialize_sizetypes): Likewise.
19616 * targhooks.c (default_get_mask_mode): Likewise.
19617 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
19619 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19620 Alan Hayward <alan.hayward@arm.com>
19621 David Sherwood <david.sherwood@arm.com>
19623 * machmode.h (opt_mode::else_blk): New function.
19624 (int_mode_for_mode): Declare.
19625 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
19626 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
19628 * cfgexpand.c (expand_debug_expr): Likewise.
19629 * combine.c (gen_lowpart_or_truncate): Likewise.
19630 (gen_lowpart_for_combine): Likewise.
19631 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
19632 * config/avr/avr.c (avr_to_int_mode): Likewise.
19633 (avr_out_plus_1): Likewise.
19634 (avr_out_plus): Likewise.
19635 (avr_out_round): Likewise.
19636 * config/i386/i386.c (ix86_split_to_parts): Likewise.
19637 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
19638 (s390_expand_vcond): Likewise.
19639 * config/spu/spu.c (spu_split_immediate): Likewise.
19640 (spu_expand_mov): Likewise.
19641 * dse.c (get_stored_val): Likewise.
19642 * expmed.c (store_bit_field_1): Likewise.
19643 (convert_extracted_bit_field): Use int_mode_for_mode instead of
19645 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
19646 (extract_low_bits): Likewise.
19647 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
19648 handling rather than repeating the check.
19649 (emit_group_store): Likewise.
19650 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
19651 * optabs.c (expand_absneg_bit): Likewise.
19652 (expand_copysign_absneg): Likewise.
19653 (expand_copysign_bit): Likewise.
19654 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
19655 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19656 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
19657 * var-tracking.c (prepare_call_arguments): Likewise.
19658 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
19659 int_mode_for_mode instead of mode_for_size.
19660 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
19662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19663 Alan Hayward <alan.hayward@arm.com>
19664 David Sherwood <david.sherwood@arm.com>
19666 * machmode.h (int_mode_for_size): New function.
19667 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
19668 instead of mode_for_size.
19669 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
19671 * combine.c (expand_field_assignment): Use int_mode_for_size
19672 instead of mode_for_size.
19673 (make_extraction): Likewise.
19674 (simplify_shift_const_1): Likewise.
19675 (simplify_comparison): Likewise.
19676 * dojump.c (do_jump): Likewise.
19677 * dwarf2out.c (mem_loc_descriptor): Likewise.
19678 * emit-rtl.c (init_derived_machine_modes): Likewise.
19679 * expmed.c (flip_storage_order): Likewise.
19680 (convert_extracted_bit_field): Likewise.
19681 * expr.c (copy_blkmode_from_reg): Likewise.
19682 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
19683 * internal-fn.c (expand_mul_overflow): Likewise.
19684 * lower-subreg.c (simple_move): Likewise.
19685 * optabs-libfuncs.c (init_optabs): Likewise.
19686 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19687 * tree.c (vector_type_mode): Likewise.
19688 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
19689 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
19690 * tree-vect-generic.c (expand_vector_parallel): Likewise.
19691 * tree-vect-stmts.c (vectorizable_load): Likewise.
19692 (vectorizable_store): Likewise.
19694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19695 Alan Hayward <alan.hayward@arm.com>
19696 David Sherwood <david.sherwood@arm.com>
19698 * coretypes.h (pod_mode): New type.
19699 (scalar_int_mode_pod): New typedef.
19700 * machmode.h (pod_mode): New class.
19701 (int_n_data_t::m): Change type to scalar_int_mode_pod.
19702 * genmodes.c (emit_mode_int_n): Update accordingly.
19703 * lower-subreg.h (target_lower_subreg): Change type to
19704 scalar_int_mode_pod.
19705 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
19706 scalar_int_mode_pod.
19708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19709 Alan Hayward <alan.hayward@arm.com>
19710 David Sherwood <david.sherwood@arm.com>
19712 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
19713 machine_mode to scalar_int_mode.
19714 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
19715 (rs6000_option_override_internal): Remove cast to int.
19716 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
19717 machine_mode to scalar_int_mode.
19718 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
19719 (rs6000_option_override_internal): Remove cast to int.
19720 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
19721 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
19723 * config/s390/s390.c (s390_expand_builtin): Likewise.
19724 * coretypes.h (scalar_int_mode): New type.
19725 (opt_scalar_int_mode): New typedef.
19726 * machmode.h (scalar_int_mode): New class.
19727 (scalar_int_mode::includes_p): New function.
19728 (byte_mode): Change type to scalar_int_mode.
19729 (word_mode): Likewise.
19730 (ptr_mode): Likewise.
19731 * emit-rtl.c (byte_mode): Likewise.
19732 (word_mode): Likewise.
19733 (ptr_mode): Likewise.
19734 (init_derived_machine_modes): Update accordingly.
19735 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
19736 and MODE_PARTIAL_INT.
19737 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
19738 opt_scalar_int_mode.
19740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19741 Alan Hayward <alan.hayward@arm.com>
19742 David Sherwood <david.sherwood@arm.com>
19744 * target.def (libgcc_floating_mode_supported_p): Take a
19746 * doc/tm.texi: Regenerate.
19747 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
19749 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19750 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
19753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19754 Alan Hayward <alan.hayward@arm.com>
19755 David Sherwood <david.sherwood@arm.com>
19757 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
19758 * doc/tm.texi: Regenerate.
19759 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
19760 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
19761 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
19762 * targhooks.h (default_floatn_mode): Likewise.
19763 * targhooks.c (default_floatn_mode): Likewise.
19764 * tree.c (build_common_tree_nodes): Update accordingly.
19766 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19767 Alan Hayward <alan.hayward@arm.com>
19768 David Sherwood <david.sherwood@arm.com>
19770 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
19771 (mode_iterator::iterate_p): Likewise.
19772 (mode_iterator::get_wider): Likewise.
19773 * expr.c (init_expr_target): Use opt_scalar_float_mode.
19775 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19776 Alan Hayward <alan.hayward@arm.com>
19777 David Sherwood <david.sherwood@arm.com>
19779 * coretypes.h (opt_scalar_float_mode): New typedef.
19780 * machmode.h (float_mode_for_size): New function.
19781 * emit-rtl.c (double_mode): Delete.
19782 (init_emit_once): Use float_mode_for_size.
19783 * stor-layout.c (layout_type): Likewise.
19784 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19787 Alan Hayward <alan.hayward@arm.com>
19788 David Sherwood <david.sherwood@arm.com>
19790 * output.h (assemble_real): Take a scalar_float_mode.
19791 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19792 * config/arm/arm.md (consttable_4): Likewise.
19793 (consttable_8): Likewise.
19794 (consttable_16): Likewise.
19795 * config/mips/mips.md (consttable_float): Likewise.
19796 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19797 * varasm.c (assemble_real): Take a scalar_float_mode.
19798 (output_constant_pool_2): Update accordingly.
19799 (output_constant): Likewise.
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 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19806 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19807 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19808 (native_encode_real): Likewise.
19809 (native_interpret_real): Likewise.
19810 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19811 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19813 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19814 Alan Hayward <alan.hayward@arm.com>
19815 David Sherwood <david.sherwood@arm.com>
19817 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19818 <scalar_float_mode>. Simplify.
19819 (gen_extend_conv_libfunc): Likewise.
19821 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19822 Alan Hayward <alan.hayward@arm.com>
19823 David Sherwood <david.sherwood@arm.com>
19825 * coretypes.h (scalar_float_mode): New type.
19826 * machmode.h (mode_traits::from_int): Use machine_mode if
19827 USE_ENUM_MODES is defined.
19828 (is_a): New function.
19830 (dyn_cast): Likewise.
19831 (scalar_float_mode): New class.
19832 (scalar_float_mode::includes_p): New function.
19833 (is_float_mode): Likewise.
19834 * gdbhooks.py (MachineModePrinter): New class.
19835 (build_pretty_printer): Use it for scalar_float_mode.
19836 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19837 (format_helper::format_helper): Turn into a template.
19838 * genmodes.c (get_mode_class): New function.
19839 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19840 or machine_mode if none.
19841 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19842 as_a <scalar_float_mode>.
19843 * dwarf2out.c (mem_loc_descriptor): Likewise.
19844 (insert_float): Likewise.
19845 (add_const_value_attribute): Likewise.
19846 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19847 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19848 (expand_unop): Update accordingly.
19849 (expand_abs_nojump): Likewise.
19850 (expand_copysign_absneg): Take a scalar_float_mode.
19851 (expand_copysign_bit): Likewise.
19852 (expand_copysign): Update accordingly.
19854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19855 Alan Hayward <alan.hayward@arm.com>
19856 David Sherwood <david.sherwood@arm.com>
19858 * coretypes.h (opt_mode): New class.
19859 * machmode.h (opt_mode): Likewise.
19860 (opt_mode::else_void): New function.
19861 (opt_mode::require): Likewise.
19862 (opt_mode::exists): Likewise.
19863 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19864 (GET_MODE_2XWIDER_MODE): Likewise.
19865 (mode_iterator::get_wider): Update accordingly.
19866 (mode_iterator::get_2xwider): Likewise.
19867 (mode_iterator::get_known_wider): Likewise, turning into a template.
19868 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19869 forcing a wider mode to exist.
19870 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19871 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19872 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19873 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19874 * lower-subreg.c (init_lower_subreg): Likewise.
19875 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19876 on the final iteration.
19877 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19878 a wider mode exists before asking for a move pattern.
19879 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19880 forcing a wider mode to exist.
19881 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19882 returning false if no such mode exists.
19883 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19884 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19885 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19886 Avoid checking for a MODE_INT if we already know the mode is not a
19888 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19889 forcing a wider mode to exist.
19890 (expmed_mult_highpart_optab): Likewise.
19891 (expmed_mult_highpart): Likewise.
19892 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19894 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19895 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19896 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19897 * internal-fn.c (expand_mul_overflow): Update use of
19898 GET_MODE_2XWIDER_MODE.
19899 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19900 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19901 GET_MODE_WIDER_MODE.
19902 (convert_plusminus_to_widen): Likewise.
19903 * tree-switch-conversion.c (array_value_type): Likewise.
19904 * var-tracking.c (emit_note_insn_var_location): Likewise.
19905 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19906 Return false inside rather than outside the loop if no wider mode
19908 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19909 and GET_MODE_2XWIDER_MODE
19910 (can_compare_p): Use else_void.
19911 * gdbhooks.py (OptMachineModePrinter): New class.
19912 (build_pretty_printer): Use it for opt_mode.
19914 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19915 Alan Hayward <alan.hayward@arm.com>
19916 David Sherwood <david.sherwood@arm.com>
19918 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19919 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19921 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19922 Alan Hayward <alan.hayward@arm.com>
19923 David Sherwood <david.sherwood@arm.com>
19925 * machmode.h (mode_traits): New structure.
19926 (get_narrowest_mode): New function.
19927 (mode_iterator::start): Likewise.
19928 (mode_iterator::iterate_p): Likewise.
19929 (mode_iterator::get_wider): Likewise.
19930 (mode_iterator::get_known_wider): Likewise.
19931 (mode_iterator::get_2xwider): Likewise.
19932 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19933 (FOR_EACH_MODE): Likewise.
19934 (FOR_EACH_MODE_FROM): Likewise.
19935 (FOR_EACH_MODE_UNTIL): Likewise.
19936 (FOR_EACH_WIDER_MODE): Likewise.
19937 (FOR_EACH_2XWIDER_MODE): Likewise.
19938 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19939 * combine.c (simplify_comparison): Likewise
19940 * config/i386/i386.c (type_natural_mode): Likewise.
19941 * cse.c (cse_insn): Likewise.
19942 * dse.c (find_shift_sequence): Likewise.
19943 * emit-rtl.c (init_derived_machine_modes): Likewise.
19944 (init_emit_once): Likewise.
19945 * explow.c (hard_function_value): Likewise.
19946 * expmed.c (extract_fixed_bit_field_1): Likewise.
19947 (extract_bit_field_1): Likewise.
19948 (expand_divmod): Likewise.
19949 (emit_store_flag_1): Likewise.
19950 * expr.c (init_expr_target): Likewise.
19951 (convert_move): Likewise.
19952 (alignment_for_piecewise_move): Likewise.
19953 (widest_int_mode_for_size): Likewise.
19954 (emit_block_move_via_movmem): Likewise.
19955 (copy_blkmode_to_reg): Likewise.
19956 (set_storage_via_setmem): Likewise.
19957 (compress_float_constant): Likewise.
19958 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19959 * optabs-query.c (get_best_extraction_insn): Likewise.
19960 * optabs.c (expand_binop): Likewise.
19961 (expand_twoval_unop): Likewise.
19962 (expand_twoval_binop): Likewise.
19963 (widen_leading): Likewise.
19964 (widen_bswap): Likewise.
19965 (expand_parity): Likewise.
19966 (expand_unop): Likewise.
19967 (prepare_cmp_insn): Likewise.
19968 (prepare_float_lib_cmp): Likewise.
19969 (expand_float): Likewise.
19970 (expand_fix): Likewise.
19971 (expand_sfix_optab): Likewise.
19972 * postreload.c (move2add_use_add2_insn): Likewise.
19973 * reg-stack.c (reg_to_stack): Likewise.
19974 * reginfo.c (choose_hard_reg_mode): Likewise.
19975 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19976 * stor-layout.c (mode_for_size): Likewise.
19977 (smallest_mode_for_size): Likewise.
19978 (mode_for_vector): Likewise.
19979 (finish_bitfield_representative): Likewise.
19980 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19981 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19982 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19983 * var-tracking.c (prepare_call_arguments): Likewise.
19985 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19986 Alan Hayward <alan.hayward@arm.com>
19987 David Sherwood <david.sherwood@arm.com>
19989 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19990 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19991 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19992 * machmode.h (mode_size): Move earlier in file.
19993 (mode_precision): Likewise.
19994 (mode_inner): Likewise.
19995 (mode_nunits): Likewise.
19996 (mode_unit_size): Likewise.
19997 (unit_unit_precision): Likewise.
19998 (mode_wider): Likewise.
19999 (mode_2xwider): Likewise.
20000 (machine_mode): New class.
20001 (mode_to_bytes): New function.
20002 (mode_to_bits): Likewise.
20003 (mode_to_precision): Likewise.
20004 (mode_to_inner): Likewise.
20005 (mode_to_unit_size): Likewise.
20006 (mode_to_unit_precision): Likewise.
20007 (mode_to_nunits): Likewise.
20008 (GET_MODE_SIZE): Use mode_to_bytes.
20009 (GET_MODE_BITSIZE): Use mode_to_bits.
20010 (GET_MODE_PRECISION): Use mode_to_precision.
20011 (GET_MODE_INNER): Use mode_to_inner.
20012 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
20013 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
20014 (GET_MODE_NUNITS): Use mode_to_nunits.
20015 * system.h (ALWAYS_INLINE): New macro.
20016 * config/powerpcspe/powerpcspe-c.c
20017 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
20018 int for arg1_mode and arg2_mode.
20020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20021 Alan Hayward <alan.hayward@arm.com>
20022 David Sherwood <david.sherwood@arm.com>
20024 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
20025 Prefix mode names with E_ in case statements.
20026 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20027 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
20028 (aarch64_split_simd_move): Likewise.
20029 (aarch64_gen_storewb_pair): Likewise.
20030 (aarch64_gen_loadwb_pair): Likewise.
20031 (aarch64_gen_store_pair): Likewise.
20032 (aarch64_gen_load_pair): Likewise.
20033 (aarch64_get_condition_code_1): Likewise.
20034 (aarch64_constant_pool_reload_icode): Likewise.
20035 (get_rsqrte_type): Likewise.
20036 (get_rsqrts_type): Likewise.
20037 (get_recpe_type): Likewise.
20038 (get_recps_type): Likewise.
20039 (aarch64_gimplify_va_arg_expr): Likewise.
20040 (aarch64_simd_container_mode): Likewise.
20041 (aarch64_emit_load_exclusive): Likewise.
20042 (aarch64_emit_store_exclusive): Likewise.
20043 (aarch64_expand_compare_and_swap): Likewise.
20044 (aarch64_gen_atomic_cas): Likewise.
20045 (aarch64_emit_bic): Likewise.
20046 (aarch64_emit_atomic_swap): Likewise.
20047 (aarch64_emit_atomic_load_op): Likewise.
20048 (aarch64_evpc_trn): Likewise.
20049 (aarch64_evpc_uzp): Likewise.
20050 (aarch64_evpc_zip): Likewise.
20051 (aarch64_evpc_ext): Likewise.
20052 (aarch64_evpc_rev): Likewise.
20053 (aarch64_evpc_dup): Likewise.
20054 (aarch64_gen_ccmp_first): Likewise.
20055 (aarch64_gen_ccmp_next): Likewise.
20056 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
20057 (alpha_emit_xfloating_libcall): Likewise.
20058 (emit_insxl): Likewise.
20059 (alpha_arg_type): Likewise.
20060 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
20061 (arc_preferred_simd_mode): Likewise.
20062 (arc_secondary_reload): Likewise.
20063 (get_arc_condition_code): Likewise.
20064 (arc_print_operand): Likewise.
20065 (arc_legitimate_constant_p): Likewise.
20066 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20067 * config/arc/arc.md (casesi_load): Likewise.
20068 (casesi_compact_jump): Likewise.
20069 * config/arc/predicates.md (proper_comparison_operator): Likewise.
20070 (cc_use_register): Likewise.
20071 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20072 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
20073 (arm_init_iwmmxt_builtins): Likewise.
20074 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
20075 (neon_expand_vector_init): Likewise.
20076 (arm_attr_length_move_neon): Likewise.
20077 (maybe_get_arm_condition_code): Likewise.
20078 (arm_emit_vector_const): Likewise.
20079 (arm_preferred_simd_mode): Likewise.
20080 (arm_output_iwmmxt_tinsr): Likewise.
20081 (thumb1_output_casesi): Likewise.
20082 (thumb2_output_casesi): Likewise.
20083 (arm_emit_load_exclusive): Likewise.
20084 (arm_emit_store_exclusive): Likewise.
20085 (arm_expand_compare_and_swap): Likewise.
20086 (arm_evpc_neon_vuzp): Likewise.
20087 (arm_evpc_neon_vzip): Likewise.
20088 (arm_evpc_neon_vrev): Likewise.
20089 (arm_evpc_neon_vtrn): Likewise.
20090 (arm_evpc_neon_vext): Likewise.
20091 (arm_validize_comparison): Likewise.
20092 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
20093 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
20094 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
20095 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
20096 (c6x_preferred_simd_mode): Likewise.
20097 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
20098 (epiphany_rtx_costs): Likewise.
20099 * config/epiphany/predicates.md (proper_comparison_operator):
20101 * config/frv/frv.c (condexec_memory_operand): Likewise.
20102 (frv_emit_move): Likewise.
20103 (output_move_single): Likewise.
20104 (output_condmove_single): Likewise.
20105 (frv_hard_regno_mode_ok): Likewise.
20106 (frv_matching_accg_mode): Likewise.
20107 * config/h8300/h8300.c (split_adds_subs): Likewise.
20108 (h8300_rtx_costs): Likewise.
20109 (h8300_print_operand): Likewise.
20110 (compute_mov_length): Likewise.
20111 (output_logical_op): Likewise.
20112 (compute_logical_op_length): Likewise.
20113 (compute_logical_op_cc): Likewise.
20114 (h8300_shift_needs_scratch_p): Likewise.
20115 (output_a_shift): Likewise.
20116 (compute_a_shift_length): Likewise.
20117 (compute_a_shift_cc): Likewise.
20118 (expand_a_rotate): Likewise.
20119 (output_a_rotate): Likewise.
20120 * config/i386/i386.c (classify_argument): Likewise.
20121 (function_arg_advance_32): Likewise.
20122 (function_arg_32): Likewise.
20123 (function_arg_64): Likewise.
20124 (function_value_64): Likewise.
20125 (ix86_gimplify_va_arg): Likewise.
20126 (ix86_legitimate_constant_p): Likewise.
20127 (put_condition_code): Likewise.
20128 (split_double_mode): Likewise.
20129 (ix86_avx256_split_vector_move_misalign): Likewise.
20130 (ix86_expand_vector_logical_operator): Likewise.
20131 (ix86_split_idivmod): Likewise.
20132 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
20133 (ix86_build_const_vector): Likewise.
20134 (ix86_build_signbit_mask): Likewise.
20135 (ix86_match_ccmode): Likewise.
20136 (ix86_cc_modes_compatible): Likewise.
20137 (ix86_expand_branch): Likewise.
20138 (ix86_expand_sse_cmp): Likewise.
20139 (ix86_expand_sse_movcc): Likewise.
20140 (ix86_expand_int_sse_cmp): Likewise.
20141 (ix86_expand_vec_perm_vpermi2): Likewise.
20142 (ix86_expand_vec_perm): Likewise.
20143 (ix86_expand_sse_unpack): Likewise.
20144 (ix86_expand_int_addcc): Likewise.
20145 (ix86_split_to_parts): Likewise.
20146 (ix86_vectorize_builtin_gather): Likewise.
20147 (ix86_vectorize_builtin_scatter): Likewise.
20148 (avx_vpermilp_parallel): Likewise.
20149 (inline_memory_move_cost): Likewise.
20150 (ix86_tieable_integer_mode_p): Likewise.
20151 (x86_maybe_negate_const_int): Likewise.
20152 (ix86_expand_vector_init_duplicate): Likewise.
20153 (ix86_expand_vector_init_one_nonzero): Likewise.
20154 (ix86_expand_vector_init_one_var): Likewise.
20155 (ix86_expand_vector_init_concat): Likewise.
20156 (ix86_expand_vector_init_interleave): Likewise.
20157 (ix86_expand_vector_init_general): Likewise.
20158 (ix86_expand_vector_set): Likewise.
20159 (ix86_expand_vector_extract): Likewise.
20160 (emit_reduc_half): Likewise.
20161 (ix86_emit_i387_round): Likewise.
20162 (ix86_mangle_type): Likewise.
20163 (ix86_expand_round_sse4): Likewise.
20164 (expand_vec_perm_blend): Likewise.
20165 (canonicalize_vector_int_perm): Likewise.
20166 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
20167 (expand_vec_perm_1): Likewise.
20168 (expand_vec_perm_interleave3): Likewise.
20169 (expand_vec_perm_even_odd_pack): Likewise.
20170 (expand_vec_perm_even_odd_1): Likewise.
20171 (expand_vec_perm_broadcast_1): Likewise.
20172 (ix86_vectorize_vec_perm_const_ok): Likewise.
20173 (ix86_expand_vecop_qihi): Likewise.
20174 (ix86_expand_mul_widen_hilo): Likewise.
20175 (ix86_expand_sse2_abs): Likewise.
20176 (ix86_expand_pextr): Likewise.
20177 (ix86_expand_pinsr): Likewise.
20178 (ix86_preferred_simd_mode): Likewise.
20179 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
20180 * config/i386/sse.md (*andnot<mode>3): Likewise.
20181 (<mask_codefor><code><mode>3<mask_name>): Likewise.
20182 (*<code><mode>3): Likewise.
20183 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
20184 (ia64_expand_atomic_op): Likewise.
20185 (ia64_arg_type): Likewise.
20186 (ia64_mode_to_int): Likewise.
20187 (ia64_scalar_mode_supported_p): Likewise.
20188 (ia64_vector_mode_supported_p): Likewise.
20189 (expand_vec_perm_broadcast): Likewise.
20190 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
20191 (iq2000_function_arg_advance): Likewise.
20192 (iq2000_function_arg): Likewise.
20193 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
20194 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
20195 (m68k_libcall_value): Likewise.
20196 (m68k_function_value): Likewise.
20197 (sched_attr_op_type): Likewise.
20198 * config/mcore/mcore.c (mcore_output_move): Likewise.
20199 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
20201 (microblaze_function_arg): Likewise.
20202 * config/mips/mips.c (mips16_build_call_stub): Likewise.
20203 (mips_print_operand): Likewise.
20204 (mips_mode_ok_for_mov_fmt_p): Likewise.
20205 (mips_vector_mode_supported_p): Likewise.
20206 (mips_preferred_simd_mode): Likewise.
20207 (mips_expand_vpc_loongson_even_odd): Likewise.
20208 (mips_expand_vec_unpack): Likewise.
20209 (mips_expand_vi_broadcast): Likewise.
20210 (mips_expand_vector_init): Likewise.
20211 (mips_expand_vec_reduc): Likewise.
20212 (mips_expand_msa_cmp): Likewise.
20213 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
20214 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
20215 (cc_flags_for_mode): Likewise.
20216 * config/msp430/msp430.c (msp430_print_operand): Likewise.
20217 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
20218 (nds32_output_casesi_pc_relative): Likewise.
20219 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20220 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
20221 (nvptx_gen_unpack): Likewise.
20222 (nvptx_gen_pack): Likewise.
20223 (nvptx_gen_shuffle): Likewise.
20224 (nvptx_gen_wcast): Likewise.
20225 (nvptx_preferred_simd_mode): Likewise.
20226 * config/pa/pa.c (pa_secondary_reload): Likewise.
20227 * config/pa/predicates.md (base14_operand): Likewise.
20228 * config/powerpcspe/powerpcspe-c.c
20229 (altivec_resolve_overloaded_builtin): Likewise.
20230 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
20232 (rs6000_preferred_simd_mode): Likewise.
20233 (output_vec_const_move): Likewise.
20234 (rs6000_expand_vector_extract): Likewise.
20235 (rs6000_split_vec_extract_var): Likewise.
20236 (reg_offset_addressing_ok_p): Likewise.
20237 (rs6000_legitimate_offset_address_p): Likewise.
20238 (rs6000_legitimize_address): Likewise.
20239 (rs6000_emit_set_const): Likewise.
20240 (rs6000_const_vec): Likewise.
20241 (rs6000_emit_move): Likewise.
20242 (spe_build_register_parallel): Likewise.
20243 (rs6000_darwin64_record_arg_recurse): Likewise.
20244 (swap_selector_for_mode): Likewise.
20245 (spe_init_builtins): Likewise.
20246 (paired_init_builtins): Likewise.
20247 (altivec_init_builtins): Likewise.
20248 (do_load_for_compare): Likewise.
20249 (rs6000_generate_compare): Likewise.
20250 (rs6000_expand_float128_convert): Likewise.
20251 (emit_load_locked): Likewise.
20252 (emit_store_conditional): Likewise.
20253 (rs6000_output_function_epilogue): Likewise.
20254 (rs6000_handle_altivec_attribute): Likewise.
20255 (rs6000_function_value): Likewise.
20256 (emit_fusion_gpr_load): Likewise.
20257 (emit_fusion_p9_load): Likewise.
20258 (emit_fusion_p9_store): Likewise.
20259 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
20260 (fusion_gpr_mem_load): Likewise.
20261 (fusion_addis_mem_combo_load): Likewise.
20262 (fusion_addis_mem_combo_store): Likewise.
20263 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
20264 (fusion_gpr_mem_load): Likewise.
20265 (fusion_addis_mem_combo_load): Likewise.
20266 (fusion_addis_mem_combo_store): Likewise.
20267 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20269 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
20270 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20271 (rs6000_preferred_simd_mode): Likewise.
20272 (output_vec_const_move): Likewise.
20273 (rs6000_expand_vector_extract): Likewise.
20274 (rs6000_split_vec_extract_var): Likewise.
20275 (reg_offset_addressing_ok_p): Likewise.
20276 (rs6000_legitimate_offset_address_p): Likewise.
20277 (rs6000_legitimize_address): Likewise.
20278 (rs6000_emit_set_const): Likewise.
20279 (rs6000_const_vec): Likewise.
20280 (rs6000_emit_move): Likewise.
20281 (rs6000_darwin64_record_arg_recurse): Likewise.
20282 (swap_selector_for_mode): Likewise.
20283 (paired_init_builtins): Likewise.
20284 (altivec_init_builtins): Likewise.
20285 (rs6000_expand_float128_convert): Likewise.
20286 (emit_load_locked): Likewise.
20287 (emit_store_conditional): Likewise.
20288 (rs6000_output_function_epilogue): Likewise.
20289 (rs6000_handle_altivec_attribute): Likewise.
20290 (rs6000_function_value): Likewise.
20291 (emit_fusion_gpr_load): Likewise.
20292 (emit_fusion_p9_load): Likewise.
20293 (emit_fusion_p9_store): Likewise.
20294 * config/rx/rx.c (rx_gen_move_template): Likewise.
20295 (flags_from_mode): Likewise.
20296 * config/s390/predicates.md (s390_alc_comparison): Likewise.
20297 (s390_slb_comparison): Likewise.
20298 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
20299 (s390_vector_mode_supported_p): Likewise.
20300 (s390_cc_modes_compatible): Likewise.
20301 (s390_match_ccmode_set): Likewise.
20302 (s390_canonicalize_comparison): Likewise.
20303 (s390_emit_compare_and_swap): Likewise.
20304 (s390_branch_condition_mask): Likewise.
20305 (s390_rtx_costs): Likewise.
20306 (s390_secondary_reload): Likewise.
20307 (__SECONDARY_RELOAD_CASE): Likewise.
20308 (s390_expand_cs): Likewise.
20309 (s390_preferred_simd_mode): Likewise.
20310 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
20311 * config/sh/sh.c (sh_print_operand): Likewise.
20312 (dump_table): Likewise.
20313 (sh_secondary_reload): Likewise.
20314 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20315 * config/sh/sh.md (casesi_worker_1): Likewise.
20316 (casesi_worker_2): Likewise.
20317 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
20318 (fcc_comparison_operator): Likewise.
20319 * config/sparc/sparc.c (sparc_expand_move): Likewise.
20320 (emit_soft_tfmode_cvt): Likewise.
20321 (sparc_preferred_simd_mode): Likewise.
20322 (output_cbranch): Likewise.
20323 (sparc_print_operand): Likewise.
20324 (sparc_expand_vec_perm_bmask): Likewise.
20325 (vector_init_bshuffle): Likewise.
20326 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
20327 (spu_vector_mode_supported_p): Likewise.
20328 (spu_expand_insv): Likewise.
20329 (spu_emit_branch_or_set): Likewise.
20330 (spu_handle_vector_attribute): Likewise.
20331 (spu_builtin_splats): Likewise.
20332 (spu_builtin_extract): Likewise.
20333 (spu_builtin_promote): Likewise.
20334 (spu_expand_sign_extend): Likewise.
20335 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
20336 (tilegx_simd_int): Likewise.
20337 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
20338 (tilepro_simd_int): Likewise.
20339 * config/v850/v850.c (const_double_split): Likewise.
20340 (v850_print_operand): Likewise.
20341 (ep_memory_offset): Likewise.
20342 * config/vax/vax.c (vax_rtx_costs): Likewise.
20343 (vax_output_int_move): Likewise.
20344 (vax_output_int_add): Likewise.
20345 (vax_output_int_subtract): Likewise.
20346 * config/visium/predicates.md (visium_branch_operator): Likewise.
20347 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
20348 (visium_print_operand_address): Likewise.
20349 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20350 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
20351 (xtensa_expand_conditional_branch): Likewise.
20352 (xtensa_copy_incoming_a7): Likewise.
20353 (xtensa_output_literal): Likewise.
20354 * dfp.c (decimal_real_maxval): Likewise.
20355 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
20357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20358 Alan Hayward <alan.hayward@arm.com>
20359 David Sherwood <david.sherwood@arm.com>
20361 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
20362 (mode_nunits_inline): Likewise.
20363 (mode_inner_inline): Likewise.
20364 (mode_unit_size_inline): Likewise.
20365 (mode_unit_precision_inline): Likewise.
20366 (emit_insn_modes_h): Likewise. Also emit a #define of the
20368 (emit_mode_wider): Add an E_ prefix to mode names.
20369 (emit_mode_complex): Likewise.
20370 (emit_mode_inner): Likewise.
20371 (emit_mode_adjustments): Likewise.
20372 (emit_mode_int_n): Likewise.
20373 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
20374 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
20375 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
20376 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
20377 (CRC32_BUILTIN, ENTRY): Likewise.
20378 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
20379 (aarch64_pop_regs): Likewise.
20380 (aarch64_process_components): Likewise.
20381 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
20382 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
20383 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
20384 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
20385 * config/arm/arm.c (arm_init_libfuncs): Likewise.
20386 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
20388 * config/i386/i386-builtin.def (pcmpestr): Likewise.
20389 (pcmpistr): Likewise.
20390 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
20391 * config/mmix/mmix.c (mmix_output_condition): Likewise.
20392 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20394 * config/rl78/rl78.c (mduc_regs): Likewise.
20395 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20396 (htm_expand_builtin): Likewise.
20397 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
20398 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
20399 * config/xtensa/xtensa.c (print_operand): Likewise.
20400 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
20401 (NUM_MODE_VECTOR_INT): Likewise.
20402 * genoutput.c (null_operand): Likewise.
20403 (output_operand_data): Likewise.
20404 * genrecog.c (print_parameter_value): Likewise.
20405 * lra.c (debug_operand_data): Likewise.
20407 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20409 * dwarf2out.c (scompare_loc_descriptor_wide)
20410 (scompare_loc_descriptor_narrow): New functions, split out from...
20411 (scompare_loc_descriptor): ...here.
20412 * expmed.c (emit_store_flag_int): New function, split out from...
20413 (emit_store_flag): ...here.
20415 2017-08-30 Richard Biener <rguenther@suse.de>
20417 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
20418 (dwarf2out_early_finish): Move setting of AT_pubnames from
20419 early debug output to early finish.
20421 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20423 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
20424 and -mdata-region to the assembler.
20426 2017-08-30 Richard Biener <rguenther@suse.de>
20428 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
20430 (gen_subprogram_die): Add DW_AT_object_pointer only early.
20431 (dwarf2out_early_global_decl): Only generate a DIE for the
20432 abstract origin if it doesn't already exist or is a declaration DIE.
20433 (resolve_addr): Do not add the linkage name twice when
20434 generating a stub DIE for the DW_TAG_GNU_call_site target.
20436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20438 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20439 Use machine_mode rather than int for arg1_mode.
20441 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20444 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
20445 that the second argument of the built-in functions to unpack
20446 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
20447 switch statement instead a lot of if statements.
20448 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
20449 Allow 64-bit values to be in Altivec registers as well as
20450 traditional floating point registers.
20451 (pack<mode>, FMOVE128_VSX iterator): Likewise.
20453 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
20455 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
20456 MAX_REGS_PER_ADDRESS == 1.
20458 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
20460 * config/i386/i386.opt (flag_fentry): Do not init to -1.
20461 * config/i386/i386.c (ix86_option_override_internal): Simplify
20462 setting of opts->x_flag_entry.
20464 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20465 Jakub Jelinek <jakub@redhat.com>
20466 Richard Biener <rguenther@suse.de>
20468 PR tree-optimization/81503
20469 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
20470 folded constant fits in the target type; reorder tests for clarity.
20472 2017-08-29 Martin Liska <mliska@suse.cz>
20474 * passes.def: Include pass_lower_switch.
20475 * stmt.c (dump_case_nodes): Remove and move to
20476 tree-switch-conversion.
20477 (case_values_threshold): Likewise.
20478 (expand_switch_as_decision_tree_p): Likewise.
20479 (emit_case_decision_tree): Likewise.
20480 (expand_case): Likewise.
20481 (balance_case_nodes): Likewise.
20482 (node_has_low_bound): Likewise.
20483 (node_has_high_bound): Likewise.
20484 (node_is_bounded): Likewise.
20485 (emit_case_nodes): Likewise.
20486 (struct simple_case_node): New struct.
20487 (add_case_node): Remove.
20488 (emit_case_dispatch_table): Use vector instead of case_list.
20489 (reset_out_edges_aux): Remove.
20490 (compute_cases_per_edge): Likewise.
20491 (expand_case): Build list of simple_case_node.
20492 (expand_sjlj_dispatch_table): Use it.
20493 * tree-switch-conversion.c (struct case_node): Moved from
20494 stmt.c and adjusted.
20495 (emit_case_nodes): Likewise.
20496 (node_has_low_bound): Likewise.
20497 (node_has_high_bound): Likewise.
20498 (node_is_bounded): Likewise.
20499 (case_values_threshold): Likewise.
20500 (reset_out_edges_aux): Likewise.
20501 (compute_cases_per_edge): Likewise.
20502 (add_case_node): Likewise.
20503 (dump_case_nodes): Likewise.
20504 (balance_case_nodes): Likewise.
20505 (expand_switch_as_decision_tree_p): Likewise.
20506 (emit_jump): Likewise.
20507 (emit_case_decision_tree): Likewise.
20508 (try_switch_expansion): Likewise.
20509 (do_jump_if_equal): Likewise.
20510 (emit_cmp_and_jump_insns): Likewise.
20511 (fix_phi_operands_for_edge): New function.
20512 (record_phi_operand_mapping): Likewise.
20513 (class pass_lower_switch): New pass.
20514 (pass_lower_switch::execute): New function.
20515 (make_pass_lower_switch): Likewise.
20516 (conditional_probability):
20517 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
20518 * tree-pass.h: Add make_pass_lower_switch.
20520 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20523 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
20526 2017-08-29 Richard Biener <rguenther@suse.de>
20528 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
20529 we do not add a DW_AT_inline attribute twice.
20530 (gen_subprogram_die): Remove code setting DW_AT_inline on
20531 DECL_ABSTRACT_P nodes.
20533 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
20535 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
20536 calls to internal functions.
20537 (gimplify_modify_expr): Likewise.
20538 * tree-call-cdce.c (use_internal_fn): Likewise.
20539 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
20540 (convert_to_divmod): Set the nothrow flag.
20541 * tree-if-conv.c (predicate_mem_writes): Likewise.
20542 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20543 (vectorizable_call): Likewise.
20544 (vectorizable_store): Likewise.
20545 (vectorizable_load): Likewise.
20546 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
20547 (vect_recog_mask_conversion_pattern): Likewise.
20549 2017-08-29 Martin Liska <mliska@suse.cz>
20552 * gcc.c (driver_handle_option): Add new argument.
20553 * opts-common.c (handle_option): Pass
20554 target_option_override_hook.
20555 * opts-global.c (lang_handle_option): Add new option.
20556 (set_default_handlers): Add new argument.
20557 (decode_options): Likewise.
20558 * opts.c (target_handle_option): Likewise.
20559 (common_handle_option): Call target_option_override_hook.
20560 * opts.h (struct cl_option_handler_func): Add hook for
20561 target option override.
20562 (struct cl_option_handlers): Likewise.
20563 (set_default_handlers): Add new argument.
20564 (decode_options): Likewise.
20565 (common_handle_option): Likewise.
20566 (target_handle_option): Likewise.
20567 * toplev.c (toplev::main): Pass targetm.target_option.override
20570 2017-08-29 Richard Biener <rguenther@suse.de>
20571 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
20573 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
20574 life to the active subtree.
20576 2017-08-28 Jeff Law <law@redhat.com>
20578 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
20579 derive_equivalences.
20580 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
20581 Code moved into....
20582 (edge_info::derive_equivalences): New private member function
20584 * tree-ssa-dom.c (class edge_info): Changed from a struct
20585 to a class. Add ctor/dtor, methods and data members.
20586 (edge_info::edge_info): Renamed from allocate_edge_info.
20587 Initialize additional members.
20588 (edge_info::~edge_info): New.
20589 (free_dom_edge_info): Delete the edge info.
20590 (record_edge_info): Use new class & associated member functions.
20591 Tighten forms for testing for edge equivalences.
20592 (record_temporary_equivalences): Iterate over the simple
20593 equivalences rather than assuming there's only one per edge.
20594 (cprop_into_successor_phis): Iterate over the simple
20595 equivalences rather than assuming there's only one per edge.
20596 (optimize_stmt): Use operand_equal_p rather than pointer
20597 equality for mini-DSE code.
20599 2017-08-28 Nathan Sidwell <nathan@acm.org>
20601 * gcc.c (execute): Fold SIGPIPE handling into switch
20602 statement. Adjust internal error message.
20604 2017-08-28 Richard Biener <rguenther@suse.de>
20607 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
20608 Do nothing for removed DIEs.
20610 2017-08-28 Richard Biener <rguenther@suse.de>
20612 PR tree-optimization/81977
20613 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
20616 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
20619 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
20620 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
20621 using targetm.gen_mem_thread_fence.
20623 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
20626 * config/i386/i386.md (*<btsc><mode>): Change operand 2
20627 predicate to register_operand. Reorder operands.
20628 (*btr<mode>): Ditto.
20629 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
20630 (*btr<mode>_mask): Ditto.
20632 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
20634 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
20635 * config/rs6000/xmmintrin.h: New file.
20636 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
20638 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20641 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
20642 parameter and_insn and return it.
20643 (recombine_lvx_pattern): Insert a copy to ensure availability of
20644 the base register of the copied masking operation at the point of
20645 the instruction replacement.
20646 (recombine_stvx_pattern): Likewise.
20648 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20650 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
20651 undocumented switches.
20652 (-mpower9-dform-vector): Likewise.
20653 (-mpower9-dform): Likewise.
20654 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
20655 comments to delete references to -mpower9-dform* switches.
20656 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
20657 Delete reference to -mpower9-dform* switches, test for
20658 -mpower9-vector instead.
20659 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
20660 (OTHER_P9_VECTOR_MASKS): Likewise.
20661 (POWERPC_MASKS): Likewise.
20662 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
20663 tests against -mpower9-dform* to -mpower9-vector. Delete code
20664 that checked for -mpower9-dform* consistancy with other options.
20665 Add test for -mpower9-misc to enable other power9 switches.
20666 (rs6000_init_hard_regno_mode_ok): Likewise.
20667 (rs6000_option_override_internal): Likewise.
20668 (rs6000_emit_prologue): Likewise.
20669 (rs6000_emit_epilogue): Likewise.
20670 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
20671 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
20672 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
20674 (emit_fusion_p9_store): Likewise.
20675 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
20676 resetting these macros if the assembler does not support ISA 3.0
20678 (TARGET_P9_DFORM_VECTOR): Likewise.
20679 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
20680 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
20682 2017-08-25 Alan Modra <amodra@gmail.com>
20684 PR rtl-optimization/81747
20685 * cse.c (cse_extended_basic_block): Don't attempt to record
20686 equivalences for degenerate conditional branches that branch
20687 to their fall-through.
20689 2017-08-24 Martin Sebor <msebor@redhat.com>
20691 PR middle-end/81908
20692 * gimple-fold.c (size_must_be_zero_p): New function.
20693 (gimple_fold_builtin_memory_op): Call it.
20695 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
20697 * config/rs6000/mm_malloc.h: New file.
20699 2017-08-24 Bin Cheng <bin.cheng@arm.com>
20701 PR tree-optimization/81913
20702 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
20703 analysis when either IVs in condition can wrap.
20705 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
20707 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
20708 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
20710 2017-08-24 Richard Biener <rguenther@suse.de>
20713 * targhooks.c (default_target_can_inline_p): Properly
20714 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
20715 is present and always compare.
20716 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
20717 infer -mfpmath=sse from TARGET_SSE_P.
20718 (ix86_can_inline_p): Properly use target_option_default_node when
20719 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
20721 2017-08-24 Richard Biener <rguenther@suse.de>
20724 * dwarf2out.c (output_die): Handle flag_generate_offload like
20726 (output_comp_unit): Likewise.
20727 (gen_array_type_die): Likewise.
20728 (dwarf2out_early_finish): Likewise.
20729 (note_variable_value_in_expr): Likewise.
20730 (dwarf2out_finish): Likewise. Adjust assert.
20731 * cgraphunit.c (symbol_table::compile): Move setting of
20732 flag_generate_offload earlier ...
20733 (symbol_table::finalize_compilation_unit): ... here, before
20734 early debug finalization.
20736 2017-08-24 Richard Biener <rguenther@suse.de>
20738 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
20739 and ipa-fnsummary.h.
20740 (ix86_can_inline_p): When ix86_fpmath flags do not match
20741 check whether the callee uses FP math at all.
20743 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
20745 PR middle-end/81931
20746 * tree-ssanames.c (get_nonzero_bits): Use element_precision
20747 instead of TYPE_PRECISION.
20749 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
20750 Alan Hayward <alan.hayward@arm.com>
20751 David Sherwood <david.sherwood@arm.com>
20753 * combine.c (make_extraction): Use subreg_offset_from_lsb.
20755 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
20757 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
20759 (ix86_frame::stack_realign_allocate): New field.
20760 (struct machine_frame_state): Modify comments.
20761 (machine_frame_state::sp_realigned_fp_end): New field.
20762 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
20763 layout calculation.
20764 (sp_valid_at): Add assertion to assure no attempt to access invalid
20765 offset of a realigned stack.
20766 (fp_valid_at): Likewise.
20767 (choose_baseaddr): Modify comments.
20768 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
20769 ix86_expand_prologue.
20770 (ix86_expand_prologue): Modify stack realignment and allocation.
20771 (ix86_expand_epilogue): Modify comments.
20772 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
20773 avx2_runtime, avx512f, and avx512f_runtime.
20775 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20777 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20778 (mstackrealign): Do not init to -1.
20779 * config/i386/i386.c (ix86_option_override_internal):
20780 Check opts_set, not opts when setting default value of
20781 opts->x_ix86_force_align_arg_pointer.
20783 2017-08-23 Richard Biener <rguenther@suse.de>
20785 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20786 lang_hooks.decl_printable_name.
20787 * print-rtl-function.c (print_rtx_function): Likewise.
20788 * tree-pretty-print.c (dump_function_header): Likewise.
20790 2017-08-23 Richard Biener <rguenther@suse.de>
20793 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20794 -g0 at compile-time.
20796 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20798 PR middle-end/19706
20799 * doc/sourcebuild.texi (Other hardware attributes):
20802 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20804 PR middle-end/19706
20805 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20806 Fix single-use check.
20808 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20810 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20812 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20814 * doc/install.texi: Modify to add more details on running selected
20817 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20819 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20820 is combined with -mabi=ms.
20821 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20824 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20826 PR tree-optimization/81488
20827 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20828 and cached_basis fields.
20829 (MAX_SPREAD): New constant.
20830 (alloc_cand_and_find_basis): Initialize new fields.
20831 (clear_visited): New function.
20832 (create_phi_basis_1): Rename from create_phi_basis, set visited
20833 and cached_basis fields.
20834 (create_phi_basis): New wrapper function.
20835 (phi_add_costs_1): Rename from phi_add_costs, add spread
20836 parameter, set visited field, short-circuit when limits reached.
20837 (phi_add_costs): New wrapper function.
20838 (record_phi_increments_1): Rename from record_phi_increments, set
20840 (record_phi_increments): New wrapper function.
20841 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20842 (phi_incr_cost): New wrapper function.
20843 (all_phi_incrs_profitable_1): Rename from
20844 all_phi_incrs_profitable, set visited field.
20845 (all_phi_incrs_profitable): New wrapper function.
20847 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20848 Alan Hayward <alan.hayward@arm.com>
20849 David Sherwood <david.sherwood@arm.com>
20851 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20852 that takes the outer and inner modes.
20853 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20854 comparison as the canonical test for a paradoxical subreg.
20855 * combine.c (simplify_set): Use paradoxical_subreg_p.
20856 (make_extraction): Likewise.
20857 (force_to_mode): Likewise.
20858 (rtx_equal_for_field_assignment_p): Likewise.
20859 (gen_lowpart_for_combine): Likewise.
20860 (simplify_comparison): Likewise.
20861 * cse.c (equiv_constant): Likewise.
20862 * expmed.c (store_bit_field_1): Likewise.
20863 * final.c (alter_subreg): Likewise.
20864 * fwprop.c (propagate_rtx): Likewise.
20865 (forward_propagate_subreg): Likewise.
20866 * ira-conflicts.c (ira_build_conflicts): Likewise.
20867 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20868 * lra-constraints.c (curr_insn_transform): Likewise.
20869 (split_reg): Likewise.
20870 * lra-eliminations.c (move_plus_up): Likewise.
20871 (lra_eliminate_regs_1): Likewise.
20872 * recog.c (general_operand): Likewise.
20873 * ree.c (combine_reaching_defs): Likewise.
20874 * reload.c (push_reload): Likewise.
20875 (find_reloads): Likewise.
20876 * reload1.c (elimination_effects): Likewise.
20877 (compute_reload_subreg_offset): Likewise.
20878 (choose_reload_regs): Likewise.
20879 * rtlanal.c (subreg_lsb_1): Likewise.
20880 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20881 (simplify_subreg): Likewise.
20882 * var-tracking.c (track_loc_p): Likewise.
20883 * emit-rtl.c (byte_lowpart_offset): Likewise.
20884 (paradoxical_subreg_p): Delete out-of-line definition.
20886 2017-08-22 Jeff Law <law@redhat.com>
20888 PR tree-optimization/81741
20889 PR tree-optimization/71947
20890 * tree-ssa-dom.c: Include tree-inline.h.
20891 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20892 equivalences if one is more expensive to compute than the other.
20893 * tree-ssa-scopedtables.h (class const_or_copies): Make
20894 record_const_or_copy_raw method private.
20895 (class avail_exprs_stack): New method simplify_binary_operation.
20896 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20897 avail_exprs_stack::simplify_binary_operation as needed.
20898 (avail_exprs_stack::simplify_binary_operation): New function.
20900 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20902 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20903 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20904 (DOT_SYMBOLS): Likewise.
20905 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20906 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20907 (RS6000_ABI_NAME): Likewise.
20908 (TARGET_CMODEL): Likewise.
20909 (TOC_SECTION_ASM_OP): Likewise.
20910 (SET_CMODEL): New macro.
20911 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20913 2017-08-22 Richard Biener <rguenther@suse.de>
20915 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20916 to allow for free-lang-data replacements similar to verify_type_variant.
20918 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20919 Alan Hayward <alan.hayward@arm.com>
20920 David Sherwood <david.sherwood@arm.com>
20922 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20923 VOIDmode for the LABEL_REF.
20925 2017-08-22 Richard Biener <rguenther@suse.de>
20927 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20929 2017-08-22 Richard Biener <rguenther@suse.de>
20931 * common.opt (feliminate-dwarf2-dups): Ignore.
20932 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20933 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20934 same_die_p_wrap, compute_section_prefix,
20935 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20936 (comdat_symbol_id, comdat_symbol_number): Likewise.
20937 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20939 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20940 (output_die): Mark unreachable path unreachable.
20941 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20942 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20943 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20944 (dwarf2out_early_finish): Likewise.
20946 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20948 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20950 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20953 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20954 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20955 (avr_attribute_table) <io, io_low, address>: Initialize
20956 .decl_required with true.
20958 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20960 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20961 undocumented debugging options.
20962 (-mvsx-scalar-double): Likewise.
20963 (-mallow-df-permute): Likewise.
20964 (-mvectorize-builtins): Likewise.
20965 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20966 (rs6000_builtin_vectorized_function): Likewise.
20967 (rs6000_builtin_md_vectorized_function): Likewise.
20968 (rs6000_opt_vars): Likewise.
20970 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20973 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20974 (*btrq_imm): Rename from *btrq.
20975 (*btcq_imm): Rename from *btcq.
20976 (btsc): New code attribute.
20977 (*<btsc><mode>): New insn pattern.
20978 (*btr<mode>): Ditto.
20979 (*<btsc><mode>_mask): New insn_and_split pattern.
20980 (*btr<mode>_mask): Ditto.
20982 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20983 Alan Hayward <alan.hayward@arm.com>
20984 David Sherwood <david.sherwood@arm.com>
20986 * function.c (pad_below): Simplify padding calculation.
20988 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20989 Alan Hayward <alan.hayward@arm.com>
20990 David Sherwood <david.sherwood@arm.com>
20992 * target.def (function_prologue): Remove frame size argument.
20993 (function_epilogue): Likewise.
20994 * doc/tm.texi: Regenerate.
20995 * final.c (final_start_function): Update call to function_prologue.
20996 (final_end_function): Update call to function_epilogue.
20997 (default_function_pro_epilogue): Remove frame size argument.
20998 * output.h (default_function_pro_epilogue): Likewise.
20999 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
21000 (arm_output_function_prologue): Likewise.
21001 * config/frv/frv.c (frv_function_prologue): Likewise.
21002 (frv_function_epilogue): Likewise.
21003 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
21004 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
21005 (ia64_output_function_epilogue): Likewise.
21006 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
21007 (m32r_output_function_epilogue): Likewise.
21008 * config/microblaze/microblaze.c (microblaze_function_prologue)
21009 (microblaze_function_epilogue): Likewise.
21010 * config/mips/mips.c (mips_output_function_prologue): Likewise.
21011 (mips_output_function_epilogue): Likewise.
21012 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
21013 (mmix_target_asm_function_epilogue): Likewise.
21014 * config/msp430/msp430.c (msp430_start_function): Likewise.
21015 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
21016 (nds32_asm_function_epilogue): Likewise.
21017 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
21018 * config/pa/pa.c (pa_output_function_prologue): Likewise.
21019 (pa_output_function_epilogue): Likewise.
21020 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
21021 (rs6000_output_function_epilogue): Likewise.
21022 * config/rl78/rl78.c (rl78_start_function): Likewise.
21023 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
21024 (rs6000_output_function_epilogue): Likewise.
21025 * config/rx/rx.c (rx_output_function_prologue): Likewise.
21026 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
21027 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
21028 (sparc_asm_function_epilogue): Likewise.
21030 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21032 * tree.h (type_has_mode_precision_p): New function.
21033 * convert.c (convert_to_integer_1): Use it.
21034 * expr.c (expand_expr_real_2): Likewise.
21035 (expand_expr_real_1): Likewise.
21036 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
21037 * match.pd: Likewise.
21038 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
21039 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
21040 * tree-tailcall.c (process_assignment): Likewise.
21041 * tree-vect-loop.c (vectorizable_reduction): Likewise.
21042 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
21043 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
21044 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
21045 (vectorizable_assignment): Likewise.
21046 (vectorizable_shift): Likewise.
21047 (vectorizable_operation): Likewise.
21048 * tree-vrp.c (register_edge_assert_for_2): Likewise.
21050 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
21052 * match.pd: Add pow (C, x) simplification.
21054 2017-08-21 Richard Biener <rguenther@suse.de>
21056 PR tree-optimization/81900
21057 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
21058 for blocks with abnormal predecessors.
21059 (compute_antic): Do not set visited flag prematurely.
21061 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
21064 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
21066 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21068 * stor-layout.h (vector_type_mode): Move to...
21069 * tree.h (vector_type_mode): ...here.
21070 * stor-layout.c (vector_type_mode): Move to...
21071 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
21073 2017-08-21 Richard Biener <rguenther@suse.de>
21075 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
21076 register_external_die hooks.
21077 (debug_false_tree_charstarstar_uhwistar): Declare.
21078 (debug_nothing_tree_charstar_uhwi): Likewise.
21079 * debug.c (do_nothing_debug_hooks): Adjust.
21080 (debug_false_tree_charstarstar_uhwistar): New do nothing.
21081 (debug_nothing_tree_charstar_uhwi): Likewise.
21082 * dbxout.c (dbx_debug_hooks): Adjust.
21083 (xcoff_debug_hooks): Likewise.
21084 * sdbout.c (sdb_debug_hooks): Likewise.
21085 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21086 * dwarf2out.c (macinfo_label_base): New global.
21087 (dwarf2out_register_external_die): New function for the
21088 register_external_die hook.
21089 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
21090 (dwarf2_debug_hooks): Use them.
21091 (dwarf2_lineno_debug_hooks): Adjust.
21092 (struct die_struct): Add with_offset flag.
21093 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
21094 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
21095 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
21096 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
21097 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
21098 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
21099 defining section names for the early LTO debug variants.
21100 (reset_indirect_string): New helper.
21101 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
21102 (print_dw_val): Add support for offsetted symbol references.
21103 (get_ultimate_context): Split out from is_cxx.
21104 (is_cxx): Use get_ultimate_context.
21105 (is_fortran): Add decl overload.
21106 (compute_comp_unit_symbol): Split out worker from
21107 compute_section_prefix.
21108 (compute_section_prefix): Call compute_comp_unit_symbol and
21109 set comdat_type_p here.
21110 (output_die): Skip DIE symbol output for the LTO added one.
21111 Handle DIE symbol references with offset.
21112 (output_comp_unit): Guard section name mangling properly.
21113 For LTO debug sections emit a symbol at the section beginning
21114 which we use to refer to its DIEs.
21115 (add_abstract_origin_attribute): For DIEs registered via
21116 dwarf2out_register_external_die directly refer to the early
21117 DIE rather than indirectly through the shadow one we created.
21118 Remove obsolete call to dwarf2out_abstract_function for
21119 non-function/block origins.
21120 (gen_array_type_die): When generating early LTO debug do
21121 not emit DW_AT_string_length.
21122 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
21123 late when in LTO. As suggested place a gcc_unreachable for
21124 the DECL_ABSTRACT_P case.
21125 (gen_subprogram_die): Avoid another specification DIE
21126 for early built declarations/definitions for the late LTO case.
21127 (gen_variable_die): Add type references for late duplicated VLA dies
21129 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
21130 we have the abstract instance already.
21131 (process_scope_var): Adjust decl DIE contexts in LTO which
21132 first puts them in limbo.
21133 (gen_decl_die): Do not generate type DIEs late apart from
21134 types for VLAs or for decls we do not yet have a DIE. Do not
21135 call dwarf2out_abstract_function late.
21136 (dwarf2out_early_global_decl): Make sure to create DIEs
21137 for abstract instances of a decl first.
21138 (dwarf2out_late_global_decl): Adjust comment.
21139 (output_macinfo_op): With multiple macro sections use
21140 macinfo_label_base to distinguish labels.
21141 (output_macinfo): Likewise. Update macinfo_label_base.
21142 Pass in the line info label.
21143 (note_variable_value_in_expr): When generating LTO resolve
21144 all variable values here by generating DIEs as needed.
21145 (init_sections_and_labels): Add early LTO debug flag parameter
21146 and generate different sections and names if set. Add generation
21147 counter for the labels so we can have multiple of them.
21148 (reset_dies): Helper to allow DIEs to be output multiple times.
21149 (dwarf2out_finish): When outputting DIEs to the fat part of an
21150 LTO object first reset DIEs.
21151 (dwarf2out_early_finish): Output early DIEs when generating LTO.
21152 (modified_type_die): Check for decl_ultimate_origin being self
21154 (gen_type_die_with_usage): Likewise.
21155 (gen_typedef_die): Allow decl_ultimate_origin being self.
21156 (set_decl_abstract_flags): Remove.
21157 (set_block_abstract_flags): Likewise.
21158 (dwarf2out_abstract_function): Treat the early generated DIEs
21159 as the abstract copy and only add DW_AT_inline and
21160 DW_AT_artificial here and call set_decl_origin_self.
21161 If the DIE has an abstract origin don't do anything.
21162 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
21163 if we have none yet (Go fails to build one, PR78628).
21164 (variably_modified_type_p): Prevent endless recursion for Ada
21165 cyclic pointer types.
21166 * lto-streamer-in.c: Include debug.h.
21167 (dref_queue): New global.
21168 (lto_read_tree_1): Stream in DIE references.
21169 (lto_input_tree): Register DIE references.
21170 (input_function): Stream DECL_DEBUG_ARGS.
21171 * lto-streamer-out.c: Include debug.h.
21172 (lto_write_tree_1): Output DIE references.
21173 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
21174 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
21175 (output_function): Stream DECL_DEBUG_ARGS.
21176 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
21177 Stream DECL_ABSTRACT_ORIGIN.
21178 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
21179 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
21180 DECL_CONTEXT for file-scope decls.
21181 * lto-streamer.h (struct dref_entry): Declare.
21182 (dref_queue): Likewise.
21183 * cfgexpand.c (pass_expand::execute): Do not call the
21184 outlining_inline_function hook here.
21185 * lto-wrapper.c (debug_obj): New global.
21186 (tool_cleanup): Unlink it if required.
21187 (debug_objcopy): New function.
21188 (run_gcc): Handle early debug sections in the IL files by
21189 extracting them to separate files, partially linkin them and
21190 feeding the result back as result to the linker.
21191 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
21192 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
21193 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
21194 sections into a separate segment.
21195 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
21197 (darwin_asm_dwarf_section): Likewise.
21198 (darwin_asm_output_dwarf_offset): Likewise.
21199 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
21201 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21202 Alan Hayward <alan.hayward@arm.com>
21203 David Sherwood <david.sherwood@arm.com>
21205 * read-md.h (md_reader::record_potential_iterator_use): Replace
21206 pointer argument with an rtx and an index.
21207 * read-rtl.c (iterator_group::apply_iterator): Likewise.
21208 (apply_mode_iterator): Likewise.
21209 (apply_code_iterator): Likewise.
21210 (apply_int_iterator): Likewise.
21211 (apply_subst_iterator): Likewise.
21212 (record_iterator_use): Likewise.
21213 (record_attribute_use): Likewise.
21214 (md_reader::record_potential_iterator_use): Likewise. Update calls
21215 to record_iterator_use and apply_iterator.
21216 (iterator_use): Replace ptr with x and index.
21217 (attribute_use): Likewise.
21218 (apply_attribute_uses): Update calls to apply_iterator.
21219 (apply_iterators): Likewise. Update initialization of iterator_use.
21220 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
21221 and record_potential_iterator_use.
21222 (rtx_reader::read_rtx_operand): Likewise.
21224 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21225 Alan Hayward <alan.hayward@arm.com>
21226 David Sherwood <david.sherwood@arm.com>
21228 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
21231 2017-08-21 Richard Biener <rguenther@suse.de>
21233 PR middle-end/81884
21234 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
21235 at struct end conservatively when comparing common bases.
21237 2017-08-21 Richard Biener <rguenther@suse.de>
21239 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
21240 (mem_ref_in_stmt): Remove.
21241 (determine_max_movement): Use ref index to get at the reference.
21242 (invariantness_dom_walker::before_dom_children): Deal with
21243 lim data already initialized.
21244 (gather_mem_refs_stmt): Initialize lim data and record ref index.
21246 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21248 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
21249 (TARGET_ISA_ROUND): Ditto.
21250 (TARGET_ROUND): Ditto.
21251 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
21252 * config/i386/i386.md: Ditto.
21253 * config/i386/sse.md: Ditto.
21254 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
21255 with OPTION_MASK_ISA_SSE4_1.
21257 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21260 * doc/extend.texi (x86 Built-in Functions): Correct the name of
21261 __builtin_ia32_lzcnt_u16.
21263 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
21266 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
21267 (rs6000_set_current_function): Rewrite function to use it.
21269 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
21272 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
21273 and TYPE_WARN_IF_NOT_ALIGN.
21274 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
21275 (handle_warn_if_not_align): New.
21276 (place_union_field): Call handle_warn_if_not_align.
21277 (place_field): Call handle_warn_if_not_align.
21278 Copy TYPE_WARN_IF_NOT_ALIGN.
21279 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
21280 (layout_type): Likewise.
21281 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
21283 (tree_decl_common): Add warn_if_not_align.
21284 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
21285 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
21286 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
21287 (DECL_WARN_IF_NOT_ALIGN): Likewise.
21288 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
21289 * doc/extend.texi: Document warn_if_not_aligned attribute.
21290 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
21292 2017-08-17 Martin Liska <mliska@suse.cz>
21295 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
21296 (get_data_dependence): Use it as pointer type.
21297 (distribute_loop): Likewise.
21299 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21301 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
21302 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
21303 (p8_vmrgow_<mode>_direct): New define_insn.
21304 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
21305 handle endianness for vmrgew and vmrgow permute patterns.
21307 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21309 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
21310 * config/rs6000/rs6000-cpus.def: Remove comment.
21311 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
21312 (POWERPC_MASKS): Likewise.
21313 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
21314 use of TARGET_VSX_TIMODE.
21315 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
21316 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
21317 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
21318 (rs6000_option_override_internal): Remove dead code.
21319 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
21320 (rs6000_legitimize_reload_address): Likewise.
21321 (rs6000_legitimate_address_p): Likewise.
21322 (rs6000_opt_masks): Delete "vsx-timode".
21323 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
21324 from function comment.
21325 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
21326 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
21327 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
21329 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
21330 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
21331 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
21332 (VSX_TI): Likewise.
21334 (define_peephole2): Likewise.
21336 2017-08-17 Martin Sebor <msebor@redhat.com>
21339 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
21340 past the end of an array.
21341 (test_pp_format): Add test cases.
21343 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
21345 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
21346 missing ECF_NOTHROW flags.
21348 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21351 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
21352 operands residing in integer registers.
21353 (*vsx_le_perm_load_<mode>): Likewise.
21354 (*vsx_le_perm_store_<mode>): Likewise.
21355 (define_peephole2): Add peepholes to optimize the above.
21357 2017-08-17 Marek Polacek <polacek@redhat.com>
21359 PR middle-end/81814
21360 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
21361 to mimic what shorten_compare did. Change the return type to bool.
21362 (fold_cond_expr_with_comparison): Update call to
21363 operand_equal_for_comparison_p.
21364 (fold_ternary_loc): Likewise.
21366 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
21368 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
21370 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
21372 2017-08-17 Richard Biener <rguenther@suse.de>
21374 * tree-ssa-structalias.c (solve_graph): When propagating
21375 to successors update the graphs succ edges and avoid duplicate work.
21377 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
21380 * config/i386/i386.c (ix86_option_override_internal): Save target
21381 specific options after ix86_stack_protector_guard_reg was changed.
21383 2017-08-17 Richard Biener <rguenther@suse.de>
21385 PR tree-optimization/81827
21386 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
21387 (new_var_info): Initialize it conservatively.
21388 (get_call_vi): Mark register vars.
21389 (new_scalar_tmp_constraint_exp): Likewise.
21390 (handle_rhs_call): Likewise.
21391 (handle_const_call): Likewise.
21392 (create_function_info_for): Likewise.
21393 (solve_constraints): Sort varinfos to separate register from
21394 non-register vars to pack points-to solution bitmaps during
21397 2017-08-17 Marek Polacek <polacek@redhat.com>
21399 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
21401 2017-08-17 Richard Biener <rguenther@suse.de>
21403 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
21404 to true when overflow is undefined and we saturated the result.
21406 2017-08-17 Alan Modra <amodra@gmail.com>
21409 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
21410 Don't use store multiple if only one reg needs saving.
21411 (interesting_frame_related_regno): New function.
21412 (rs6000_frame_related): Don't emit frame info for regs that
21414 (rs6000_emit_epilogue): Likewise.
21416 2017-08-16 Nathan Sidwell <nathan@acm.org>
21418 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
21419 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
21420 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
21421 * tree.c (free_lang_data_in_type): Use else-if chain. Always
21422 clear TYPE_LANG_1. Remove obsolete member-function stripping.
21423 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
21424 (verify_type): Adjust for TYPE_BINFO move.
21425 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
21426 process TYPE_BINFO directly.
21427 (hash_tree): Likewise.
21428 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21430 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21433 2017-08-16 David Malcolm <dmalcolm@redhat.com>
21435 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
21437 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
21440 * config/i386/i386.md (*anddi_1_btr): Change predicates of
21441 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
21442 Add ix86_binary_operator_ok to insn constraint.
21443 (*iordi_1_bts): Ditto.
21444 (*xordi_1_btc): Ditto.
21445 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
21446 Update corresponding peephole2 pattern.
21450 2017-08-16 Bin Cheng <bin.cheng@arm.com>
21452 PR tree-optimization/81832
21453 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
21454 copy loop header which has IFN_LOOP_DIST_ALIAS call.
21456 2017-08-16 Marek Polacek <polacek@redhat.com>
21459 * fold-const.c (fold_indirect_ref_1): Restore original behavior
21460 regarding size_zero_node.
21462 2017-08-16 Martin Liska <mliska@suse.cz>
21465 * config.gcc: Respect previously set extra_objs in case
21468 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
21470 PR tree-optimization/81835
21471 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
21472 the phi SSA_NAME. Check that the condition in a COND_EXPR does
21473 not depend on the phi.
21475 2017-08-16 Alan Modra <amodra@gmail.com>
21477 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
21480 2017-08-16 Alan Modra <amodra@gmail.com>
21482 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
21483 (save_reg_p): ..into this. Update all callers.
21484 (first_reg_to_save): Simplify.
21486 2017-08-16 Alan Modra <amodra@gmail.com>
21488 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
21491 2017-08-15 Joseph Myers <joseph@codesourcery.com>
21495 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
21496 constant count if that count is less than 32.
21498 2017-08-15 Nathan Sidwell <nathan@acm.org>
21500 * gcc.c (execute): Emit friendlier message if inferior is killed
21501 by an external cause.
21503 2017-08-15 Richard Biener <rguenther@suse.de>
21505 PR tree-optimization/81790
21506 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
21507 CONSTRUCTORs from simplifying and VN.
21509 2017-08-14 Martin Sebor <msebor@redhat.com>
21511 * builtin-attrs.def: Add comments.
21513 2017-08-14 Martin Sebor <msebor@redhat.com>
21516 * doc/extend.texi (attribute nonstring): Document new attribute.
21518 2017-08-14 Martin Sebor <msebor@redhat.com>
21521 * tree-diagnostic.c (default_tree_printer): Handle %G.
21522 * gimple-pretty-print.h (percent_G_format): Declare new function.
21523 * gimple-pretty-print.c (percent_G_format): Define.
21524 * tree-pretty-print.c (percent_K_format): Add argument.
21526 2017-08-14 Martin Sebor <msebor@redhat.com>
21528 PR translation/79998
21529 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
21530 Remove a stray space.
21532 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
21535 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
21536 (*iordi_1_bts): Ditto.
21537 (*xordi_1_btc): Ditto.
21539 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21542 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
21543 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21545 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
21546 quoted strings, and make more translator-friendly.
21547 (darwin_rs6000_override_options): Likewise.
21548 (rs6000_option_override_internal): Likewise.
21549 (rs6000_return_in_memory): Fix overlong line.
21550 (init_cmulative_args): Use quoted strings, and make more
21551 translator-friendly.
21552 (rs6000_pass_by_reference): Fix overlong line.
21553 (def_builtin): Use quoted strings.
21554 (altivec_expand_predicate_builtin): Use quoted strings, and make
21555 more translator-friendly.
21556 (htm_expand_builtin): Use quoted strings.
21557 (cpu_expand_builtin): Use quoted strings, and make more
21558 translator-friendly.
21559 (altivec_expand_builtin): Likewise.
21560 (paired_expand_predicate_builtin): Likewise.
21561 (rs6000_invalid_builtin): Likewise.
21562 (builtin_function_type): Use quoted strings.
21563 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
21564 more translator-friendly.
21565 (rs6000_trampoline_init): Likewise.
21566 (rs6000_handle_altivec_attribute): Likewise.
21567 (rs6000_inner_target_options): Use quoted strings.
21568 (rs6000_disable_incompatible_switches): Likewise.
21569 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
21570 strings, and make more translator-friendly.
21571 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
21573 2017-08-14 Bin Cheng <bin.cheng@arm.com>
21575 PR tree-optimization/81799
21576 * tree-loop-distribution.c (version_loop_by_alias_check): Force
21577 cond_expr to simple gimple operand.
21579 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
21581 PR middle-end/46932
21582 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
21584 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
21588 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
21589 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
21590 TARGET_GASISR_PROLOGUES.
21591 * config/avr/avr.c (avr_option_override): Same.
21592 (avr_pass_pre_proep::execute): Same.
21594 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
21597 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
21598 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
21600 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
21602 * config/i386/i386.md (*load_tp_<mode>): Redefine as
21603 define_insn_and_split. Split to a memory load from 0 in
21604 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
21605 using PTR mode iterator.
21606 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
21607 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
21608 (*add_tp_<mode>): Redefine as define_insn_and_split.
21609 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
21610 address space. Merge with *add_tp_x32 using PTR mode iterator.
21611 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
21612 Split to an add with a memory load from 0 in
21613 DEFAULT_TLS_SEG_REG address space.
21615 2017-08-12 Andrew Pinski <apinski@cavium.com>
21617 * config/aarch64/aarch64-option-extensions.def (rdma):
21618 Fix feature string to what Linux prints out in /proc/cpuinfo.
21620 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
21623 * dwarf2out.c (modified_type_die): For C typedef types that have
21624 an ultimate origin, process the ultimate origin instead of the
21626 (gen_typedef_die): Assert that input DECLs have no ultimate
21628 (gen_type_die_with_usage): For typedef variants that have an
21629 ultimate origin, just call gen_decl_die on the original DECL.
21630 (process_scope_var): Avoid creating DIEs for local typedefs and
21631 concrete static variables.
21633 2017-08-12 Alan Modra <amodra@gmail.com>
21637 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
21638 match gnu-user.h startfile.
21639 (ENDFILE_LINUX_SPEC): Similarly.
21641 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
21644 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
21646 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
21648 2017-08-11 Tamar Christina <tamar.christina@arm.com>
21649 * config/aarch64/aarch64.md (mov<mode>): Change.
21650 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21651 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
21652 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
21654 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
21656 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
21657 for storage order barriers.
21659 2017-08-11 Martin Liska <mliska@suse.cz>
21662 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
21663 variables of void type.
21665 2017-08-11 Martin Liska <mliska@suse.cz>
21667 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
21668 TARGET_SUPPORTS_ALIASES.
21669 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
21670 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
21671 (optimize_weakref): Likewise.
21672 * symtab.c (symtab_node::noninterposable_alias): Likewise.
21673 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
21674 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
21676 2017-08-11 Martin Liska <mliska@suse.cz>
21679 * config/i386/i386.c (make_resolver_func): Do complete
21680 refactoring of the function.
21682 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21685 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
21686 * config/i386/i386.c (ix86_stack_protect_guard): Use
21687 ix86_stack_protect_guard_symbol_str to generate varible declaration.
21688 * doc/invoke.texi (x86 Options): Document
21689 -mstack-protector-guard-symbol= option.
21691 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21693 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
21694 * config/i386/i386.c (ix86_split_stack_guard): New function.
21695 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
21696 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
21697 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
21698 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
21699 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
21700 (split_stack_space_check): Call ix86_split_stack_guard.
21702 2017-08-10 Martin Sebor <msebor@redhat.com>
21704 * print-tree.c (print_node): Print location using the established
21706 Replace spaces with colons.
21707 (debug_raw, debug): Ditto.
21709 2017-08-10 Martin Sebor <msebor@redhat.com>
21712 * pretty-print.c (pp_format): Correct the handling of %s precision.
21714 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
21717 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21719 (ix86_finalize_stack_frame_flags): This. Also clear
21720 frame_pointer_needed if -fno-omit-frame-pointer is used without
21722 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21723 with ix86_finalize_stack_frame_flags.
21724 (ix86_expand_epilogue): Likewise.
21725 (ix86_expand_split_stack_prologue): Likewise.
21726 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
21728 2017-08-10 Martin Liska <mliska@suse.cz>
21731 * c-attribs.c (handle_target_attribute):
21732 Report warning for an empty string argument of target attribute.
21734 2017-08-09 Jakub Jelinek <jakub@redhat.com>
21737 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
21739 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
21740 or DECL_NONLOCAL labels.
21741 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
21742 or DECL_NONLOCAL labels here.
21744 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
21746 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
21747 to indicate when early gimple folding has been disabled.
21748 (rs6000_gimple_fold_builtin): Add debug content.
21749 (rs6000_invalid_builtin): Fix whitespace.
21750 (rs6000_expand_builtin): Fix whitespace.
21751 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
21753 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
21756 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
21757 SAVE_MULTIPLE if not all the registers that saves, should be saved.
21759 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
21761 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
21762 (qdf24xx): Likewise.
21763 * config/aarch64/aarch64.md: Include falkor.md.
21764 * config/aarch64/falkor.md: New.
21766 2017-08-09 Marek Polacek <polacek@redhat.com>
21769 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
21770 * diagnostic.c (emit_diagnostic): Add a comment.
21771 (emit_diagnostic_valist): New function.
21773 2017-08-09 Marek Polacek <polacek@redhat.com>
21776 * input.c (make_location): New overload.
21777 * input.h (make_location): Declare.
21779 2017-08-08 Alan Modra <amodra@gmail.com>
21780 H.J. Lu <hongjiu.lu@intel.com>
21783 * gcc.c (NO_PIE_SPEC): Delete.
21784 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21786 (LINK_PIE_SPEC): ..to here.
21787 (LINK_COMMAND_SPEC): Support -no-pie.
21788 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21789 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21791 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21792 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21793 (ENDFILE_CRTEND_SPEC): Similarly.
21795 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21798 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21799 (mstack-protector-guard-offset=): Ditto.
21800 * config/i386/i386.c (ix86_option_override): Handle
21801 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21803 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21804 ix86_stack_protect_guard_offset variables.
21805 (TARGET_STACK_PROTECT_GUARD): Always define.
21806 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21807 and -mstack-protector-guard-offset= options.
21809 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21811 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21812 boundary case for the last candidate.
21814 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21816 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21817 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21819 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21821 PR middle-end/19706
21822 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21823 * config/aarch64/aarch64-builtins.c
21824 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21825 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21826 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21828 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21829 Andrew Pinski <pinskia@gmail.com>
21831 PR middle-end/19706
21832 * internal-fn.def (XORSIGN): New.
21833 * optabs.def (xorsign_optab): New.
21834 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21835 (convert_expand_mult_copysign): New.
21836 (pass_optimize_widening_mul::execute): Call
21837 convert_expand_mult_copysign.
21839 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21841 PR tree-optimization/81354
21842 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21843 Insert on edges rather than explicitly creating landing pads.
21844 (analyze_candidates_and_replace): Commit edge inserts.
21846 2017-08-08 Richard Biener <rguenther@suse.de>
21848 PR middle-end/81719
21849 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21850 (expand_simple_operations): Also look through ADDR_EXPRs with
21851 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21853 2017-08-08 Richard Biener <rguenther@suse.de>
21855 PR tree-optimization/81723
21856 * tree-vect-slp.c (struct bst_traits): New hash traits.
21857 (bst_fail): New global.
21858 (vect_build_slp_tree_2): New worker, split out from ...
21859 (vect_build_slp_tree): ... this now wrapping it with using
21860 bst_fail set to cache SLP tree build fails. Properly handle
21862 (vect_analyze_slp_instance): Allocate and free bst_fail.
21864 2017-08-08 Martin Liska <mliska@suse.cz>
21867 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21868 LABEL_DECLs that can be from a different function.
21870 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21872 PR tree-optimization/81744
21873 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21874 loop's number of iterations.
21876 2017-08-08 Martin Liska <mliska@suse.cz>
21878 * asan.c: Include header files.
21879 * attribs.c (build_decl_attribute_variant): New function moved
21881 (build_type_attribute_qual_variant): Likewise.
21882 (cmp_attrib_identifiers): Likewise.
21883 (simple_cst_list_equal): Likewise.
21884 (omp_declare_simd_clauses_equal): Likewise.
21885 (attribute_value_equal): Likewise.
21886 (comp_type_attributes): Likewise.
21887 (build_type_attribute_variant): Likewise.
21888 (lookup_ident_attribute): Likewise.
21889 (remove_attribute): Likewise.
21890 (merge_attributes): Likewise.
21891 (merge_type_attributes): Likewise.
21892 (merge_decl_attributes): Likewise.
21893 (merge_dllimport_decl_attributes): Likewise.
21894 (handle_dll_attribute): Likewise.
21895 (attribute_list_equal): Likewise.
21896 (attribute_list_contained): Likewise.
21897 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21898 (lookup_attribute_by_prefix): Likewise.
21899 * bb-reorder.c: Include header files.
21900 * builtins.c: Likewise.
21901 * calls.c: Likewise.
21902 * cfgexpand.c: Likewise.
21903 * cgraph.c: Likewise.
21904 * cgraphunit.c: Likewise.
21905 * convert.c: Likewise.
21906 * dwarf2out.c: Likewise.
21907 * final.c: Likewise.
21908 * fold-const.c: Likewise.
21909 * function.c: Likewise.
21910 * gimple-expr.c: Likewise.
21911 * gimple-fold.c: Likewise.
21912 * gimple-pretty-print.c: Likewise.
21913 * gimple.c: Likewise.
21914 * gimplify.c: Likewise.
21915 * hsa-common.c: Likewise.
21916 * hsa-gen.c: Likewise.
21917 * internal-fn.c: Likewise.
21918 * ipa-chkp.c: Likewise.
21919 * ipa-cp.c: Likewise.
21920 * ipa-devirt.c: Likewise.
21921 * ipa-fnsummary.c: Likewise.
21922 * ipa-inline.c: Likewise.
21923 * ipa-visibility.c: Likewise.
21925 * lto-cgraph.c: Likewise.
21926 * omp-expand.c: Likewise.
21927 * omp-general.c: Likewise.
21928 * omp-low.c: Likewise.
21929 * omp-offload.c: Likewise.
21930 * omp-simd-clone.c: Likewise.
21931 * opts-global.c: Likewise.
21932 * passes.c: Likewise.
21933 * predict.c: Likewise.
21934 * sancov.c: Likewise.
21935 * sanopt.c: Likewise.
21936 * symtab.c: Likewise.
21937 * toplev.c: Likewise.
21938 * trans-mem.c: Likewise.
21939 * tree-chkp.c: Likewise.
21940 * tree-eh.c: Likewise.
21941 * tree-into-ssa.c: Likewise.
21942 * tree-object-size.c: Likewise.
21943 * tree-parloops.c: Likewise.
21944 * tree-profile.c: Likewise.
21945 * tree-ssa-ccp.c: Likewise.
21946 * tree-ssa-live.c: Likewise.
21947 * tree-ssa-loop.c: Likewise.
21948 * tree-ssa-sccvn.c: Likewise.
21949 * tree-ssa-structalias.c: Likewise.
21950 * tree-ssa.c: Likewise.
21951 * tree-streamer-in.c: Likewise.
21952 * tree-vectorizer.c: Likewise.
21953 * tree-vrp.c: Likewise.
21954 * tsan.c: Likewise.
21955 * ubsan.c: Likewise.
21956 * varasm.c: Likewise.
21957 * varpool.c: Likewise.
21958 * tree.c: Remove functions moved to attribs.[ch].
21959 * tree.h: Likewise.
21960 * config/aarch64/aarch64.c: Add attrs.h header file.
21961 * config/alpha/alpha.c: Likewise.
21962 * config/arc/arc.c: Likewise.
21963 * config/arm/arm.c: Likewise.
21964 * config/avr/avr.c: Likewise.
21965 * config/bfin/bfin.c: Likewise.
21966 * config/c6x/c6x.c: Likewise.
21967 * config/cr16/cr16.c: Likewise.
21968 * config/cris/cris.c: Likewise.
21969 * config/darwin.c: Likewise.
21970 * config/epiphany/epiphany.c: Likewise.
21971 * config/fr30/fr30.c: Likewise.
21972 * config/frv/frv.c: Likewise.
21973 * config/ft32/ft32.c: Likewise.
21974 * config/h8300/h8300.c: Likewise.
21975 * config/i386/winnt.c: Likewise.
21976 * config/ia64/ia64.c: Likewise.
21977 * config/iq2000/iq2000.c: Likewise.
21978 * config/lm32/lm32.c: Likewise.
21979 * config/m32c/m32c.c: Likewise.
21980 * config/m32r/m32r.c: Likewise.
21981 * config/m68k/m68k.c: Likewise.
21982 * config/mcore/mcore.c: Likewise.
21983 * config/microblaze/microblaze.c: Likewise.
21984 * config/mips/mips.c: Likewise.
21985 * config/mmix/mmix.c: Likewise.
21986 * config/mn10300/mn10300.c: Likewise.
21987 * config/moxie/moxie.c: Likewise.
21988 * config/msp430/msp430.c: Likewise.
21989 * config/nds32/nds32-isr.c: Likewise.
21990 * config/nds32/nds32.c: Likewise.
21991 * config/nios2/nios2.c: Likewise.
21992 * config/nvptx/nvptx.c: Likewise.
21993 * config/pa/pa.c: Likewise.
21994 * config/pdp11/pdp11.c: Likewise.
21995 * config/powerpcspe/powerpcspe.c: Likewise.
21996 * config/riscv/riscv.c: Likewise.
21997 * config/rl78/rl78.c: Likewise.
21998 * config/rx/rx.c: Likewise.
21999 * config/s390/s390.c: Likewise.
22000 * config/sh/sh.c: Likewise.
22001 * config/sol2.c: Likewise.
22002 * config/sparc/sparc.c: Likewise.
22003 * config/spu/spu.c: Likewise.
22004 * config/stormy16/stormy16.c: Likewise.
22005 * config/tilegx/tilegx.c: Likewise.
22006 * config/tilepro/tilepro.c: Likewise.
22007 * config/v850/v850.c: Likewise.
22008 * config/vax/vax.c: Likewise.
22009 * config/visium/visium.c: Likewise.
22010 * config/xtensa/xtensa.c: Likewise.
22012 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
22015 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
22016 constraints since the -mupper-regs-* switches have been
22018 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
22019 into a vector from a double word element that was extracted from
22020 another vector, and eliminate extra XXPERMDI instructions.
22021 (vsx_concat_<mode>_2): Likewise.
22022 (vsx_concat_<mode>_3): Likewise.
22023 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
22024 concat to allow optimizing inserts from previous extracts.
22026 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
22028 * config/i386/i386.c (ix86_stack_protect_guard): Generate
22029 memory reference to a SSP offset in TLS address space.
22030 (ix86_print_operand) <case '@'>: Remove.
22031 (ix86_print_operand_punct_valid_p): Remove '@' code.
22032 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
22033 UNSPEC_SP_TLS_TEST.
22034 (stack_tls_protect_set_<mode>): Remove.
22035 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
22036 (stack_tls_protect_test_<mode>): Remove.
22037 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
22039 2017-08-07 Olivier Hainque <hainque@adacore.com>
22042 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
22044 2017-08-07 Douglas Rupp <rupp@adacore.com>
22046 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
22047 variable was referenced as multidir in command.
22049 2017-08-07 Jakub Jelinek <jakub@redhat.com>
22052 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
22055 2017-08-07 Martin Liska <mliska@suse.cz>
22057 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
22058 * config/rl78/rl78.c: Add include of attribs.h.
22059 * config/sh/sh.c: Likewise.
22060 * config/v850/v850.c: Likewise.
22062 2017-08-07 Tom de Vries <tom@codesourcery.com>
22064 PR middle-end/78266
22065 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
22067 2017-08-07 Martin Liska <mliska@suse.cz>
22069 * config/mips/mips.c: Include attribs.h.
22071 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
22074 * doc/invoke.texi: Document change in behvaior for -Ofast for
22077 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
22079 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
22081 (aarch64_pop_regs): Likewise.
22082 (aarch64_gen_load_pair): Likewise.
22083 (aarch64_save_callee_saves): Likewise.
22084 (aarch64_restore_callee_saves): Likewise.
22086 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22088 * config/i386/i386.c: Revert the last change.
22090 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22093 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
22095 (ix86_finalize_stack_frame_flags): This. Also clear
22096 frame_pointer_needed if -fno-omit-frame-pointer is used without
22098 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
22099 with ix86_finalize_stack_frame_flags.
22100 (ix86_expand_epilogue): Likewise.
22101 (ix86_expand_split_stack_prologue): Likewise.
22103 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22106 * config/i386/i386.c (get_builtin_code_for_version): Set priority
22107 to P_AES for Westmere.
22109 2017-08-07 Jonathan Yong <10walls@gmail.com>
22111 * config/i386/mingw.opt (fset-stack-executable): Removed.
22112 * config/i386/cygming.opt (fset-stack-executable): Moved
22114 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
22116 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
22118 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
22120 2017-08-07 Marek Polacek <polacek@redhat.com>
22122 PR middle-end/81737
22123 * fold-const.c (fold_indirect_ref_1): Check type_domain.
22125 2017-08-07 Martin Liska <mliska@suse.cz>
22127 * attribs.h (canonicalize_attr_name): New function.
22128 (cmp_attribs): Move from c-format.c and adjusted.
22129 (is_attribute_p): Moved from tree.h.
22130 * tree-inline.c: Add new includes.
22131 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
22132 (private_is_attribute_p): Remove.
22133 (private_lookup_attribute): Likewise.
22134 (private_lookup_attribute_by_prefix): Simplify.
22135 (remove_attribute): Use is_attribute_p.
22136 * tree.h: Remove removed declarations.
22138 2017-08-07 Jakub Jelinek <jakub@redhat.com>
22140 PR middle-end/81698
22141 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
22142 instead of computing it in the function. Formatting fix.
22143 (expand_case): Don't rely on default_edge being the first edge,
22144 clear it if removing it, pass default_edge to
22145 emit_case_dispatch_table.
22146 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
22149 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
22151 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
22152 insn in the function, emit NOP after the insn.
22154 2017-08-06 Tom de Vries <tom@codesourcery.com>
22156 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
22159 2017-08-06 Tom de Vries <tom@codesourcery.com>
22161 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
22164 2017-08-04 Yury Gribov <tetra2005@gmail.com>
22166 PR tree-optimization/57371
22167 * match.pd: New pattern.
22169 2017-08-04 Marek Polacek <polacek@redhat.com>
22171 PR middle-end/81695
22172 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
22173 perform the computation in offset_int.
22175 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22177 PR tree-optimization/81136
22178 * tree-vectorizer.h: Include tree-hash-traits.h.
22179 (vec_base_alignments): New typedef.
22180 (vec_info): Add a base_alignments field.
22181 (vect_record_base_alignments): Declare.
22182 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
22184 (DR_IS_CONDITIONAL_IN_STMT): New macro.
22185 (create_data_ref): Add an is_conditional_in_stmt argument.
22186 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
22187 the is_conditional_in_stmt field.
22188 (data_ref_loc): Add an is_conditional_in_stmt field.
22189 (get_references_in_stmt): Set the is_conditional_in_stmt field.
22190 (find_data_references_in_stmt): Update call to create_data_ref.
22191 (graphite_find_data_references_in_stmt): Likewise.
22192 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
22193 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
22194 (vect_record_base_alignment): New function.
22195 (vect_record_base_alignments): Likewise.
22196 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
22197 for nested statements even if we fail to compute a misalignment.
22198 Use pooled base alignments for unconditional references.
22199 (vect_find_same_alignment_drs): Compare base addresses instead
22201 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
22202 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
22204 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22206 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
22207 Add an explicit name for the enum. Use auto_vec for slp_instances
22208 and grouped_stores.
22209 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
22211 (_bb_vec_info): Add a constructor and destructor.
22212 (vinfo_for_stmt): Return NULL for uids of -1 as well.
22213 (destroy_loop_vec_info): Delete.
22214 (vect_destroy_datarefs): Likewise.
22215 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
22216 (vec_info::vec_info): New function.
22217 (vec_info::~vec_info): Likewise.
22218 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
22219 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
22220 destroy_loop_vec_info.
22221 * tree-vect-loop.c (new_loop_vec_info): Replace with...
22222 (_loop_vec_info::_loop_vec_info): ...this.
22223 (destroy_loop_vec_info): Replace with...
22224 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
22225 the stmt_vec_infos. Leave handling of vec_info information to its
22226 destructor. Remove explicit vector releases.
22227 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
22228 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
22229 * tree-vect-slp.c (new_bb_vec_info): Replace with...
22230 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
22231 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
22232 (destroy_bb_vec_info): Replace with...
22233 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
22234 information to its destructor.
22235 (vect_slp_analyze_bb_1): Use new and delete instead of
22236 new_bb_vec_info and destroy_bb_vec_info.
22237 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
22240 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22242 * tree-data-ref.h (subscript): Add access_fn field.
22243 (data_dependence_relation): Add could_be_independent_p.
22244 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
22245 (same_access_functions): Move to tree-data-ref.c.
22246 * tree-data-ref.c (ref_contains_union_access_p): New function.
22247 (access_fn_component_p): Likewise.
22248 (access_fn_components_comparable_p): Likewise.
22249 (dr_analyze_indices): Add a reference to access_fn_component_p.
22250 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
22252 (constant_access_functions): Likewise.
22253 (add_other_self_distances): Likewise.
22254 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
22255 (initialize_data_dependence_relation): Use XCNEW and remove
22256 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
22257 of access functions that have the same type. Allow the
22258 subsequence to end with different bases in some circumstances.
22259 Record the chosen access functions in SUB_ACCESS_FN.
22260 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
22261 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
22262 (subscript_dependence_tester_1): Likewise dra and drb.
22263 (build_classic_dist_vector): Update calls accordingly.
22264 (subscript_dependence_tester): Likewise.
22265 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
22266 DDR_COULD_BE_INDEPENDENT_P.
22267 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
22268 comp_alias_ddrs instead of may_alias_ddrs.
22269 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
22271 (vect_analyze_data_ref_dependence): Use it if
22272 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
22273 distance vectors if that fails.
22274 (dependence_distance_ge_vf): New function.
22275 (vect_prune_runtime_alias_test_list): Use it. Don't clear
22276 LOOP_VINFO_MAY_ALIAS_DDRS.
22278 2017-08-04 Richard Biener <rguenther@suse.de>
22280 PR middle-end/81705
22281 * fold-const.c (fold_binary_loc): Properly restrict
22282 minus_var0 && minus_var1 case when associating undefined overflow
22285 2017-08-04 Tom de Vries <tom@codesourcery.com>
22287 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
22289 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22291 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22292 Don't start diagnostic messages with a capital letter.
22293 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22295 (rs6000_invalid_builtin): Likewise.
22296 (rs6000_trampoline_init): Likewise.
22298 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22301 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
22302 after setting changeable df flags.
22304 2017-08-03 Richard Biener <rguenther@suse.de>
22306 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
22307 up if the use is in USE - X.
22309 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
22311 * toplev.c (dumpfile.h): New include.
22312 (internal_error_reentered): New static function. Use it...
22313 (internal_error_function): ...here to handle reentered internal_error.
22315 2017-08-03 Richard Biener <rguenther@suse.de>
22317 PR middle-end/81148
22318 * fold-const.c (split_tree): Add minus_var and minus_con
22319 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
22320 here but always use minus_*.
22321 (associate_trees): Assert we never associate with MINUS_EXPR
22322 and NULL first operand. Do not recurse for PLUS_EXPR operands
22323 when associating as MINUS_EXPR either.
22324 (fold_binary_loc): Track minus_var and minus_con.
22326 2017-08-03 Tom de Vries <tom@codesourcery.com>
22329 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
22330 ACCEL_COMPILER, apply finish_options on
22331 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
22333 2017-08-03 Tom de Vries <tom@codesourcery.com>
22336 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
22337 function_entry_patch_area_size > 0.
22339 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22342 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
22343 instead of 10??LU, perform unit multiplication in wide_int,
22344 don't change alloc_object_size_limit if the limit is larger
22347 PR tree-optimization/81655
22348 PR tree-optimization/81588
22349 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
22350 the case when ranges[i].low and high are 1 for unsigned type with
22353 PR middle-end/81052
22354 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
22355 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
22357 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22359 * tree-vrp.h: Add include guard.
22361 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
22364 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
22365 (ud2): New insn pattern.
22366 * config/i386/i386.c (ix86_expand_epilogue):
22367 For naked functions, generate ud2 instead of trap insn.
22369 2017-08-02 Marek Polacek <polacek@redhat.com>
22372 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
22374 2017-08-02 Tom de Vries <tom@codesourcery.com>
22375 Cesar Philippidis <cesar@codesourcery.com>
22377 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
22378 Add missing edge probabilities.
22380 2017-08-02 Tamar Christina <tamar.christina@arm.com>
22382 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
22383 Correct endianness.
22385 2017-08-02 Jakub Jelinek <jakub@redhat.com>
22387 PR middle-end/79499
22388 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
22389 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
22390 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
22391 prologue_seq sequences - if any.
22393 2017-08-02 Richard Biener <rguenther@suse.de>
22395 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
22396 via vectors if supported, integer extracts via punning if supported
22397 or otherwise vector extracts.
22399 2017-08-02 Richard Biener <rguenther@suse.de>
22401 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
22403 (bitmap_insert_into_set): ... this.
22405 2017-08-02 Richard Biener <rguenther@suse.de>
22407 PR tree-optimization/81633
22409 2015-08-17 Alan Hayward <alan.hayward@arm.com>
22411 PR tree-optimization/71752
22412 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
22414 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
22416 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
22417 (machine_function::call_ms2sysv_pad_out): Remove field.
22418 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
22419 (ix86_compute_frame_layout): Likewise.
22421 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
22424 * config/i386/i386.c (ix86_set_func_type): Disallow naked
22425 attribute with interrupt attribute.
22427 2017-08-01 Andrew Pinski <apinski@cavium.com>
22429 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
22430 BIT_INSERT_EXPR's operand 1
22431 to see if the types precision matches.
22433 2017-08-01 Martin Liska <mliska@suse.cz>
22435 PR middle-end/70140
22436 * builtins.c (expand_builtin_memcpy_args): Remove.
22437 (expand_builtin_memcpy): Call newly added function
22438 expand_builtin_memory_copy_args.
22439 (expand_builtin_memcpy_with_bounds): Likewise.
22440 (expand_builtin_mempcpy): Remove last argument.
22441 (expand_builtin_mempcpy_with_bounds): Likewise.
22442 (expand_builtin_memory_copy_args): New function created from
22443 expand_builtin_mempcpy_args with small modifications.
22444 (expand_builtin_mempcpy_args): Remove.
22445 (expand_builtin_stpcpy): Remove unused argument.
22446 (expand_builtin): Likewise.
22447 (expand_builtin_with_bounds): Likewise.
22449 2017-08-01 Martin Liska <mliska@suse.cz>
22452 Make mempcpy more optimal (PR middle-end/70140).
22454 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22457 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
22458 __builtin_vec_cmpne verify both arguments are compatible vectors
22459 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
22460 verify arg1_type is a pointer or array type. For __builtin_vec_st,
22461 move computation of aligned to after checking the argument types.
22465 * config/rs6000/vsx.md (vextract_fp_from_shorth,
22466 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
22469 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
22470 Doug Rupp <rupp@adacore.com>
22471 Olivier Hainque <hainque@adacore.com>
22473 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
22474 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
22476 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
22477 for TARGET_OS_CPP_BUILTIN.
22478 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
22479 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
22481 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
22482 passing required abi options to the assembler for EABI configurations.
22483 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
22484 of .text.hot and .text.unlikely sections for kernel modules when
22485 using ARM style exceptions.
22486 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
22487 options. Add EXTRA_CC1_SPEC.
22488 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
22490 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
22492 (ARM_TARGET2_DWARF_FORMAT): Define.
22493 * config/arm/t-vxworks: Adjust multilib control to removal of the
22494 Diab command line options.
22496 2017-08-01 Martin Liska <mliska@suse.cz>
22498 PR gcov-profile/81561
22499 * gcov.c (unblock): Make unblocking safe as we need to preserve
22500 index correspondence of blocks and block_lists.
22502 2017-08-01 Richard Biener <rguenther@suse.de>
22504 PR tree-optimization/81181
22505 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
22506 (compute_antic): ... end of iteration here.
22508 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
22510 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
22511 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
22512 (ftree-slp-vectorize): Likewise.
22513 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
22514 can no longer be set independent of flag_tree_loop_vectorize.
22515 * omp-general.c (emp_max_vf): Likewise.
22516 * opts.c (enable_fdo_optimizations): Remove references to
22517 flag_tree_vectorize, these are now implicit.
22518 (common_handle_option): Remove handling for OPT_ftree_vectorize,
22519 and leave it for the options machinery.
22521 2017-08-01 Martin Liska <mliska@suse.cz>
22523 PR middle-end/70140
22524 * builtins.c (expand_builtin_memcpy_args): Remove.
22525 (expand_builtin_memcpy): Call newly added function
22526 expand_builtin_memory_copy_args.
22527 (expand_builtin_memcpy_with_bounds): Likewise.
22528 (expand_builtin_mempcpy): Remove last argument.
22529 (expand_builtin_mempcpy_with_bounds): Likewise.
22530 (expand_builtin_memory_copy_args): New function created from
22531 expand_builtin_mempcpy_args with small modifications.
22532 (expand_builtin_mempcpy_args): Remove.
22533 (expand_builtin_stpcpy): Remove unused argument.
22534 (expand_builtin): Likewise.
22535 (expand_builtin_with_bounds): Likewise.
22537 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22540 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
22541 print "ds:" only for immediates in generic address space.
22543 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22546 * config/i386/i386.c (ix86_funciton_naked): New prototype.
22547 (ix86_function_ok_for_sibcall): Return false for naked functions.
22549 2017-08-01 Richard Biener <rguenther@suse.de>
22551 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
22552 (compute_antic): Seed worklist with exit block predecessors.
22553 * cfganal.c (dfs_find_deadend): For a cycle return the source
22554 of the edge closing it.
22556 2017-08-01 Tamar Christina <tamar.christina@arm.com>
22558 * config/aarch64/aarch64.c
22559 (aarch64_can_const_movi_rtx_p): Move 0 check.
22561 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22563 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
22564 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
22566 * match.pd: Ditto in address comparison pattern.
22568 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22570 PR tree-optimization/81627
22571 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
22572 closed ssa form for store-store chain.
22574 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22576 PR tree-optimization/81620
22577 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
22578 for store-store chain.
22580 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22582 PR tree-optimization/81588
22583 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
22584 ranges[i].in_p, invert comparison code ccode. For >/>=,
22585 swap rhs1 and rhs2 and comparison code unconditionally,
22586 for </<= don't do that. Don't swap rhs1/rhs2 again if
22587 ranges[i].in_p, instead invert comparison code ccode if
22588 opcode or oe->rank is BIT_IOR_EXPR.
22591 * optabs.def (vec_extract_optab, vec_init_optab): Change from
22592 a direct optab to conversion optab.
22593 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
22594 with GET_MODE_INNER as last argument instead of optab_handler.
22595 * expmed.c (extract_bit_field_1): Likewise. Use vector from
22596 vector extraction if possible and optab is available.
22597 * expr.c (store_constructor): Use convert_optab_handler instead
22598 of optab_handler. Use vector initialization from smaller
22599 vectors if possible and optab is available.
22600 * tree-vect-stmts.c (vectorizable_load): Likewise.
22601 * doc/md.texi (vec_extract, vec_init): Document that the optabs
22602 now have two modes.
22603 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
22604 of vec_init from half-sized vectors with the same element mode.
22605 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
22606 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
22607 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
22608 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
22609 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
22610 after mode in gen_vec_extract* calls.
22611 (vec_extract<mode>): Renamed to ...
22612 (vec_extract<mode><ssescalarmodelower>): ... this.
22613 (vec_extract<mode><ssehalfvecmodelower>): New expander.
22614 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
22615 element mode after mode in gen_vec_init* calls.
22616 (VEC_INIT_HALF_MODE): New mode iterator.
22617 (vec_init<mode>): Renamed to ...
22618 (vec_init<mode><ssescalarmodelower>): ... this.
22619 (vec_init<mode><ssehalfvecmodelower>): New expander.
22620 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
22621 (vec_extractv2sfsf): ... this.
22622 (vec_initv2sf): Renamed to ...
22623 (vec_initv2sfsf): ... this.
22624 (vec_extractv2si): Renamed to ...
22625 (vec_extractv2sisi): ... this.
22626 (vec_initv2si): Renamed to ...
22627 (vec_initv2sisi): ... this.
22628 (vec_extractv4hi): Renamed to ...
22629 (vec_extractv4hihi): ... this.
22630 (vec_initv4hi): Renamed to ...
22631 (vec_initv4hihi): ... this.
22632 (vec_extractv8qi): Renamed to ...
22633 (vec_extractv8qiqi): ... this.
22634 (vec_initv8qi): Renamed to ...
22635 (vec_initv8qiqi): ... this.
22636 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
22637 (vec_init<mode>): Renamed to ...
22638 (vec_init<mode><VEC_base_l>): ... this.
22639 (vec_extract<mode>): Renamed to ...
22640 (vec_extract<mode><VEC_base_l>): ... this.
22641 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
22642 (vec_initv2sfsf): ... this.
22643 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
22644 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22645 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
22646 element mode after mode in gen_vec_init* calls.
22647 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
22648 (vec_init<mode><Vel>): ... this.
22649 (vec_extract<mode>): Renamed to ...
22650 (vec_extract<mode><Vel>): ... this.
22651 * config/aarch64/iterators.md (Vel): New mode attribute.
22652 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
22653 Add element mode after mode in gen_vec_extract* calls.
22654 * config/s390/vector.md (non_vec_l): New mode attribute.
22655 (vec_extract<mode>): Renamed to ...
22656 (vec_extract<mode><non_vec_l>): ... this.
22657 (vec_init<mode>): Renamed to ...
22658 (vec_init<mode><non_vec_l>): ... this.
22659 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
22660 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
22662 * config/arm/iterators.md (V_elem_l): New mode attribute.
22663 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
22664 (vec_extract<mode><V_elem_l>): ... this.
22665 (vec_extractv2di): Renamed to ...
22666 (vec_extractv2didi): ... this.
22667 (vec_init<mode>): Renamed to ...
22668 (vec_init<mode><V_elem_l>): ... this.
22669 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
22670 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
22671 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
22672 Add element mode after gen_vec_extract* calls.
22673 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
22674 (vec_init<mode><unitmode>): ... this.
22675 (vec_extract<mode>): Renamed to ...
22676 (vec_extract<mode><unitmode>): ... this.
22677 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
22678 (vec_init<mode><unitmode>): ... this.
22679 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
22680 (vec_initv2sfsf): ... this.
22681 (vec_extractv2sf): Renamed to ...
22682 (vec_extractv2sfsf): ... this.
22683 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
22684 Add element mode after gen_vec_extract* calls.
22685 * config/mips/mips.md (unitmode): New mode iterator.
22686 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
22687 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
22688 * config/spu/spu.md (inner_l): New mode attribute.
22689 (vec_init<mode>): Renamed to ...
22690 (vec_init<mode><inner_l>): ... this.
22691 (vec_extract<mode>): Renamed to ...
22692 (vec_extract<mode><inner_l>): ... this.
22693 * config/sparc/sparc.md (veltmode): New mode iterator.
22694 (vec_init<VMALL:mode>): Renamed to ...
22695 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
22696 * config/ia64/vect.md (vec_initv2si): Renamed to ...
22697 (vec_initv2sisi): ... this.
22698 (vec_initv2sf): Renamed to ...
22699 (vec_initv2sfsf): ... this.
22700 (vec_extractv2sf): Renamed to ...
22701 (vec_extractv2sfsf): ... this.
22702 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
22703 (vec_init<mode>): Renamed to ...
22704 (vec_init<mode><VEC_base_l>): ... this.
22705 (vec_extract<mode>): Renamed to ...
22706 (vec_extract<mode><VEC_base_l>): ... this.
22707 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
22708 (vec_initv2sfsf): ... this.
22709 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
22710 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22711 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
22712 gen_vec_init* calls.
22714 2017-08-01 Richard Biener <rguenther@suse.de>
22716 PR tree-optimization/81297
22717 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
22718 TREE_OVERFLOW from INTEGER_CSTs.
22720 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
22722 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
22724 2017-07-31 Carl Love <cel@us.ibm.com>
22726 * config/rs6000/rs6000-c: Add support for built-in functions
22727 vector signed char vec_xl_be (signed long long, signed char *);
22728 vector unsigned char vec_xl_be (signed long long, unsigned char *);
22729 vector signed int vec_xl_be (signed long long, signed int *);
22730 vector unsigned int vec_xl_be (signed long long, unsigned int *);
22731 vector signed long long vec_xl_be (signed long long, signed long long *);
22732 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
22733 vector signed short vec_xl_be (signed long long, signed short *);
22734 vector unsigned short vec_xl_be (signed long long, unsigned short *);
22735 vector double vec_xl_be (signed long long, double *);
22736 vector float vec_xl_be (signed long long, float *);
22737 * config/rs6000/altivec.h (vec_xl_be): Add #define.
22738 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
22739 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
22741 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
22742 (altivec_expand_builtin): Add switch statement to call
22743 altivec_expand_xl_be for each builtin.
22744 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
22745 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
22746 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
22747 __builtin_vsx_le_be_v16qi.
22748 * doc/extend.texi: Update the built-in documentation file for the
22749 new built-in functions.
22751 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22754 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
22756 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22758 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22760 * config.gcc: Add z14.
22761 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
22762 CPU model numbers for z13s and z14.
22763 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
22765 * config/s390/s390-opts.h (enum processor_type): Rename
22766 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22767 * config/s390/s390.c (processor_table): Add field for CPU name to
22768 be passed to Binutils.
22769 (s390_asm_output_machine_for_arch): Use the new field in
22770 processor_table for Binutils.
22771 (s390_expand_builtin): Replace arch12 with z14.
22772 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22773 (s390_get_sched_attrmask): Likewise.
22774 (s390_get_unit_mask): Likewise.
22775 * config/s390/s390.opt: Add z14 to processor_type enum.
22777 2017-07-31 Martin Jambor <mjambor@suse.cz>
22780 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22781 regardless of optimization level.
22783 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22784 Martin Liska <mliska@suse.cz>
22786 * predict.def: Remove old comment and adjust probability.
22787 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22788 PREDICT statements.
22790 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22793 * config/i386/i386.c (ix86_function_naked): New function.
22794 (ix86_can_use_return_insn_p): Return false for naked functions.
22795 (ix86_expand_prologue): Skip prologue for naked functions.
22796 (ix86_expand_epilogue): Skip epilogue for naked functions
22797 and emit trap instruction.
22798 (ix86_warn_func_return): New function.
22799 (ix86_attribute_table): Add "naked" attribute specification.
22800 (TARGET_WARN_FUNC_RETURN): Define.
22801 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22803 2017-07-31 Martin Liska <mliska@suse.cz>
22805 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22806 (dump_gimple_bb_header): Always dump BB info.
22807 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22809 2017-07-31 Martin Liska <mliska@suse.cz>
22812 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22813 also with current_function_decl non-null equality.
22815 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22818 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22819 change type to the element type, instead add eltype variable and
22820 use it where we are interested in the element type.
22822 PR tree-optimization/81603
22823 * ipa-polymorphic-call.c
22824 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22825 offset arithmetic in offset_int, bail out if the resulting bit offset
22826 doesn't fit into shwi.
22828 2017-07-31 Martin Liska <mliska@suse.cz>
22830 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22831 (gimplify_save_expr): Fix comment.
22833 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22836 * config/i386/i386.c (ix86_function_arg): Update arguments for
22838 (ix86_compute_frame_layout): Set the initial stack offset to
22839 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22840 INCOMING_FRAME_SP_OFFSET.
22841 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22842 stack before exception handler returns.
22843 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22844 the 'ERROR_CODE' for exception handler.
22846 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22848 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22849 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22850 (ASM_OUTPUT_REG_POP): Ditto.
22851 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22852 instead of asm_fprintf to output pure string.
22854 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22856 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22857 to imported_module_or_decl hook.
22858 (debug_nothing_tree_tree_tree_bool): Remove.
22859 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22860 * debug.c (do_nothing_debug_hooks): Use
22861 debug_nothing_tree_tree_tree_bool_bool instead of
22862 debug_nothing_tree_tree_tree_bool.
22863 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22864 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22865 * sdbout.c (sdb_debug_hooks): Likewise.
22866 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22867 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22869 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22870 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22871 attribute, don't add anything.
22873 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22875 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22876 (fold_build2_stat_loc): Likewise.
22877 (fold_build3_stat_loc): Likewise.
22878 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22879 (fold_build1_loc): Remove macro.
22880 (fold_build2_loc): Likewise.
22881 (fold_build3_loc): Likewise.
22883 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22885 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22886 (gimple_build_debug_bind_source_stat): Likewise.
22887 * gimple.h (gimple_build_debug_bind): Remove macro.
22888 (gimple_build_debug_bind_source): Likewise.
22890 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22892 * bitmap.c (bitmap_alloc): Adjust.
22893 (bitmap_gc_alloc): Likewise.
22894 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22896 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22898 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22899 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22900 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22901 (bitmap_gc_alloc_stat): Likewise.
22902 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22904 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22906 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22907 * rtl.h (shallow_copy_rtx): Remove macro.
22909 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22911 * emit-rtl.c (gen_raw_REG): Adjust.
22912 * gengenrtl.c (gendef): Likewise.
22913 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22914 * rtl.h (rtx_alloc): Remove macro.
22916 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22918 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22919 (build_tree_list_stat): Likewise.
22920 * tree.h (build_tree_list): Remove macro.
22921 (build_tree_list_vec): Likewise.
22923 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22925 * tree.c (make_vector_stat): Remove _stat from name.
22926 (build_vector_stat): Likewise.
22927 * tree.h (make_vector_stat): Remove macro.
22928 (build_vector_stat): Likewise.
22930 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22932 * tree.h (build_var_debug_value): Remove prototype.
22934 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22936 * tree.c (tree_cons_stat): Remove _stat from name.
22937 * tree.h (tree_cons): Remove macro.
22939 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22941 * tree.c (build_vl_exp_stat): Remove _stat from name.
22942 * tree.h (build_vl_exp): Remove macro.
22944 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22946 * tree.c (build_decl_stat): Remove _stat from name.
22947 * tree.h (build_decl): Remove macro.
22949 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22951 * gimple.c (gimple_build_with_ops_stat): Adjust.
22952 (gimple_alloc_stat): Remove _stat from name.
22953 * gimple.h (gimple_alloc): Remove macro.
22955 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22957 * tree.c (make_tree_vec_stat): Remove _stat from name.
22958 (grow_tree_vec_stat): Likewise.
22959 * tree.h (make_tree_vec_stat): Adjust prototype.
22960 (grow_tree_vec_stat): Likewise.
22961 (make_tree_vec): Remove macro.
22962 (grow_tree_vec): Likewise.
22964 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22966 * fold-const.c (fold_build1_stat_loc): Adjust.
22967 (fold_build2_stat_loc): Likewise.
22968 (fold_build3_stat_loc): Likewise.
22969 * tree.c (build0_stat): Remove _stat from name.
22970 (build1_stat): Likewise.
22971 (build2_stat): Likewise.
22972 (build3_stat): Likewise.
22973 (build4_stat): Likewise.
22974 (build5_stat): Likewise.
22975 * tree.h (build1_loc): Remove macro, and rename _stat function
22977 (build2_loc): Likewise.
22978 (build3_loc): Likewise.
22979 (build4_loc): Likewise.
22980 (build5_loc): Likewise.
22982 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22984 * tree.c (make_int_cst_stat): Remove _stat from name.
22985 * tree.h (make_int_cst_stat): Adjust prototype.
22986 (make_int_cst): Remove macro.
22988 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22990 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22991 * tree.h (make_tree_binfo_stat): Adjust prototype.
22992 (make_tree_binfo): Remove.
22994 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22996 * tree.c (copy_node_stat): Rename to copy_node.
22997 (build_distinct_type_copy): Adjust.
22998 * tree.h (copy_node_stat): Adjust prototype.
22999 (copy_node): Remove macro.
23001 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23003 * tree.c (make_node_stat): rename to make_node.
23004 (build_tree_list_stat): Adjust.
23005 (build0_stat): Likewise.
23006 (build2_stat): Likewise.
23007 (build3_stat): Likewise.
23008 (build4_stat): Likewise.
23009 (build5_stat): Likewise.
23010 (build_decl_stat): Likewise.
23011 * tree.h (make_node_stat): Adjust prototype.
23012 (make_node): remove macro.
23014 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
23016 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
23017 (PPC_FEATURE2_SCV): Likewise.
23018 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
23020 2017-07-28 Tamar Christina <tamar.christina@arm.com>
23022 * config/aarch64/aarch64.c
23023 (aarch64_internal_mov_immediate): Add new special pattern.
23024 * config/aarch64/aarch64.md (*movdi_aarch64):
23025 Add reg/32bit const mov case.
23027 2017-07-28 Tamar Christina <tamar.christina@arm.com>
23028 Richard Sandiford <richard.sandiford@linaro.org>
23030 * config/aarch64/aarch64.md (mov<mode>): Generalize.
23031 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
23032 Add integer and movi cases.
23033 (movi-split-hf-df-sf split, fp16): New.
23034 (enabled): Added TARGET_FP_F16INST.
23035 * config/aarch64/iterators.md (GPF_HF): New.
23036 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
23038 2017-07-28 Tamar Christina <tamar.christina@arm.com>
23040 * config/aarch64/aarch64.c
23041 (aarch64_simd_container_mode): Add prototype.
23042 (aarch64_expand_mov_immediate): Add HI support.
23043 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
23044 (aarch64_can_const_movi_rtx_p): New.
23045 (aarch64_preferred_reload_class):
23046 Remove restrictions of using FP registers for certain SIMD operations.
23047 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
23048 (aarch64_valid_floating_const): Add integer move validation.
23049 (aarch64_simd_imm_scalar_p): Remove.
23050 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
23051 (aarch64_legitimate_constant_p): Expand list of supported cases.
23052 * config/aarch64/aarch64-protos.h
23053 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
23054 (aarch64_reinterpret_float_as_int): New.
23055 (aarch64_simd_imm_scalar_p): Remove.
23056 * config/aarch64/constraints.md (Uvi): New.
23057 (Dd): Split into Ds and new Dd.
23058 * config/aarch64/aarch64.md (*movsi_aarch64):
23060 (*movdi_aarch64): Add SIMD mov case.
23062 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23064 * tree-predcom.c: (struct chain): Handle store-store chain in which
23065 stores for elimination only store loop invariant values.
23066 (execute_pred_commoning_chain): Ditto.
23067 (prepare_initializers_chain_store_elim): Ditto.
23068 (prepare_finalizers): Ditto.
23069 (is_inv_store_elimination_chain): New function.
23070 (initialize_root_vars_store_elim_1): New function.
23072 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23074 * tree-predcom.c: Revise general description of the pass.
23075 (enum chain_type): New enum type for store elimination.
23076 (struct chain): New field supporting store elimination.
23077 (struct component): Ditto.
23078 (dump_chain): Dump store-stores chain.
23079 (release_chain): Release resources.
23080 (split_data_refs_to_components): Compute and create component
23081 contains only stores for elimination.
23082 (get_chain_last_ref_at): New function.
23083 (make_invariant_chain): Initialization.
23084 (make_rooted_chain): Specify chain type in parameter and record it.
23085 (add_looparound_copies): Skip for store-stores chain.
23086 (determine_roots_comp): Compute type of chain and pass it to
23088 (initialize_root_vars_store_elim_2): New function.
23089 (finalize_eliminated_stores): New function.
23090 (remove_stmt): Handle store for elimination.
23091 (execute_pred_commoning_chain): Execute predictive commoning on
23092 store-store chains.
23093 (determine_unroll_factor): Skip unroll for store-stores chain.
23094 (prepare_initializers_chain_store_elim): New function.
23095 (prepare_initializers_chain): Hanlde store-store chain.
23096 (prepare_finalizers_chain, prepare_finalizers): New function.
23097 (tree_predictive_commoning_loop): Return integer value indicating
23098 if loop is unrolled or lcssa form is corrupted.
23099 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
23101 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23103 * tree-predcom.c (initialize_root): Delete.
23104 (execute_pred_commoning_chain): Initialize root vars and replace
23105 reference of non-combined chain directly, rather than call above
23108 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23110 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
23111 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
23113 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23115 * tree-predcom.c (struct chain): New field init_seq.
23116 (release_chain): Release init_seq.
23117 (prepare_initializers_chain): Record intialization stmts in above
23119 (insert_init_seqs): New function.
23120 (tree_predictive_commoning_loop): Call insert_init_seqs.
23122 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23124 * tree-predcom.c (determine_roots_comp): Skip trivial components.
23126 2017-07-28 Richard Biener <rguenther@suse.de>
23128 * match.pd: Remove superfluous :c.
23129 * genmatch.c (simplify::id): Add member.
23130 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
23132 (current_id): New global.
23133 (dt_node::parent): Move from ...
23134 (dt_operand::parent): ... here. Add for_id member.
23135 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
23136 (decision_tree::find_node): Relax order requirement when
23137 merging DT_TRUE nodes to ones inbetween the current simplify
23138 and the one we try to merge with. Add diagnostic whenever
23139 we need to enforce pattern order by not merging.
23140 (decision_tree::insert): Set current_id.
23141 (decision_tree::print_node): Dump parent node and for_id.
23142 (parser::last_id): Add member.
23143 (parser::push_simplify): Assign unique id.
23144 (parser::parser): Initialize last_id.
23146 2017-07-28 Martin Liska <mliska@suse.cz>
23149 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
23150 gimple_build_debug_bind.
23152 2017-07-28 Richard Biener <rguenther@suse.de>
23154 PR tree-optimization/81502
23155 * match.pd: Add pattern combining BIT_INSERT_EXPR with
23157 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
23159 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
23160 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
23161 for BIT_FIELD_REF args.
23162 * fold-const.c (make_bit_field_ref): Likewise.
23163 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
23165 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23168 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
23169 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
23170 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
23171 Or it into SANITIZER_UNDEFINED.
23172 * ubsan.c: Include gimple-fold.h and varasm.h.
23173 (ubsan_expand_ptr_ifn): New function.
23174 (instrument_pointer_overflow): New function.
23175 (maybe_instrument_pointer_overflow): New function.
23176 (instrument_object_size): Formatting fix.
23177 (pass_ubsan::execute): Call instrument_pointer_overflow
23178 and maybe_instrument_pointer_overflow.
23179 * internal-fn.c (expand_UBSAN_PTR): New function.
23180 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
23181 * sanitizer.def (__ubsan_handle_pointer_overflow,
23182 __ubsan_handle_pointer_overflow_abort): New builtins.
23183 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
23184 * internal-fn.def (UBSAN_PTR): New internal function.
23185 * opts.c (sanitizer_opts): Add pointer-overflow.
23186 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
23187 * fold-const.c (build_range_check): Compute pointer range check in
23188 integral type if pointer arithmetics would be needed. Formatting
23191 2017-07-28 Martin Liska <mliska@suse.cz>
23194 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
23195 parameters that are of a variable-length.
23197 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
23199 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
23201 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
23202 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
23203 (CRT_CALL_STATIC_FUNCTION): Likewise.
23204 (ASM_DEFAULT_SPEC): New define.
23205 (ASM_SPEC32): Likewise.
23206 (ASM_SPEC64): Likewise.
23207 (ASM_SPEC_COMMON): Likewise.
23208 (ASM_SPEC): Likewise.
23209 (INVALID_64BIT): Likewise.
23210 (LINK_OS_DEFAULT_SPEC): Likewise.
23211 (LINK_OS_SPEC32): Likewise.
23212 (LINK_OS_SPEC64): Likewise.
23213 (POWERPC_LINUX): Likewise.
23214 (PTRDIFF_TYPE): Likewise.
23215 (RESTORE_FP_PREFIX): Likewise.
23216 (RESTORE_FP_SUFFIX): Likewise.
23217 (SAVE_FP_PREFIX): Likewise.
23218 (SAVE_FP_SUFFIX): Likewise.
23219 (SIZE_TYPE): Likewise.
23220 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
23221 (TARGET_64BIT): Likewise.
23222 (TARGET_64BIT): Likewise.
23223 (TARGET_AIX): Likewise.
23224 (WCHAR_TYPE_SIZE): Likewise.
23225 (WCHAR_TYPE): Undefine.
23226 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
23227 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
23228 (CPP_OS_RTEMS_SPEC): Delete.
23229 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
23230 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
23232 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
23234 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23236 PR tree-optimization/81578
23237 * tree-parloops.c (build_new_reduction): Bail out if
23238 reduction_code isn't one of the standard OpenMP reductions.
23239 Move the details printing after that decision.
23241 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23243 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
23244 related to reload_in_progress.
23245 (splat_input_operand): Likewise.
23246 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
23248 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
23250 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
23251 (TARGET_INSTANTIATE_DECLS): Likewise.
23252 (legitimate_indexed_address_p): Delete reload_in_progress code.
23253 (rs6000_debug_legitimate_address_p): Likewise.
23254 (rs6000_eliminate_indexed_memrefs): Likewise.
23255 (rs6000_emit_le_vsx_store): Likewise.
23256 (rs6000_emit_move_si_sf_subreg): Likewise.
23257 (rs6000_emit_move): Likewise.
23258 (register_to_reg_type): Likewise.
23259 (rs6000_pre_atomic_barrier): Likewise.
23260 (rs6000_machopic_legitimize_pic_address): Likewise.
23261 (rs6000_allocate_stack_temp): Likewise.
23262 (rs6000_address_for_fpconvert): Likewise.
23263 (rs6000_address_for_altivec): Likewise.
23264 (rs6000_secondary_memory_needed_rtx): Delete function.
23265 (rs6000_check_sdmode): Likewise.
23266 (rs6000_alloc_sdmode_stack_slot): Likewise.
23267 (rs6000_instantiate_decls): Likewise.
23268 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
23269 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
23270 Delete reload_in_progress.
23271 (*vec_reload_and_plus_<mptrsize>): Likewise.
23272 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
23273 (vsx_div_v2di): Likewise.
23274 (vsx_udiv_v2di): Likewise.
23276 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23278 * config/rs6000/rs6000.opt (mlra): Replace with stub.
23279 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
23280 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
23281 (rs6000_debug_reg_global): Delete print of LRA status.
23282 (rs6000_option_override_internal): Delete dead LRA related code.
23283 (rs6000_lra_p): Delete function.
23284 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
23286 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
23288 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
23289 * config/riscv/rtems.h: New file.
23291 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23292 Sudakshina Das <sudi.das@arm.com>
23294 * config/aarch64/aarch64.md
23295 (define_split for and<mode>3nr_compare): Move
23296 non aarch64_logical_operand to a register.
23297 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
23298 register immediate operand to a register.
23299 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
23301 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23303 PR middle-end/81564
23304 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
23306 2017-07-27 Richard Biener <rguenther@suse.de>
23308 PR tree-optimization/81573
23309 PR tree-optimization/81494
23310 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
23311 multi defuse cycle case.
23313 2017-07-27 Richard Biener <rguenther@suse.de>
23315 PR tree-optimization/81571
23316 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
23319 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
23321 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
23322 earlier and only if MASK_FPU is set. Adjust formatting.
23324 2017-07-27 Martin Liska <mliska@suse.cz>
23326 * opt-functions.awk: Add validation of value of Init.
23327 * optc-gen.awk: Pass new argument.
23329 2017-07-27 Martin Liska <mliska@suse.cz>
23331 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
23332 Fix wrong condition.
23334 2017-07-27 Martin Liska <mliska@suse.cz>
23336 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
23337 BBs and edges seen by autoFDO.
23339 2017-07-27 Richard Biener <rguenther@suse.de>
23341 PR tree-optimization/81502
23342 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
23343 with incompatible but same sized type.
23344 (execute_update_addresses_taken): Likewise.
23346 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
23348 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
23349 flag_tree_loop_vectorize rather than flag_tree_vectorize.
23351 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23354 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
23355 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
23356 Change s_operand to memory_operand.
23358 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
23360 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
23361 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
23362 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
23363 Emit instructions rather than returning an expression. Handle TFmode
23364 and KFmode by casting to TImode.
23365 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
23366 (rs6000_emit_le_vsx_store): Likewise.
23367 * config/rs6000/vsx.md (VSX_TI): New iterator.
23368 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
23369 (*vsx_le_undo_permute_<mode>): Likewise.
23370 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
23371 emit the split sequence.
23372 (*vsx_le_perm_store_<mode>): Likewise.
23374 2017-07-27 Jakub Jelinek <jakub@redhat.com>
23376 PR tree-optimization/81555
23377 PR tree-optimization/81556
23378 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
23379 if true, force CHANGED for the recursive invocation.
23380 (reassociate_bb): Remember original length of ops array, pass
23381 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
23383 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
23384 attributes for noipa attribute. For naked attribute use
23385 lookup_attribute first before lookup_attribute_spec.
23386 * final.c (rest_of_handle_final): Disable IPA RA for functions with
23388 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
23389 for functions with noipa attribute.
23390 (cgraph_externally_visible_p): Return true for functions with noipa
23392 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
23393 for functions with noipa attribute.
23394 * doc/extend.texi: Document noipa function attribute.
23395 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
23396 also for functions with noipa attribute.
23397 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
23399 2017-07-26 Andrew Pinski <apinski@cavium.com>
23401 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
23402 vec_unalign_load_cost and vec_unalign_store_cost.
23404 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
23406 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
23407 -mvsx-small-integer option.
23408 (ISA_3_0_MASKS_IEEE): Likewise.
23409 (OTHER_VSX_VECTOR_MASKS): Likewise.
23410 (POWERPC_MASKS): Likewise.
23411 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
23412 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
23413 code, only testing for DImode being allowed in non-VSX floating
23415 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
23416 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
23418 (rs6000_option_override_internal): Delete -mvsx-small-integer.
23419 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
23420 TARGET_P8_VECTOR test.
23421 (rs6000_secondary_reload_simple_move): Likewise.
23422 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
23423 since TARGET_P9_VECTOR was already tested.
23424 (rs6000_opt_masks): Remove -mvsx-small-integer.
23425 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
23426 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23428 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
23429 test for TARGET_VEXTRACTUB was used, and that uses
23431 (p9 extract splitter): Likewise.
23432 (vsx_extract_<mode>_di_p9): Likewise.
23433 (vsx_extract_<mode>_store_p9): Likewise.
23434 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
23435 for TARGET_P9_VECTOR was used. Delete code that is now dead with
23436 the elimination of TARGET_VSX_SMALL_INTEGER.
23437 (vsx_extract_<mode>_p8): Likewise.
23438 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
23439 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
23440 (vsx_set_<mode>_p9): Likewise.
23441 (vsx_set_v4sf_p9): Likewise.
23442 (vsx_set_v4sf_p9_zero): Likewise.
23443 (vsx_insert_extract_v4sf_p9): Likewise.
23444 (vsx_insert_extract_v4sf_p9_2): Likewise.
23445 * config/rs6000/rs6000.md (sign extend splitter): Change
23446 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
23447 (floatsi<mode>2_lfiwax_mem): Likewise.
23448 (floatunssi<mode>2_lfiwzx_mem): Likewise.
23449 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
23450 since a test for TARGET_P9_VECTOR was used.
23451 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23452 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
23453 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23454 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23455 TARGET_P8_VECTOR test.
23456 (fix_trunc<mode>si2_stfiwx): Likewise.
23457 (fix_trunc<mode>si2_internal): Likewise.
23458 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
23459 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23461 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23462 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23463 TARGET_P8_VECTOR test.
23464 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23465 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
23466 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23468 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23469 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
23470 since a test for TARGET_P9_VECTOR was used.
23471 (splitter for loading small constants): Likewise.
23473 2017-07-26 Andrew Pinski <apinski@cavium.com>
23475 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
23478 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
23481 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
23482 (fp_valid_at): Likewise.
23484 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23486 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
23487 (qdf24xx_addrcost_table): Likewise.
23488 (cortexa57_tunings): Update to use generic_branch_cost.
23489 (cortexa72_tunings): Likewise.
23490 (cortexa73_tunings): Likewise.
23491 (qdf24xx_tunings): Likewise.
23493 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23495 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
23496 (thunderx2t99_branch_cost): Likewise.
23497 (cortexa35_tunings): Update to use generic_branch_cost.
23498 (cortexa53_tunings): Likewise.
23499 (cortexa57_tunings): Likewise.
23500 (cortexa72_tunings): Likewise.
23501 (cortexa73_tunings): Likewise.
23502 (thunderx2t99_tunings): Likewise.
23504 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23506 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
23507 (sparc_option_override): Honour MASK_FSMULD.
23508 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
23509 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
23510 * config/sparc/sparc.opt (mfsmuld): New option.
23511 * doc/invoke.texi (mfsmuld): Document option.
23513 2017-07-26 Marek Polacek <polacek@redhat.com>
23515 PR middle-end/70992
23516 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
23518 2017-07-26 Richard Biener <rguenther@suse.de>
23520 * gimple-match-head.c (do_valueize): Return OP if valueize
23522 (get_def): New helper to get at the def stmt of a SSA name
23523 if valueize allows.
23524 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
23525 do_valueize to get at the def stmt.
23526 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
23528 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
23530 PR middle-end/46932
23531 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
23533 2017-07-26 Martin Liska <mliska@suse.cz>
23536 * function.c (expand_function_start): Make expansion of
23537 nonlocal_goto_save_area after parm_birth_insn.
23539 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23541 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
23542 from all CPU target flags enable members.
23544 2017-07-26 Richard Biener <rguenther@suse.de>
23546 * genmatch.c (dt_simplify::gen): Make iterator vars const.
23547 (decision_tree::gen): Make 'type' const.
23548 (write_predicate): Likewise.
23550 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23552 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
23553 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
23554 (rs6000_option_override_internal): Likewise.
23555 (rs6000_expand_vector_set): Likewise.
23556 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
23557 (TARGET_UPPER_REGS_SF): Likewise.
23558 (TARGET_UPPER_REGS_DI): Likewise.
23559 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
23560 (TARGET_DIRECT_MOVE_64BIT): Likewise.
23561 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
23562 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23563 (Splitters for DI constants in Altivec registers): Likewise.
23564 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
23565 (vsx_set_v4sf_p9): Likewise.
23566 (vsx_set_v4sf_p9_zero): Likewise.
23567 (vsx_insert_extract_v4sf_p9): Likewise.
23568 (vsx_insert_extract_v4sf_p9_2): Likewise.
23570 2017-07-25 Carl Love <cel@us.ibm.com>
23572 * doc/extend.texi: Update the built-in documentation file for the
23573 existing built-in functions
23574 vector signed char vec_cnttz (vector signed char);
23575 vector unsigned char vec_cnttz (vector unsigned char);
23576 vector signed short vec_cnttz (vector signed short);
23577 vector unsigned short vec_cnttz (vector unsigned short);
23578 vector signed int vec_cnttz (vector signed int);
23579 vector unsigned int vec_cnttz (vector unsigned int);
23580 vector signed long long vec_cnttz (vector signed long long);
23581 vector unsigned long long vec_cnttz (vector unsigned long long);
23583 2017-07-25 Andrew Pinski <apinski@cavium.com>
23585 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
23586 accesses where the use is for the first operand of a BIT_INSERT.
23588 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23591 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23592 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23594 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23596 * config/i386/gstabs.h: Delete.
23597 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
23599 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
23601 * config/i386/i386.c (ix86_decompose_address): Do not check for
23602 register RTX when looking at index_reg or base_reg.
23603 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
23605 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
23607 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
23608 to update EH info here.
23610 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23612 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
23614 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23616 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
23618 2017-07-25 Torsten Duwe <duwe@suse.de>
23620 * common.opt: Introduce -fpatchable-function-entry
23621 command line option, and its variables function_entry_patch_area_size
23622 and function_entry_patch_area_start.
23623 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
23624 including a two-value parser.
23625 * target.def (print_patchable_function_entry): New target hook.
23626 * targhooks.h (default_print_patchable_function_entry): New function.
23627 * targhooks.c (default_print_patchable_function_entry): Likewise.
23628 * toplev.c (process_options): Switch off IPA-RA if
23629 patchable function entries are being generated.
23630 * varasm.c (assemble_start_function): Look at the
23631 patchable-function-entry command line switch and current
23632 function attributes and maybe generate NOP instructions by
23633 calling the print_patchable_function_entry hook.
23634 * doc/extend.texi: Document patchable_function_entry attribute.
23635 * doc/invoke.texi: Document -fpatchable_function_entry
23636 command line option.
23637 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
23639 * doc/tm.texi: Re-generate.
23641 2017-07-25 Jakub Jelinek <jakub@redhat.com>
23644 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
23645 TARGET_AVX512DQ rather than TARGET_AVX512BW.
23647 2017-07-25 Tamar Christina <tamar.christina@arm.com>
23649 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
23651 2017-07-25 Richard Biener <rguenther@suse.de>
23653 PR tree-optimization/81455
23654 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
23655 not walk in cycles when looking for guards.
23657 2017-07-25 Richard Biener <rguenther@suse.de>
23659 PR tree-optimization/81529
23660 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
23661 when optimizing backedge uses.
23663 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
23665 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
23667 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
23668 to dl_section_ref. On AIX, append an expression to subtract
23669 the size of the section length to dl_section_ref.
23671 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
23673 * configure.ac: If any of the config.* scripts fail, exit 1.
23674 * configure: Regenerate.
23676 2017-07-25 Richard Biener <rguenther@suse.de>
23678 PR middle-end/81546
23679 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
23680 of immediate uses, be more verbose on errors.
23682 2017-07-25 Richard Biener <rguenther@suse.de>
23684 PR tree-optimization/81510
23685 * tree-vect-loop.c (vect_is_simple_reduction): When the
23686 reduction stmt is not inside the loop bail out.
23688 2017-07-25 Richard Biener <rguenther@suse.de>
23690 PR tree-optimization/81303
23691 * tree-vect-loop-manip.c (vect_loop_versioning): Build
23692 profitability check against LOOP_VINFO_NITERSM1.
23694 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23696 * domwalk.c (cmp_bb_postorder): Simplify.
23697 (sort_bbs_postorder): New function. Use it...
23698 (dom_walker::walk): ...here to optimize common cases.
23700 2017-07-25 Martin Liska <mliska@suse.cz>
23703 * ipa-visibility.c (function_and_variable_visibility): Make the
23704 redirection just on target that supports aliasing.
23705 Fix GNU coding style.
23707 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
23710 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
23712 * config/arm/rtems.h: Likewise.
23713 * config/bfin/rtems.h: Likewise.
23714 * config/i386/rtemself.h: Likewise.
23715 * config/lm32/rtems.h: Likewise.
23716 * config/m32c/rtems.h: Likewise.
23717 * config/m68k/rtemself.h: Likewise.
23718 * config/microblaze/rtems.h: Likewise.
23719 * config/mips/rtems.h: Likewise.
23720 * config/moxie/rtems.h: Likewise.
23721 * config/nios2/rtems.h: Likewise.
23722 * config/powerpcspe/rtems.h: Likewise.
23723 * config/rs6000/rtems.h: Likewise.
23724 * config/rtems.h: Likewise.
23725 * config/sh/rtems.h: Likewise.
23726 * config/sh/rtemself.h: Likewise.
23727 * config/sparc/rtemself.h: Likewise.
23729 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23732 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
23733 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
23734 * tree-ssa-structalias.c (alias_get_name): Same.
23736 2017-07-25 Bin Cheng <bin.cheng@arm.com>
23739 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
23740 instructions if no du chain is found.
23742 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23744 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
23746 2017-07-25 Richard Biener <rguenther@suse.de>
23748 PR middle-end/81505
23749 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
23752 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23754 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
23755 upper-regs options.
23756 (ISA_2_7_MASKS_SERVER): Likewise.
23757 (ISA_3_0_MASKS_IEEE): Likewise.
23758 (OTHER_P8_VECTOR_MASKS): Likewise.
23759 (OTHER_VSX_VECTOR_MASKS): Likewise.
23760 (POWERPC_MASKS): Likewise.
23761 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
23762 duplicate list of options.
23763 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
23764 explicit -mupper-regs options.
23765 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
23766 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
23767 alias for -mupper-regs-df.
23768 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
23769 (rs6000_init_hard_regno_mode_ok): Likewise.
23770 (rs6000_option_override_internal): Likewise.
23771 (rs6000_opt_masks): Likewise.
23772 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
23773 options in terms of whether -mvsx or -mpower8-vector was used.
23774 (TARGET_UPPER_REGS_DI): Likewise.
23775 (TARGET_UPPER_REGS_SF): Likewise.
23776 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23777 -mupper-regs-* options.
23779 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23781 * passes.c (emergency_dump_function): Print some empty lines and a
23782 header before the RTL dump.
23784 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23786 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23788 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23791 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23792 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23794 2017-07-24 Carl Love <cel@us.ibm.com>
23796 * config/rs6000/rs6000-c.c: Add support for built-in functions
23797 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23798 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23799 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23800 vec_extract_fp_from_shortl): Add defines for the two builtins.
23801 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23802 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23804 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23805 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23806 * doc/extend.texi: Update the built-in documentation file for the
23807 new built-in function.
23809 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23812 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23814 * doc/generic.texi: Likewise.
23815 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23816 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23818 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23820 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23821 (aarch64_mls_elt_merge<mode>): Likewise.
23823 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23825 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23826 having __cxa_atexit.
23828 2017-07-23 Michael Collison <michael.collison@arm.com>
23830 * config/arm/arm.c (arm_option_override): Deprecate
23831 use of -mstructure-size-boundary.
23832 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23833 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23835 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23838 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23839 Reduce cost estimate for direct moves.
23841 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23844 * config/i386/i386.c (ix86_option_override_internal): Disable
23845 BMI, BMI2 and TBM instructions for -m16.
23847 2017-07-21 Carl Love <cel@us.ibm.com>
23849 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23850 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23851 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23852 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23853 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23854 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23855 VMULOSW): New enum "unspec" values.
23856 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23857 altivec_vmulosw): New patterns.
23858 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23859 VMULOSW): Add definitions.
23861 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23863 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23864 (qdf24xx): Likewise.
23865 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23866 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23867 (AARCH64_FL_V8_1): Renumber.
23868 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23869 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23870 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23871 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23872 rdma to feature modifiers list.
23874 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23876 PR middle-end/56727
23877 * ipa-visibility (function_and_variable_visibility): Convert
23878 recursive PLT call to direct call if appropriate.
23880 2017-07-21 Andrew Pinski <apinski@cavium.com>
23882 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23883 operand 1 to see if the types precision matches.
23884 * fold-const.c (operand_equal_p): Likewise.
23886 2017-07-21 Richard Biener <rguenther@suse.de>
23888 PR tree-optimization/81303
23889 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23890 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23891 (vect_peeling_hash_get_lowest_cost): Adjust.
23892 (vect_enhance_data_refs_alignment): Likewise. Use
23893 vect_get_peeling_costs_all_drs to compute the penalty for no
23894 peeling to match up costs.
23896 2017-07-21 Richard Biener <rguenther@suse.de>
23898 PR tree-optimization/81500
23899 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23900 we didn't identify a reduction path.
23902 2017-07-21 Tom de Vries <tom@codesourcery.com>
23903 Cesar Philippidis <cesar@codesourcery.com>
23905 PR gcov-profile/81442
23906 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23909 2017-07-21 Tom de Vries <tom@codesourcery.com>
23912 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23914 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23915 nvptx_override_options_after_change.
23917 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23919 * dwarf2out.c (output_file_names): Avoid double testing for
23920 dwarf_version >= 5.
23922 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23924 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23926 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23928 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23930 (try_crossjump_to_edge): Do not punt on partitioned functions.
23932 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23934 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23935 Put all BBs reachable only via paths crossing cold region to cold
23937 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23939 2016-07-21 Richard Biener <rguenther@suse.de>
23941 PR tree-optimization/81303
23942 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23943 into account prologue and epilogue iterations when raising
23944 min_profitable_iters to sth at least covering one vector iteration.
23946 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23948 * config/arm/arm.c (arm_test_cpu_arch_dat):
23951 2017-07-20 Nathan Sidwell <nathan@acm.org>
23953 Remove TYPE_METHODS.
23954 * tree.h (TYPE_METHODS): Delete.
23955 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23956 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23957 (dbxout_type_methods): Scan TYPE_FIELDS.
23958 (dbxout_type): Don't check TYPE_METHODS here.
23959 * function.c (use_register_for_decl): Always ignore register for
23960 class types when not optimizing.
23961 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23962 * tree.c (free_lang_data_in_type): Stitch out member functions and
23963 templates from TYPE_FIELDS.
23964 (build_distinct_type_copy, verify_type_variant,
23965 verify_type): Member fns are on TYPE_FIELDS.
23966 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23967 * tree-pretty-print.c (dump_generic_node): Likewise.
23969 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23972 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23973 V2TImode and V4TImode.
23974 (ix86_expand_vector_extract): Likewise.
23975 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23976 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23977 (ssescalarmode): Handle V4TImode and V2TImode.
23978 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23979 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23980 (VEXTRACTI128_MODE): New mode iterator.
23981 (splitter for *vec_extractv?ti first element): New.
23982 (VEC_INIT_MODE): New mode iterator.
23983 (vec_init<mode>): Consolidate 3 expanders into one using
23984 VEC_INIT_MODE mode iterator.
23986 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23988 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23989 non_spilled_static_chain_regno_p.
23991 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23993 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23995 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23997 * bb-reorder.c (connect_traces): Allow copying of blocks within
24000 2017-07-20 Richard Biener <rguenther@suse.de>
24002 * gimple.h (gimple_phi_result): Add gphi * overload.
24003 (gimple_phi_result_ptr): Likewise.
24004 (gimple_phi_arg): Likewise. Adjust index assert to only
24005 allow actual argument accesses rather than all slots available
24007 (gimple_phi_arg_def): Add gphi * overload.
24008 * tree-phinodes.c (make_phi_node): Initialize only actual
24010 (resize_phi_node): Clear memory not covered by old node,
24011 do not initialize excess argument slots.
24012 (reserve_phi_args_for_new_edge): Initialize new argument slot
24015 2017-07-20 Bin Cheng <bin.cheng@arm.com>
24017 PR tree-optimization/81388
24019 2016-07-21 Bin Cheng <bin.cheng@arm.com>
24021 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
24022 by removing computation of may_be_zero.
24024 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24025 Tom de Vries <tom@codesourcery.com>
24027 PR middle-end/81030
24028 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
24029 when gimple level profile disagrees with what RTL expander did.
24031 2017-07-20 Richard Biener <rguenther@suse.de>
24033 PR tree-optimization/61171
24034 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
24035 (vect_analyze_stmt): Add slp instance parameter.
24036 (vectorizable_reduction): Likewise.
24037 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
24038 (vect_is_simple_reduction): Deal with chains not detected
24039 as SLP reduction chain, specifically not properly associated
24040 chains containing a mix of plus/minus.
24041 (get_reduction_op): Remove.
24042 (get_initial_defs_for_reduction): Simplify, pass in whether
24043 this is a reduction chain, pass in the SLP node for the PHIs.
24044 (vect_create_epilog_for_reduction): Get the SLP instance as
24046 (vectorizable_reduction): Get the SLP instance as arg.
24047 During analysis remember the SLP node with the PHIs in the
24048 instance. Simplify getting at the vectorized reduction PHIs.
24049 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
24050 through SLP instance.
24051 (vect_slp_analyze_operations): Likewise.
24052 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
24053 (vect_transform_stmt): Likewise.
24055 2017-07-20 Tom de Vries <tom@codesourcery.com>
24057 PR tree-optimization/81489
24058 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
24059 read of phi arg location to before loop that modifies phi.
24061 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24063 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
24066 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24068 PR middle-end/81331
24069 * except.c (execute): Fix ordering issue.
24071 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
24073 PR rtl-optimization/81423
24074 * combine.c (make_compound_operation_int): Don't try to optimize
24075 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
24077 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
24079 PR rtl-optimization/81423
24080 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
24081 with a constant that is -1 in the truncated to mode.
24083 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24085 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
24086 (determine_unlikely_bbs): ... here.
24087 * predict.h (propagate_unlikely_bbs_forward): Declare.
24088 * cfgexpand.c (pass_expand::execute): Use it.
24089 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
24091 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
24092 propagate_unlikely_bbs_forward.
24094 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24096 PR middle-end/81331
24097 * except.c (maybe_add_nop_after_section_switch): New function.
24100 2017-07-19 Tom de Vries <tom@codesourcery.com>
24102 * gimple.h (gimple_phi_set_arg): Make assert more strict.
24104 2017-07-19 Tom de Vries <tom@codesourcery.com>
24106 * gimple.h (gimple_phi_arg): Make assert more strict.
24108 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
24110 * config.gcc (powerpc*-*-*): Add mmintrin.h.
24111 * config/rs6000/mmintrin.h: New file.
24112 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
24114 2017-07-19 Jakub Jelinek <jakub@redhat.com>
24116 PR tree-optimization/81346
24117 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
24119 2017-07-19 Tom de Vries <tom@codesourcery.com>
24121 * config/nvptx/nvptx.md (VECIM): Add V2DI.
24123 2017-07-19 Tom de Vries <tom@codesourcery.com>
24125 * config/nvptx/nvptx-modes.def: Add V2DImode.
24126 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
24127 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
24128 (nvptx_output_mov_insn): Handle lack of mov.b128.
24129 (nvptx_print_operand): Handle 'H' and 'L' codes.
24130 (nvptx_vector_mode_supported): Allow V2DImode.
24131 (nvptx_preferred_simd_mode): New function.
24132 (nvptx_data_alignment): New function.
24133 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
24134 nvptx_preferred_simd_mode.
24135 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
24137 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
24139 2017-07-19 Tom de Vries <tom@codesourcery.com>
24141 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
24142 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
24143 (nvptx_vector_mode_supported): New function. Allow V2SImode.
24144 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
24145 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
24146 (mov<VECIM>_insn): New define_insn.
24147 (define_expand "mov<VECIM>): New define_expand.
24149 2017-07-19 Tom de Vries <tom@codesourcery.com>
24151 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
24153 2017-07-19 Jakub Jelinek <jakub@redhat.com>
24155 PR tree-optimization/81346
24156 * fold-const.h (fold_div_compare, range_check_type): Declare.
24157 * fold-const.c (range_check_type): New function.
24158 (build_range_check): Use range_check_type.
24159 (fold_div_compare): No longer static, rewritten into
24160 a match.pd helper function.
24161 (fold_comparison): Don't call fold_div_compare here.
24162 * match.pd (X / C1 op C2): New optimization using fold_div_compare
24163 as helper function.
24165 2017-07-19 Nathan Sidwell <nathan@acm.org>
24167 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
24168 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
24169 * tree.c (find_decls_types_r, verify_type): Use
24170 TYPE_{MIN,MAX}_VALUE_RAW.
24171 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
24172 (hash_tree): Likewise.
24173 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
24175 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
24178 2017-07-18 Tom de Vries <tom@codesourcery.com>
24180 PR middle-end/81464
24181 * omp-expand.c (expand_omp_for_static_chunk): Handle
24182 equal-argument loop exit phi.
24184 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
24187 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
24188 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
24189 operand 2 predicate.
24190 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
24191 operand 2 predicate.
24192 (ror,rol -> rorx splitters): Use const_int_operand as
24193 operand 2 predicate.
24195 2017-06-18 Richard Biener <rguenther@suse.de>
24197 PR tree-optimization/81410
24198 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
24199 the gap in the ! slp_perm SLP case after each group.
24201 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24203 PR middle-end/81463
24204 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
24207 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24209 PR middle-end/81462
24210 * predict.c (set_even_probabilities): Cleanup; do not affect
24211 probabilities that are already known.
24212 (combine_predictions_for_bb): Call even when count is set.
24214 2017-07-18 Nathan Sidwell <nathan@acm.org>
24216 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
24219 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24222 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
24223 optimization for loop niter analysis.
24225 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
24228 * config/avr/avr.c (avr_optimize_casesi): Don't use
24229 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
24231 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
24233 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
24234 body_cost_vec from _vect_peel_extended_info.
24235 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
24236 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
24239 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24241 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
24243 2017-07-18 Richard Biener <rguenther@suse.de>
24245 PR tree-optimization/80620
24246 PR tree-optimization/81403
24247 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
24248 info when re-using a VN table entry.
24250 2017-07-18 Richard Biener <rguenther@suse.de>
24252 PR tree-optimization/81418
24253 * tree-vect-loop.c (vectorizable_reduction): Properly compute
24254 vectype_in. Verify that with lane-reducing reduction operations
24255 we have a single def-use cycle.
24257 2017-07-17 Carl Love <cel@us.ibm.com>
24259 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
24261 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24262 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24263 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24264 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24265 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24266 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24267 VMULOSW): New enum "unspec" values.
24268 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24269 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24270 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24271 altivec_vmulosw): New patterns.
24272 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24273 VMULOSW): Add definitions.
24275 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
24277 * config/alpha/alpha.c: Include predict.h.
24279 2017-07-17 Yury Gribov <tetra2005@gmail.com>
24281 * tree-vrp.c (compare_assert_loc): Fix comparison function
24282 to return predictable results.
24284 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24286 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
24288 (subdi3): Likewise.
24289 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
24290 * doc/invoke.texi (mexpand-adddi): Update text.
24292 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24294 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
24295 that also clobbers the CC register. The old expand code is moved
24297 (*arc_clzsi2): ... here.
24298 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
24299 the CC register. The old expand code is moved to ...
24300 (arc_ctzsi2): ... here.
24302 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24304 * config/arc/arc.opt (mindexed-loads): Use initial value
24305 TARGET_INDEXED_LOADS_DEFAULT.
24306 (mauto-modify-reg): Use initial value
24307 TARGET_AUTO_MODIFY_REG_DEFAULT.
24308 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24309 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24310 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24311 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24313 2017-07-17 Martin Liska <mliska@suse.cz>
24316 * opts.c (finish_options): Do not allow -fgnu-tm
24317 w/ -fsanitize={kernel-,}address. Say sorry.
24319 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24322 * tree-loop-distribution.c (classify_partition): Only assert on
24323 numer of iterations.
24324 (merge_dep_scc_partitions): Delete prameter. Update function call.
24325 (distribute_loop): Remove code handling loop with unknown niters.
24326 (pass_loop_distribution::execute): Skip loop with unknown niters.
24328 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24331 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
24332 function sort_partitions_by_post_order.
24334 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24336 PR tree-optimization/81374
24337 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
24338 the max index of basic blocks, rather than number of basic blocks.
24340 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24342 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
24344 (arc_legitimate_pic_operand_p): Likewise.
24345 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
24347 (arc_needs_pcl_p): Likewise.
24348 (arc_legitimate_pc_offset_p): Likewise.
24349 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
24350 function is also used in constrains.md.
24351 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
24352 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
24353 PLUS. Only return true/false in known cases, otherwise assert.
24354 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
24355 is already called in arc_legitimate_constant_p.
24356 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
24358 (LEGITIMATE_PIC_OPERAND_P): Use
24359 arc_raw_symbolic_reference_mentioned_p function.
24360 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
24365 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24366 Andrew Burgess <andrew.burgess@embecosm.com>
24368 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
24369 (arc_return_address_register): New function.
24370 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
24371 (arc_handle_fndecl_attribute): Add naked attribute.
24372 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
24373 (TARGET_WARN_FUNC_RETURN): Likewise.
24374 (arc_allocate_stack_slots_for_args): New function.
24375 (arc_warn_func_return): Likewise.
24376 (machine_function): Change type fn_type.
24377 (arc_compute_function_type): Consider new naked function type,
24378 change function return type.
24379 (arc_must_save_register): Adapt to handle new
24380 arc_compute_function_type's return type.
24381 (arc_expand_prologue): Likewise.
24382 (arc_expand_epilogue): Likewise.
24383 (arc_return_address_regs): Delete.
24384 (arc_return_address_register): New function.
24385 (arc_epilogue_uses): Use above function.
24386 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
24387 (arc_function_type): Change encoding, add naked type.
24388 (ARC_INTERRUPT_P): Change to handle the new encoding.
24389 (ARC_FAST_INTERRUPT_P): Likewise.
24390 (ARC_NORMAL_P): Define.
24391 (ARC_NAKED_P): Likewise.
24392 (arc_compute_function_type): Delete prototype.
24393 * config/arc/arc.md (in_ret_delay_slot): Use
24394 arc_return_address_register function.
24395 (simple_return): Likewise.
24396 (p_return_i): Likewise.
24398 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24400 PR tree-optimization/81428
24401 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
24402 can't be built for those types.
24404 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24406 Remove stuff dead since r239246.
24408 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
24409 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
24410 (avr_inform_devices): Remove dead stuff.
24412 2017-07-17 Tamar Christina <tamar.christina@arm.com>
24414 * config/arm/arm_neon.h: Fix softp typo.
24416 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24418 PR tree-optimization/81365
24419 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
24420 aggregate moves onto bb predecessor edges, make sure there are no
24421 loads that could alias the lhs in between the start of bb and the
24424 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24427 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
24428 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
24429 [LSHIFTRT, outer_code = TRUNCATE]: Same.
24431 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24433 PR tree-optimization/81396
24434 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
24435 (init_symbolic_number): Initialize it to 1.
24436 (perform_symbolic_merge): Add n_ops from both operands into the new
24438 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
24439 without base_addr as useless if they need more than one operation.
24440 (bswap_replace): Handle !bswap case for NULL base_addr.
24442 2017-07-17 Tom de Vries <tom@codesourcery.com>
24445 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
24448 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
24450 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
24451 conditional builtin define __FIX_LEON3FT_B2BST.
24453 2017-07-17 Daniel Cederman <cederman@gaisler.com>
24455 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
24456 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
24459 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24461 PR rtl-optimization/81424
24462 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
24463 to remove potential trapping from operands if -fnon-call-exceptions.
24465 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24467 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
24468 profile_proability for scalling.
24469 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
24471 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24473 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
24475 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24477 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
24478 fixpoint arithmetics.
24480 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24482 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
24483 fixpoint arithmetics.
24485 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24487 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
24488 fixpoint arithmetics.
24490 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24492 * profile-count.h (profile_probability::from_reg_br_prob_note,
24493 profile_probability::to_reg_br_prob_note): New functions.
24494 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
24495 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
24496 * predict.c (probability_reliable_p): Update.
24497 (edge_probability_reliable_p): Update.
24498 (br_prob_note_reliable_p): Update.
24499 (invert_br_probabilities): Update.
24500 (add_reg_br_prob_note): New function.
24501 (combine_predictions_for_insn): Update.
24502 * asan.c (asan_clear_shadow): Update.
24503 * cfgbuild.c (compute_outgoing_frequencies): Update.
24504 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24505 (update_br_prob_note): Update.
24506 (rtl_verify_edges): Update.
24507 (purge_dead_edges): Update.
24508 (fixup_reorder_chain): Update.
24509 * emit-rtl.c (try_split): Update.
24510 * ifcvt.c (cond_exec_process_insns): Update.
24511 (cond_exec_process_if_block): Update.
24512 (dead_or_predicable): Update.
24513 * internal-fn.c (expand_addsub_overflow): Update.
24514 (expand_neg_overflow): Update.
24515 (expand_mul_overflow): Update.
24516 * loop-doloop.c (doloop_modify): Update.
24517 * loop-unroll.c (compare_and_jump_seq): Update.
24518 * optabs.c (emit_cmp_and_jump_insn_1): Update.
24519 * predict.h: Update.
24520 * reorg.c (mostly_true_jump): Update.
24522 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
24523 * config/alpha/alpha.c (emit_unlikely_jump): Update.
24524 * config/arc/arc.c: (emit_unlikely_jump): Update.
24525 * config/arm/arm.c: (emit_unlikely_jump): Update.
24526 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
24527 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
24528 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
24529 (ix86_print_operand): Update.
24530 (ix86_split_fp_branch): Update.
24531 (predict_jump): Update.
24532 * config/ia64/ia64.c (ia64_print_operand): Update.
24533 * config/mmix/mmix.c (mmix_print_operand): Update.
24534 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
24535 (rs6000_expand_split_stack_prologue): Update.
24536 * config/rs6000/rs6000.c: Update.
24537 * config/s390/s390.c (s390_expand_vec_strlen): Update.
24538 (s390_expand_vec_movstr): Update.
24539 (s390_expand_cs_tdsi): Update.
24540 (s390_expand_split_stack_prologue): Update.
24541 * config/sh/sh.c (sh_print_operand): Update.
24542 (expand_cbranchsi4): Update.
24543 (expand_cbranchdi4): Update.
24544 * config/sparc/sparc.c (output_v9branch): Update.
24545 * config/spu/spu.c (get_branch_target): Update.
24546 (ea_load_store_inline): Update.
24547 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
24548 * config/tilepro/tilepro.c: Update.
24550 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24552 * gimplify.c (mostly_copy_tree_r): Revert latest change.
24553 (gimplify_save_expr): Likewise.
24555 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24557 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
24559 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24561 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
24563 * timevar.def (TV_IPA_FNSUMMARY): Define.
24565 2017-07-16 Daniel Cederman <cederman@gaisler.com>
24567 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
24568 to back store errata sensitive sequence from being generated.
24569 (sqrtdf2_fix): Likewise.
24571 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24573 * tree-ssa-threadupdate.c (compute_path_counts,
24574 update_joiner_offpath_counts): Use profile_probability.
24576 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
24579 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24581 * config/arm/arm-c.c (arm_cpu_builtins): Define
24582 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24584 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
24586 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24587 array entries to represent __ieee128 versions of the
24588 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
24589 scalar_extract_sig, and scalar_insert_exp built-in functions.
24590 (altivec_resolve_overloaded_builtin): Add special case handling
24591 for the __builtin_scalar_insert_exp function, as represented by
24592 the P9V_BUILTIN_VEC_VSIEDP constant.
24593 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
24594 exponent support for __ieee128 argument.
24595 (VSESQP): Add scalar extract signature support for __ieee128
24597 (VSTDCNQP): Add scalar test negative support for __ieee128
24599 (VSIEQP): Add scalar insert exponent support for __int128 argument
24600 with __ieee128 result.
24601 (VSIEQPF): Add scalar insert exponent support for __ieee128
24602 argument with __ieee128 result.
24603 (VSTDCQP): Add scalar test data class support for __ieee128
24605 (VSTDCNQP): Add overload support for scalar test negative with
24606 __ieee128 argument.
24607 (VSTDCQP): Add overload support for scalar test data class
24608 __ieee128 argument.
24609 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
24610 UNSPEC_VSX_SXSIGDP.
24611 (UNSPEC_VSX_SIEXPQP): New constant.
24612 (xsxexpqp): New insn for VSX scalar extract exponent quad
24614 (xsxsigqp): New insn for VSX scalar extract significand quad
24616 (xsiexpqpf): New insn for VSX scalar insert exponent quad
24617 precision with floating point argument.
24618 (xststdcqp): New expand for VSX scalar test data class quad
24620 (xststdcnegqp): New expand for VSX scalar test negative quad
24622 (xststdcqp): New insn to match expansions for VSX scalar test data
24623 class quad precision and VSX scalar test negative quad precision.
24624 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
24625 special case operand checking to enforce that second operand of
24626 VSX scalar test data class with quad precision argument is a 7-bit
24628 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24629 prototypes and descriptions of __ieee128 versions of
24630 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
24631 scalar_test_data_class, and scalar_test_neg built-in functions.
24633 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24635 PR tree-optimization/81162
24636 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
24637 replace a negate with an add.
24639 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
24641 * doc/invoke.texi (arm/-mcpu): Document +crypto.
24643 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24645 * config/arm/arm-c.c (arm_cpu_builtins): Define
24646 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24648 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24650 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
24651 (armv8-r): Set ARM Cortex-R52 as default CPU.
24652 * config/arm/arm-tables.opt: Regenerate.
24653 * config/arm/arm-tune.md: Regenerate.
24654 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24656 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
24657 extension for -mcpu=cortex-r52.
24659 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24661 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
24662 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
24663 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
24664 (fp-armv8): Define it as FP_ARMv8 only.
24665 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
24666 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
24668 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
24669 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
24670 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
24671 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
24672 than TARGET_FPU_ARMV8.
24673 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
24674 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
24675 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
24677 * config/arm/neon.md (neon_vrint): Likewise.
24678 (neon_vcvt): Likewise.
24679 (neon_<fmaxmin_op><mode>): Likewise.
24680 (<fmaxmin><mode>3): Likewise.
24681 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
24682 * config/arm/predicates.md (arm_cond_move_operator): Check against
24683 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
24685 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
24687 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
24688 to top of function.
24690 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24692 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
24693 loop in comment with memset.
24695 2017-07-14 Martin Liska <mliska@suse.cz>
24697 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
24698 * dwarf2out.c (is_java): Remove the function.
24699 (output_pubname): Remove usage of the function.
24700 (lower_bound_default): Remove usage of DW_LANG_Java.
24701 (gen_compile_unit_die): Likewise.
24702 * gcc.c: Remove compiler defaults for .java and .zip files.
24703 * gimple-expr.c (remove_suffix): Change as there's no longer
24704 extension than 4-letter one.
24705 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
24706 (gimplify_save_expr): Likewise.
24707 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
24708 as it's possible even for other languages than Java.
24709 * langhooks.h (struct lang_hooks): Remove Java from a comment.
24710 * lto-opts.c (lto_write_options): Remove reference to Java.
24711 * opts.c (strip_off_ending): Update file extension handling.
24712 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
24713 * tree-eh.c (lower_resx): Likewise.
24714 * tree.c (free_lang_data_in_type): Remove dead code.
24715 (find_decls_types_r): Likewise.
24716 (build_common_builtin_nodes): Remove Java from a comment.
24717 (verify_type): Remove dead code.
24718 * varasm.c (assemble_external): Remove Java from a comment.
24720 2017-07-14 Martin Liska <mliska@suse.cz>
24722 * opts.c (finish_options): Add quotes.
24723 (common_handle_option): Likewise.
24725 2017-07-14 Martin Liska <mliska@suse.cz>
24727 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
24728 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
24729 Remove N_SO_PASCAL.
24730 * dwarf2out.c (lower_bound_default): Do not handle
24732 (gen_compile_unit_die): Likewise.
24733 * gcc.c: Remove default extension binding for GNU Pascal.
24734 * stmt.c: Remove Pascal language from a comment.
24735 * xcoffout.c: Likewise.
24737 2017-07-13 David Malcolm <dmalcolm@redhat.com>
24740 * diagnostic-show-locus.c (fixit_cmp): New function.
24741 (layout::layout): Sort m_fixit_hints.
24742 (column_range::column_range): Assert that the values are valid.
24743 (struct char_span): New struct.
24744 (correction::overwrite): New method.
24745 (struct source_line): New struct.
24746 (line_corrections::add_hint): Add assertions. Reimplement memcpy
24747 calls in terms of classes source_line and char_span, and
24748 correction::overwrite.
24749 (selftest::test_overlapped_fixit_printing_2): New function.
24750 (selftest::diagnostic_show_locus_c_tests): Call it.
24752 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
24754 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
24755 early if there is no lhs.
24757 2017-07-13 Martin Liska <mliska@suse.cz>
24759 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
24760 (gen_reference_type_die): Likewise.
24761 * stor-layout.c: Remove Pascal-related comment.
24763 2017-07-13 Martin Liska <mliska@suse.cz>
24765 * opts.c (finish_options): Add quotes to error messages.
24766 (parse_sanitizer_options): Likewise.
24768 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
24770 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
24772 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
24774 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24776 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24778 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24780 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24782 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24785 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24786 provides the hardware capability bits, define the macro
24787 __BUILTIN_CPU_SUPPORTS__.
24788 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24789 if GLIBC does not provide the hardware capability bits. Add a
24790 gcc_unreachable call if the built-in cpu function is neither
24791 __builtin_cpu_is nor __builtin_cpu_supports.
24792 (rs6000_get_function_versions_dispatcher): Change the warning
24793 that an old GLIBC is used which does not export the capability
24794 bits to be an error.
24795 * doc/extend.texi (target_clones attribute): Document the
24796 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24797 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24798 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24799 the macros defined by GCC if the newer GLIBC is available.
24801 2017-07-12 Jeff Law <law@redhat.com>
24803 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24804 remaining includes slightly.
24805 * config/riscv/riscv-builtins.c: Include profile-count.h.
24807 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24810 * config/avr/avr.c (avr_set_current_function): In diagnostic
24811 messages: Quote keywords and (parts of) identifiers.
24812 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24815 2017-07-12 Carl Love <cel@us.ibm.com>
24817 * config/rs6000/rs6000-c.c: Add support for built-in functions
24818 vector bool char vec_revb (vector bool char);
24819 vector bool short vec_revb (vector short char);
24820 vector bool int vec_revb (vector bool int);
24821 vector bool long long vec_revb (vector bool long long);
24822 * doc/extend.texi: Update the built-in documentation file for the
24823 new built-in functions.
24825 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24827 * config/s390/s390.md: Remove movcc splitter.
24829 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24831 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24832 load/store on condition.
24834 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24837 * config/avr/avr.c (avr_encode_section_info)
24838 [progmem && !TREE_READONLY]: Error if progmem object needs
24841 2017-07-11 Michael Collison <michael.collison@arm.com>
24843 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24846 2017-07-11 Carl Love <cel@us.ibm.com>
24848 * config/rs6000/rs6000-c.c: Add support for builtins
24849 vector unsigned int vec_parity_lsbb (vector signed int);
24850 vector unsigned int vec_parity_lsbb (vector unsigned int);
24851 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24852 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24853 vector unsigned long long vec_parity_lsbb (vector signed long long);
24854 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24855 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24856 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24857 * doc/extend.texi: Update the built-in documentation file for the
24858 new built-in functions.
24860 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24862 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24863 (layout::m_primary_loc): New field.
24864 (layout::layout): Initialize new field. Move location filtering
24865 logic from here to...
24866 (layout::maybe_add_location_range): ...this new method. Add
24867 support for filtering to just the lines already specified by other
24869 (layout::will_show_line_p): New method.
24870 (gcc_rich_location::add_location_if_nearby): New method.
24871 (selftest::test_add_location_if_nearby): New test function.
24872 (selftest::diagnostic_show_locus_c_tests): Call it.
24873 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24876 2017-07-11 Tom de Vries <tom@codesourcery.com>
24878 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24879 (bb_first_real_insn): New function.
24880 (nvptx_single): Add extra initialization of broadcasted condition
24883 2017-07-11 Nathan Sidwell <nathan@acm.org>
24885 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24887 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24889 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24890 Binutils doc as TEXI will mess them up.
24891 * doc/invoke.texi (AVR Options): Same here.
24893 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24895 * config/sparc/sparc.opt (mfix-ut700): New option.
24896 (mfix-gr712rc): Likewise.
24897 (sparc_fix_b2bst): New variable.
24898 * doc/invoke.texi (SPARC options): Document them.
24899 (ARM options): Fix warnings.
24900 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24901 instructions to prevent sequences that can trigger the store-store
24902 errata for certain LEON3FT processors.
24903 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24904 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24905 * config/sparc/sparc.md (fix_b2bst): New attribute.
24906 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24908 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24911 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24913 (*rsqrtsf2_sse): Ditto.
24915 (div<mode>3): Macroize insn from divdf3 and divsf3
24916 using MODEF mode iterator.
24918 2017-07-10 Martin Sebor <msebor@redhat.com>
24920 PR tree-optimization/80397
24921 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24922 instead of testing for equality to INTEGER_TYPE.
24924 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24926 * config.gcc: Remove uclibc from arc target spec.
24928 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24930 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24932 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24935 * lto-wrapper.c (remove_option): New function.
24936 (merge_and_complain): Merge PIC/PIE options more realistically.
24938 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24940 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24944 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24945 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24946 * config.in: Regenerate.
24947 * configure: Regenerate.
24948 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24949 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24950 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24951 (TARGET_GASISR_PROLOGUES): ...target mask.
24952 * common/config/avr/avr-common.c
24953 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24954 Set -mgas-isr-prologues.
24955 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24956 INSERT_PASS_BEFORE for it.
24957 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24958 * config/avr/avr.c (avr_option_override)
24959 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24960 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24961 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24962 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24963 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24965 (make_avr_pass_pre_proep): New function.
24966 (emit_push_sfr) <treg>: Add argument to function and use it
24967 instead of TMP_REG.
24968 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24969 and set machine->gasisr.yes.
24970 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24971 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24972 __gcc_isr.n_pushed to .L__stack_usage.
24973 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24974 (avr_asm_final_postscan_insn): ...this new static function.
24975 * config/avr/avr.h (machine_function)
24976 <is_no_gccisr, use_L__stack_usage>: New fields.
24977 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24978 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24979 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24980 (gasisr, *gasisr): New expander and insn.
24981 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24982 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24983 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24985 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24987 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24990 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24992 Move jump-tables out of .text again.
24995 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24996 (ASM_OUTPUT_ADDR_VEC): New function.
24997 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24998 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24999 INSN_ADDRESSes as asm comment.
25000 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
25001 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
25002 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
25003 * config/avr/avr.md (*tablejump): Adjust comment.
25004 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
25005 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
25007 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
25008 (avr_output_addr_vec): New proto.
25009 (avr_log_t) <insn_addresses>: New field.
25011 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
25014 * config/i386/i386.c (ix86_function_arg_advance): Set
25015 outgoing_args_on_stack to true if there are outgoing arguments
25017 (ix86_function_arg): Likewise.
25018 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
25019 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
25020 * config/i386/i386.h (machine_function): Add
25021 outgoing_args_on_stack.
25023 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
25025 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
25026 supporting pthreds.
25027 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
25029 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
25031 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
25032 (REAL_H): Remove $(MACHMODE_H).
25033 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
25035 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
25036 $(MACHMODE_H) and double-int.h.
25037 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
25039 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
25042 2017-07-07 Andrew Pinski <apinski@cavium.com>
25044 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
25045 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
25047 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
25049 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
25050 Add warning if GCC was not configured to link against a GLIBC that
25051 exports the hardware capability bits.
25052 (make_resolver_func): Make resolver function private and not a
25053 COMDAT function. Create the name with clone_function_name instead
25054 of make_unique_name.
25057 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
25058 correct operand in doing the split.
25060 2017-07-07 Carl Love <cel@us.ibm.com>
25062 * config/rs6000/rs6000-c: Add support for built-in function
25063 vector unsigned short vec_pack_to_short_fp32 (vector float,
25065 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
25066 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
25067 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
25068 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
25069 (convert_4f32_8i16): Add define_expand.
25070 * doc/extend.texi: Update the built-in documentation file for the
25071 new built-in function.
25073 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25075 * config/sparc/m8.md: New file.
25076 * config/sparc/sparc.md: Include m8.md.
25078 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25080 * config/sparc/sparc.opt: New option -mvis4b.
25081 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
25082 (sparc_option_override): Handle VIS4B.
25083 (enum sparc_builtins): Define
25084 SPARC_BUILTIN_DICTUNPACK{8,16,32},
25085 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
25086 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
25087 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
25088 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
25089 (check_constant_argument): New function.
25090 (sparc_vis_init_builtins): Define builtins
25091 __builtin_vis_dictunpack{8,16,32},
25092 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
25093 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
25094 __builtin_vis_fpcmpde{8,16,32}shl and
25095 __builtin_vis_fpcmpur{8,16,32}shl.
25096 (sparc_expand_builtin): Check that the constant operands to
25097 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
25098 constant and in range.
25099 * config/sparc/sparc-c.c (sparc_target_macros): Handle
25101 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
25102 (SPARC_IMM5_P): Likewise.
25103 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
25104 (enabled): Handle vis4b.
25105 (UNSPEC_DICTUNPACK): New unspec.
25106 (UNSPEC_FPCMPSHL): Likewise.
25107 (UNSPEC_FPUCMPSHL): Likewise.
25108 (UNSPEC_FPCMPDESHL): Likewise.
25109 (UNSPEC_FPCMPURSHL): Likewise.
25110 (cpu_feature): New CPU feature `vis4b'.
25111 (dictunpack{8,16,32}): New insns.
25112 (FPCSMODE): New mode iterator.
25113 (fpcscond): New code iterator.
25114 (fpcsucond): Likewise.
25115 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
25116 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
25117 (fpcmpde{8,16,32}{si,di}shl): Likewise.
25118 (fpcmpur{8,16,32}{si,di}shl): Likewise.
25119 * config/sparc/constraints.md: Define constraints `q' for unsigned
25120 2-bit integer constants and `t' for unsigned 5-bit integer
25122 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
25124 (imm5_operand_dictunpack16): Likewise.
25125 (imm5_operand_dictunpack32): Likewise.
25126 (imm2_operand): Likewise.
25127 * doc/invoke.texi (SPARC Options): Document -mvis4b.
25128 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
25129 ditunpack* and fpcmp*shl builtins.
25131 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25133 * config.gcc: Handle m8 in --with-{cpu,tune} options.
25134 * config.in: Add HAVE_AS_SPARC6 define.
25135 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
25137 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
25139 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
25140 (CPP_CPU_SPEC): Handle m8.
25141 (ASM_CPU_SPEC): Likewise.
25142 * config/sparc/sparc-opts.h (enum processor_type): Add
25144 * config/sparc/sparc.c (m8_costs): New struct.
25145 (sparc_option_override): Handle TARGET_CPU_m8.
25146 (sparc32_initialize_trampoline): Likewise.
25147 (sparc64_initialize_trampoline): Likewise.
25148 (sparc_issue_rate): Likewise.
25149 (sparc_register_move_cost): Likewise.
25150 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
25151 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
25152 (ASM_CPU64_DEFAULT_SPEC): Likewise.
25153 (CPP_CPU_SPEC): Handle M8.
25154 (ASM_CPU_SPEC): Likewise.
25155 (AS_M8_FLAG): Define.
25156 * config/sparc/sparc.md: Add m8 to the cpu attribute.
25157 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
25158 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
25160 * configure: Regenerate.
25161 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
25164 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25166 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
25168 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
25169 ("*movdi_insn_sp32"): Do not set v3pipe.
25170 ("*movsi_insn"): Likewise.
25171 ("*movdi_insn_sp64"): Likewise.
25172 ("*movsf_insn"): Likewise.
25173 ("*movdf_insn_sp32"): Likewise.
25174 ("*movdf_insn_sp64"): Likewise.
25175 ("*zero_extendsidi2_insn_sp64"): Likewise.
25176 ("*sign_extendsidi2_insn"): Likewise.
25177 ("*mov<VM32:mode>_insn"): Likewise.
25178 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25179 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25180 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25181 ("<vlop:code><VL:mode>3"): Likewise.
25182 ("*not_<vlop:code><VL:mode>3"): Likewise.
25183 ("*nand<VL:mode>_vis"): Likewise.
25184 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
25185 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
25186 ("one_cmpl<VL:mode>2"): Likewise.
25187 ("faligndata<VM64:mode>_vis"): Likewise.
25188 ("alignaddrsi_vis"): Likewise.
25189 ("alignaddrdi_vis"): Likweise.
25190 ("alignaddrlsi_vis"): Likewise.
25191 ("alignaddrldi_vis"): Likewise.
25192 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25193 ("bmaskdi_vis"): Likewise.
25194 ("bmasksi_vis"): Likewise.
25195 ("bshuffle<VM64:mode>_vis"): Likewise.
25196 ("cmask8<P:mode>_vis"): Likewise.
25197 ("cmask16<P:mode>_vis"): Likewise.
25198 ("cmask32<P:mode>_vis"): Likewise.
25199 ("pdistn<P:mode>_vis"): Likewise.
25200 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25202 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25204 * config/sparc/sparc.md ("subtype"): New insn attribute.
25205 ("*wrgsr_sp64"): Set insn subtype.
25206 ("*rdgsr_sp64"): Likewise.
25207 ("alignaddrsi_vis"): Likewise.
25208 ("alignaddrdi_vis"): Likewise.
25209 ("alignaddrlsi_vis"): Likewise.
25210 ("alignaddrldi_vis"): Likewise.
25211 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25212 ("fexpand_vis"): Likewise.
25213 ("fpmerge_vis"): Likewise.
25214 ("faligndata<VM64:mode>_vis"): Likewise.
25215 ("bshuffle<VM64:mode>_vis"): Likewise.
25216 ("cmask8<P:mode>_vis"): Likewise.
25217 ("cmask16<P:mode>_vis"): Likewise.
25218 ("cmask32<P:mode>_vis"): Likewise.
25219 ("fchksm16_vis"): Likewise.
25220 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
25221 ("fmean16_vis"): Likewise.
25222 ("fp<plusminus_insn>64_vis"): Likewise.
25223 ("<plusminus_insn>v8qi3"): Likewise.
25224 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25225 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
25226 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
25227 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
25228 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
25229 ("*movqi_insn"): Likewise.
25230 ("*movhi_insn"): Likewise.
25231 ("*movsi_insn"): Likewise.
25232 ("movsi_pic_gotdata_op"): Likewise.
25233 ("*movdi_insn_sp32"): Likewise.
25234 ("*movdi_insn_sp64"): Likewise.
25235 ("movdi_pic_gotdata_op"): Likewise.
25236 ("*movsf_insn"): Likewise.
25237 ("*movdf_insn_sp32"): Likewise.
25238 ("*movdf_insn_sp64"): Likewise.
25239 ("*zero_extendhisi2_insn"): Likewise.
25240 ("*zero_extendqihi2_insn"): Likewise.
25241 ("*zero_extendqisi2_insn"): Likewise.
25242 ("*zero_extendqidi2_insn"): Likewise.
25243 ("*zero_extendhidi2_insn"): Likewise.
25244 ("*zero_extendsidi2_insn_sp64"): Likewise.
25245 ("ldfsr"): Likewise.
25246 ("prefetch_64"): Likewise.
25247 ("prefetch_32"): Likewise.
25248 ("tie_ld32"): Likewise.
25249 ("tie_ld64"): Likewise.
25250 ("*tldo_ldub_sp32"): Likewise.
25251 ("*tldo_ldub1_sp32"): Likewise.
25252 ("*tldo_ldub2_sp32"): Likewise.
25253 ("*tldo_ldub_sp64"): Likewise.
25254 ("*tldo_ldub1_sp64"): Likewise.
25255 ("*tldo_ldub2_sp64"): Likewise.
25256 ("*tldo_ldub3_sp64"): Likewise.
25257 ("*tldo_lduh_sp32"): Likewise.
25258 ("*tldo_lduh1_sp32"): Likewise.
25259 ("*tldo_lduh_sp64"): Likewise.
25260 ("*tldo_lduh1_sp64"): Likewise.
25261 ("*tldo_lduh2_sp64"): Likewise.
25262 ("*tldo_lduw_sp32"): Likewise.
25263 ("*tldo_lduw_sp64"): Likewise.
25264 ("*tldo_lduw1_sp64"): Likewise.
25265 ("*tldo_ldx_sp64"): Likewise.
25266 ("*mov<VM32:mode>_insn"): Likewise.
25267 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25268 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25270 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25272 * config/sparc/sparc.md ("type"): New insn type viscmp.
25273 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
25275 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
25276 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
25277 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25278 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
25280 ("n7_vis_logical_11cycle"): Likewise.
25281 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
25282 * config/sparc/niagara2.md ("niag3_vis": Likewise.
25283 * config/sparc/niagara.md ("niag_vis"): Likewise.
25284 * config/sparc/ultra3.md ("us3_fga"): Likewise.
25285 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
25287 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25289 * config/sparc/sparc.md: New instruction type `bmask'.
25290 (bmaskdi_vis): Use the `bmask' type.
25291 (bmasksi_vis): Likewise.
25292 * config/sparc/ultra3.md (us3_array): Likewise.
25293 * config/sparc/niagara7.md (n7_array): Likewise.
25294 * config/sparc/niagara4.md (n4_array): Likewise.
25295 * config/sparc/niagara2.md (niag2_vis): Likewise.
25296 (niag3_vis): Likewise.
25297 * config/sparc/niagara.md (niag_vis): Likewise.
25299 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25301 * ipa-comdats.c: Remove optimize check from gate.
25302 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
25303 for functions not optimized.
25304 (ipa_fn_summary_read): Skip optimize check.
25305 (ipa_fn_summary_write): Likewise.
25306 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
25308 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
25310 (can_inline_edge_p): Check flag_pcc_struct_return for match.
25311 (check_callers): Give up on caller which is not optimized.
25312 (inline_small_functions): Likewise.
25313 (ipa_inline): Do not give up when not optimizing.
25314 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
25315 away unoptimizes cdtors.
25316 (whole_program_function_and_variable_visibility): Do
25317 ipa_discover_readonly_nonaddressable_vars in LTO mode.
25318 * ipa.c (process_references): Do not check optimize.
25319 (symbol_table::remove_unreachable_nodes): Update optimize check.
25320 (set_writeonly_bit): Update optimize check.
25321 (pass_ipa_cdtor_merge::gate): Do not check optimize.
25322 (pass_ipa_single_use::gate): Remove.
25324 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25326 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
25327 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
25328 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
25329 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
25330 permute_load, permute_store, adjust_extract, adjust_splat,
25331 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
25332 replace_swap_with_copy, dump_swap_insn_table,
25333 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
25334 recombine_lvx_pattern, recombine_stvx_pattern,
25335 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
25336 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
25337 to file rs6000-p8swap.c.
25338 * config/rs6000/rs6000-p8swap.c: New file.
25339 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
25340 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
25341 and rs6000*-*-* targets.
25343 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25345 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
25347 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25349 * lto-wrapper.c (merge_and_complain): Do not merge
25350 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
25351 fsigned_zeros, ftrapping_math, fwrapv.
25352 (append_compiler_options): Do not track these options.
25353 (append_linker_options): Likewie
25355 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25357 * cgraphunit.c (cgraph_node::finalize_function): When
25358 !flag_toplevel_reorde set no_reorder flag.
25359 (varpool_node::finalize_decl): Likewise.
25360 (symbol_table::compile): Drop no toplevel reorder path.
25362 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25364 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
25365 edges; zero probability is not better than uninitialized.
25367 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25369 * asan.h (asan_sanitize_allocas_p): Declare.
25370 * asan.c (asan_sanitize_allocas_p): New function.
25371 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
25372 (handle_builtin_alloca): Likewise.
25373 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
25374 if !asan_sanitize_allocas_p.
25375 * params.def (asan-instrument-allocas): Add new option.
25376 * params.h (ASAN_PROTECT_ALLOCAS): Define.
25377 * opts.c (common_handle_option): Disable allocas sanitization for
25380 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25382 * asan.c: Include gimple-fold.h.
25383 (get_last_alloca_addr): New function.
25384 (handle_builtin_stackrestore): Likewise.
25385 (handle_builtin_alloca): Likewise.
25386 (asan_emit_allocas_unpoison): Likewise.
25387 (get_mem_refs_of_builtin_call): Add new parameter, remove const
25388 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
25389 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
25390 (instrument_builtin_call): Pass gimple iterator to
25391 get_mem_refs_of_builtin_call.
25392 (last_alloca_addr): New global.
25393 * asan.h (asan_emit_allocas_unpoison): Declare.
25394 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
25395 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
25396 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
25397 if function calls alloca.
25398 * gimple-fold.c (replace_call_with_value): Remove static keyword.
25399 * gimple-fold.h (replace_call_with_value): Declare.
25400 * internal-fn.c: Include asan.h.
25401 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
25402 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
25404 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25406 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
25407 (C_SELFTEST_FLAGS): New.
25408 (CPP_SELFTEST_FLAGS): New.
25409 (SELFTEST_DEPS): New, from deps of s-selftest.
25410 (C_SELFTEST_DEPS): New, from deps of s-selftest.
25411 (CPP_SELFTEST_DEPS): New.
25412 (selftest): Add dependency on s-selftest-c++.
25413 (s-selftest): Rename to...
25414 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
25415 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
25416 than SELFTEST_FLAGS.
25417 (selftest-gdb): Rename to...
25418 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
25420 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
25421 (selftest-valgrind): Rename to...
25422 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
25424 (selftest-valgrind): Reintroduce as an alias for
25425 selftest-c-valgrind.
25426 (s-selftest-c++): New.
25427 (selftest-c++-gdb): New.
25428 (selftest-c++-valgrind): New.
25430 2017-07-06 Olivier Hainque <hainque@adacore.com>
25432 * gcc.c (process_command): When deciding if undefined variables
25433 should be ignored when processing specs, accept "gcc -v" as well.
25435 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25437 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
25438 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
25440 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25442 * config/arm/arm-cpus.in (armv8-r): Add new entry.
25443 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
25444 * config/arm/arm-tables.opt: Regenerate.
25445 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
25447 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
25449 2017-07-06 Carl Love <cel@us.ibm.com>
25451 * ChangeLog: Clean up from mid air collision
25453 2017-07-06 Carl Love <cel@us.ibm.com>
25455 * config/rs6000/rs6000-c.c: Add support for built-in functions
25456 vector signed int vec_subc (vector signed int, vector signed int);
25457 vector signed __int128 vec_subc (vector signed __int128,
25458 vector signed __int128);
25459 vector unsigned __int128 vec_subc (vector unsigned __int128,
25460 vector unsigned __int128);
25461 vector signed int vec_sube (vector signed int, vector signed int,
25462 vector signed int);
25463 vector unsigned int vec_sube (vector unsigned int,
25464 vector unsigned int,
25465 vector unsigned int);
25466 vector signed __int128 vec_sube (vector signed __int128,
25467 vector signed __int128,
25468 vector signed__int128);
25469 vector unsigned __int128 vec_sube (vector unsigned __int128,
25470 vector unsigned __int128,
25471 vector unsigned __int128);
25472 vector signed int vec_subec (vector signed int, vector signed int,
25473 vector signed int);
25474 vector unsigned int vec_subec (vector unsigned int,
25475 vector unsigned int,
25476 vector unsigned int);
25477 vector signed __int128 vec_subec (vector signed __int128,
25478 vector signed __int128,
25479 vector signed__int128);
25480 vector unsigned __int128 vec_subec (vector unsigned __int128,
25481 vector unsigned __int128,
25482 vector unsigned __int128);
25483 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
25484 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
25485 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
25486 BU_ALTIVEC_OVERLOAD_X definitions.
25487 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
25488 * doc/extend.texi: Update the built-in documentation file for the new
25489 built-in functions.
25491 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25494 * diagnostic-show-locus.c (layout::layout): Use start and finish
25495 spelling location for the start and finish of each range.
25496 * genmatch.c (linemap_client_expand_location_to_spelling_point):
25497 Add unused aspect param.
25498 * input.c (expand_location_1): Add "aspect" param, and use it
25499 to access the correct part of the location.
25500 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
25502 (expand_location_to_spelling_point): Likewise.
25503 (linemap_client_expand_location_to_spelling_point): Add "aspect"
25504 param, and pass it to expand_location_1.
25506 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
25508 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
25509 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
25510 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
25511 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
25512 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
25513 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
25514 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
25515 _mm_maskz_getmant_ss): New intrinsics.
25516 (__builtin_ia32_getexpss128_mask): Changed to ...
25517 __builtin_ia32_getexpss128_round ... this.
25518 (__builtin_ia32_getexpsd128_mask): Changed to ...
25519 __builtin_ia32_getexpsd128_round ... this.
25520 * config/i386/i386-builtin-types.def
25521 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
25522 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
25523 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
25524 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
25525 __builtin_ia32_getmantss_mask_round): New builtins.
25526 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
25527 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
25528 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
25529 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
25530 * config/i386/sse.md
25531 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
25532 avx512f_sgetexp<mode><mask_scalar_name>
25533 <round_saeonly_scalar_name> ... this.
25534 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
25535 %0, %1, %2<round_saeonly_op3>}): Changed to ...
25536 vgetexp<ssescalarmodesuffix>
25537 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25538 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
25539 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
25540 avx512f_vgetmant<mode><mask_scalar_name>
25541 <round_saeonly_scalar_name> ... this.
25542 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
25543 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
25544 vgetmant<ssescalarmodesuffix>
25545 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
25546 %0<mask_scalar_operand4>, %1, %2
25547 <round_saeonly_scalar_mask_op4>, %3} ... this.
25548 * config/i386/subst.md (mask_scalar_operand4,
25549 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
25550 round_saeonly_scalar_nimm_predicate): New subst attributes.
25552 2017-07-06 Julia Koval <julia.koval@intel.com>
25554 * config/i386/i386.c (ix86_erase_embedded_rounding):
25555 Remove code for old rounding pattern.
25557 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
25559 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
25561 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
25563 * doc/sourcebuild.texi (Test Directives, Variants of
25564 dg-require-support): Add documentation for dg-require-stack-check.
25566 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
25568 * config/i386/subst.md (mask_scalar, round_scalar,
25569 round_saeonly_scalar): New meta-templates.
25570 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
25571 round_scalar_mask_operand3, round_scalar_mask_op3,
25572 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
25573 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
25574 round_saeonly_scalar_constraint,
25575 round_saeonly_scalar_prefix): New subst attribute.
25576 * config/i386/sse.md
25577 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
25578 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
25579 <round_scalar_name> ... this.
25580 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
25581 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
25582 <round_scalar_name> ... this.
25583 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
25584 <sse>_vm<code><mode>3<mask_scalar_name>
25585 <round_saeonly_scalar_name> ... this.
25586 (v<plusminus_mnemonic><ssescalarmodesuffix>
25587 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25588 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25589 v<plusminus_mnemonic><ssescalarmodesuffix>
25590 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25591 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25592 (v<multdiv_mnemonic><ssescalarmodesuffix>
25593 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25594 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25595 v<multdiv_mnemonic><ssescalarmodesuffix>
25596 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25597 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25598 (v<maxmin_float><ssescalarmodesuffix>
25599 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
25600 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
25601 v<maxmin_float><ssescalarmodesuffix>
25602 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25603 %0<mask_scalar_operand3>, %1, %<iptr>2
25604 <round_saeonly_scalar_mask_op3>} ... this.
25606 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
25608 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
25609 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
25611 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
25612 Alan Hayward <alan.hayward@arm.com>
25613 David Sherwood <david.sherwood@arm.com>
25615 * combine.c (simplify_if_then_else): Remove "enum" before
25617 * compare-elim.c (can_eliminate_compare): Likewise.
25618 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
25620 (aarch64_lookup_simd_builtin_type): Likewise.
25621 (aarch64_simd_builtin_type): Likewise.
25622 (aarch64_init_simd_builtin_types): Likewise.
25623 (aarch64_simd_expand_args): Likewise.
25624 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
25626 (aarch64_reverse_mask): Likewise.
25627 (aarch64_simd_emit_reg_reg_move): Likewise.
25628 (aarch64_gen_adjusted_ldpstp): Likewise.
25629 (aarch64_ccmp_mode_to_code): Likewise.
25630 (aarch64_operands_ok_for_ldpstp): Likewise.
25631 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25632 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
25634 (aarch64_min_divisions_for_recip_mul): Likewise.
25635 (aarch64_reassociation_width): Likewise.
25636 (aarch64_get_condition_code_1): Likewise.
25637 (aarch64_simd_emit_reg_reg_move): Likewise.
25638 (aarch64_simd_attr_length_rglist): Likewise.
25639 (aarch64_reverse_mask): Likewise.
25640 (aarch64_operands_ok_for_ldpstp): Likewise.
25641 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25642 (aarch64_gen_adjusted_ldpstp): Likewise.
25643 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
25645 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
25646 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
25647 (arm_lookup_simd_builtin_type): Likewise.
25648 (arm_simd_builtin_type): Likewise.
25649 (arm_init_simd_builtin_types): Likewise.
25650 (arm_expand_builtin_args): Likewise.
25651 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
25652 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
25653 (ft32_setup_incoming_varargs): Likewise.
25654 (ft32_function_arg): Likewise.
25655 (ft32_function_arg_advance): Likewise.
25656 (ft32_pass_by_reference): Likewise.
25657 (ft32_arg_partial_bytes): Likewise.
25658 (ft32_valid_pointer_mode): Likewise.
25659 (ft32_addr_space_pointer_mode): Likewise.
25660 (ft32_addr_space_legitimate_address_p): Likewise.
25661 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
25663 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
25664 (ix86_emit_outlined_ms2sysv_restore): Likewise.
25665 (iamcu_alignment): Likewise.
25666 (canonicalize_vector_int_perm): Likewise.
25667 (ix86_noce_conversion_profitable_p): Likewise.
25668 (ix86_mpx_bound_mode): Likewise.
25669 (ix86_operands_ok_for_move_multiple): Likewise.
25670 * config/microblaze/microblaze-protos.h
25671 (microblaze_expand_conditional_branch_reg): Likewise.
25672 * config/microblaze/microblaze.c
25673 (microblaze_expand_conditional_branch_reg): Likewise.
25674 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
25676 (rs6000_reassociation_width): Likewise.
25677 (rs6000_invalid_binary_op): Likewise.
25678 (fusion_p9_p): Likewise.
25679 (emit_fusion_p9_load): Likewise.
25680 (emit_fusion_p9_store): Likewise.
25681 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
25683 (riscv_hard_regno_mode_ok_p): Likewise.
25684 (riscv_address_insns): Likewise.
25685 (riscv_split_symbol): Likewise.
25686 (riscv_legitimize_move): Likewise.
25687 (riscv_function_value): Likewise.
25688 (riscv_hard_regno_nregs): Likewise.
25689 (riscv_expand_builtin): Likewise.
25690 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
25691 (riscv_build_integer): Likewise.
25692 (riscv_split_integer): Likewise.
25693 (riscv_legitimate_constant_p): Likewise.
25694 (riscv_cannot_force_const_mem): Likewise.
25695 (riscv_regno_mode_ok_for_base_p): Likewise.
25696 (riscv_valid_base_register_p): Likewise.
25697 (riscv_valid_offset_p): Likewise.
25698 (riscv_valid_lo_sum_p): Likewise.
25699 (riscv_classify_address): Likewise.
25700 (riscv_legitimate_address_p): Likewise.
25701 (riscv_address_insns): Likewise.
25702 (riscv_load_store_insns): Likewise.
25703 (riscv_force_binary): Likewise.
25704 (riscv_split_symbol): Likewise.
25705 (riscv_force_address): Likewise.
25706 (riscv_legitimize_address): Likewise.
25707 (riscv_move_integer): Likewise.
25708 (riscv_legitimize_const_move): Likewise.
25709 (riscv_legitimize_move): Likewise.
25710 (riscv_address_cost): Likewise.
25711 (riscv_subword): Likewise.
25712 (riscv_output_move): Likewise.
25713 (riscv_canonicalize_int_order_test): Likewise.
25714 (riscv_emit_int_order_test): Likewise.
25715 (riscv_function_arg_boundary): Likewise.
25716 (riscv_pass_mode_in_fpr_p): Likewise.
25717 (riscv_pass_fpr_single): Likewise.
25718 (riscv_pass_fpr_pair): Likewise.
25719 (riscv_get_arg_info): Likewise.
25720 (riscv_function_arg): Likewise.
25721 (riscv_function_arg_advance): Likewise.
25722 (riscv_arg_partial_bytes): Likewise.
25723 (riscv_function_value): Likewise.
25724 (riscv_pass_by_reference): Likewise.
25725 (riscv_setup_incoming_varargs): Likewise.
25726 (riscv_print_operand): Likewise.
25727 (riscv_elf_select_rtx_section): Likewise.
25728 (riscv_save_restore_reg): Likewise.
25729 (riscv_for_each_saved_reg): Likewise.
25730 (riscv_register_move_cost): Likewise.
25731 (riscv_hard_regno_mode_ok_p): Likewise.
25732 (riscv_hard_regno_nregs): Likewise.
25733 (riscv_class_max_nregs): Likewise.
25734 (riscv_memory_move_cost): Likewise.
25735 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
25736 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
25737 (rl78_addr_space_address_mode): Likewise.
25738 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
25740 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
25741 (rs6000_reassociation_width): Likewise.
25742 (rs6000_invalid_binary_op): Likewise.
25743 (fusion_p9_p): Likewise.
25744 (emit_fusion_p9_load): Likewise.
25745 (emit_fusion_p9_store): Likewise.
25746 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
25747 (ok_for_simple_move_operands): Likewise.
25748 (ok_for_simple_move_strict_operands): Likewise.
25749 (ok_for_simple_arith_logic_operands): Likewise.
25750 (visium_legitimize_reload_address): Likewise.
25751 (visium_select_cc_mode): Likewise.
25752 (output_cbranch): Likewise.
25753 (visium_split_double_move): Likewise.
25754 (visium_expand_copysign): Likewise.
25755 (visium_expand_int_cstore): Likewise.
25756 (visium_expand_fp_cstore): Likewise.
25757 * config/visium/visium.c (visium_pass_by_reference): Likewise.
25758 (visium_function_arg): Likewise.
25759 (visium_function_arg_advance): Likewise.
25760 (visium_libcall_value): Likewise.
25761 (visium_setup_incoming_varargs): Likewise.
25762 (visium_legitimate_constant_p): Likewise.
25763 (visium_legitimate_address_p): Likewise.
25764 (visium_legitimize_address): Likewise.
25765 (visium_secondary_reload): Likewise.
25766 (visium_register_move_cost): Likewise.
25767 (visium_memory_move_cost): Likewise.
25768 (prepare_move_operands): Likewise.
25769 (ok_for_simple_move_operands): Likewise.
25770 (ok_for_simple_move_strict_operands): Likewise.
25771 (ok_for_simple_arith_logic_operands): Likewise.
25772 (visium_function_value_1): Likewise.
25773 (rtx_ok_for_offset_p): Likewise.
25774 (visium_legitimize_reload_address): Likewise.
25775 (visium_split_double_move): Likewise.
25776 (visium_expand_copysign): Likewise.
25777 (visium_expand_int_cstore): Likewise.
25778 (visium_expand_fp_cstore): Likewise.
25779 (visium_split_cstore): Likewise.
25780 (visium_select_cc_mode): Likewise.
25781 (visium_split_cbranch): Likewise.
25782 (output_cbranch): Likewise.
25783 (visium_print_operand_address): Likewise.
25784 * expmed.c (flip_storage_order): Likewise.
25785 * expmed.h (emit_cstore): Likewise.
25786 (flip_storage_order): Likewise.
25787 * genrecog.c (validate_pattern): Likewise.
25788 * hsa-gen.c (gen_hsa_addr): Likewise.
25789 * internal-fn.c (expand_arith_overflow): Likewise.
25790 * ira-color.c (allocno_copy_cost_saving): Likewise.
25791 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25792 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25793 (process_invariant_for_inheritance): Likewise.
25794 * lra-eliminations.c (move_plus_up): Likewise.
25795 * omp-low.c (lower_oacc_reductions): Likewise.
25796 * simplify-rtx.c (simplify_subreg): Likewise.
25797 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25798 (TARGET_CHKP_BOUND_MODE): Likewise..
25799 * targhooks.c (default_chkp_bound_mode): Likewise.
25800 (default_setup_incoming_vararg_bounds): Likewise.
25801 * targhooks.h (default_chkp_bound_mode): Likewise.
25802 (default_setup_incoming_vararg_bounds): Likewise.
25803 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25804 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25805 (have_whole_vector_shift): Likewise.
25806 * tree-vect-stmts.c (vectorizable_load): Likewise.
25807 * doc/tm.texi: Regenerate.
25809 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25811 Graceful degrade if Binutils PR21472 is not available.
25814 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25815 .rodata in flash test fails.
25816 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25817 * confgure: Regenerate.
25818 * config.in: Regenerate.
25819 * config/avr/avr.c (avr_asm_named_section)
25820 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25821 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25822 (avr_asm_init_sections): Same.
25824 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25826 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25827 (fma<VH:mode>4_intrinsic): Likewise.
25828 (*fmsub<VCVTF:mode>4): Likewise.
25829 (*fmsub<VH:mode>4_intrinsic): Likewise.
25831 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25834 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25835 Don't depend on "optimize > 0".
25836 (out_movhi_r_mr, out_movqi_mr_r): Same.
25837 (out_movhi_mr_r, out_movqi_r_mr): Same.
25838 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25839 io_address_operand on "optimize > 0".
25841 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25843 * tree-loop-distribution.c: Add general explanantion on the pass.
25844 (generate_loops_for_partition): Mark distributed loop.
25845 (pg_add_dependence_edges): New parameter. Handle alias data
25846 dependence specially and record it in the parameter if asked.
25847 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25848 (init_partition_graph_vertices, add_partition_graph_edge): New.
25849 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25850 (free_partition_graph_vdata, build_partition_graph): New.
25851 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25852 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25853 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25854 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25855 (version_for_distribution_p, finalize_partitions): New.
25856 (distribute_loop): Handle alias data dependence specially. Factor
25857 out loop fusion code as functions and call these functions.
25859 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25861 * tree-loop-distribution.c (classify_partition): New parameter and
25862 better handle reduction statement.
25863 (rdg_build_partitions): Revise comment.
25864 (distribute_loop): Compute statements in all partitions and pass it
25865 to classify_partition.
25867 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25869 * tree-loop-distribution.c (enum partition_type): New.
25870 (struct partition): New field type.
25871 (partition_merge_into): Add parameter. Update partition type.
25872 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25873 (build_rdg_partition_for_vertex): Compute partition type.
25874 (rdg_build_partitions): Dump partition type.
25875 (distribute_loop): Update calls to partition_merge_into.
25877 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25879 * tree-loop-distribution.c (struct ddr_hasher): New.
25880 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25882 (classify_partition): Call get_data_dependence.
25883 (pg_add_dependence_edges): Ditto.
25884 (distribute_loop): Release data dependence hash table.
25886 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25888 * tree-loop-distribution.c (ref_base_address): Delete.
25889 (similar_memory_accesses): Rename ...
25890 (share_memory_accesses): ... to this. Check if partitions access
25891 the same memory reference.
25892 (distribute_loop): Call share_memory_accesses.
25894 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25896 * tree-loop-distribution.c (struct partition): New field recording
25897 its data reference.
25898 (partition_alloc, partition_free): Init and release data refs.
25899 (partition_merge_into): Merge data refs.
25900 (build_rdg_partition_for_vertex): Collect data refs for partition.
25901 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25903 (distribute_loop): Remve data refs from vertice data of partition
25906 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25908 * tree-loop-distribution.c (params.h): Include header file.
25909 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25910 (datarefs_vec): New global var.
25911 (create_rdg_vertices): Use datarefs_vec directly.
25912 (free_rdg): Don't free data references.
25913 (build_rdg): Update use. Don't free data references.
25914 (distribute_loop): Compute global variable for data references.
25915 Bail out if there are too many data references.
25917 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25919 * tree-loop-distribution.c (loop_nest): New global var.
25920 (build_rdg): Use loop directly, rather than loop nest.
25921 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25923 (distribute_loop): Compute global variable loop nest. Update use.
25925 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25927 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25928 (partition_merge_into): New parameter. Dump reason for fusion.
25929 (distribute_loop): Update use of partition_merge_into.
25931 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25933 * tree-loop-distribution.c (bb_top_order_index): New.
25934 (bb_top_order_index_size, bb_top_order_cmp): New.
25935 (stmts_from_loop): Use topological order.
25936 (pass_loop_distribution::execute): Compute and release topological
25937 order for basic blocks.
25939 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25941 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25944 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25946 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25947 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25948 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25949 * internal-fn.def (LOOP_DIST_ALIAS): New.
25950 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25951 (fold_loop_internal_call): ... this.
25952 (vect_loop_dist_alias_call): New function.
25953 (set_uid_loop_bbs): Call fold_loop_internal_call.
25954 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25957 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25960 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25961 Require dead FLAGS_REG at the beginning of a peephole.
25963 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25966 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25967 arguments in the call to __builtin_ia32_sbb_u32.
25968 (_subborrow_u64): Swap _X and _Y arguments in the call to
25969 __builtin_ia32_sbb_u64.
25971 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25974 * tree-vrp.c (compare_assert_loc): Turn into a function template
25975 with stable template parameter. Only test if a->e is NULL,
25976 !a->e == !b->e has been verified already. Use e == NULL or
25977 e != NULL instead of e or ! e tests. If stable is true, don't use
25978 iterative_hash_expr, on the other side allow a or b or both NULL
25979 and sort the NULLs last.
25980 (process_assert_insertions): Sort using compare_assert_loc<false>
25981 instead of compare_assert_loc, later sort using
25982 compare_assert_loc<true> before calling process_assert_insertions_for
25983 in a loop. Use break instead of continue once seen NULL pointer.
25985 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25987 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25988 Cortex-R7 and Cortex-R8 processors.
25990 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25992 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25993 uninitialized while src is not.
25995 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25997 * common/config/arm/arm-common.c: Adjust include path for
25999 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
26000 (arm-cpu.h): Create in build directory. Adjust dependency rules.
26001 (arm-cpu-data.h): Likewise.
26002 (arm-cpu-cdata.h): Likewise.
26003 * config/arm/arm-cpu.h: Delete.
26004 * config/arm/arm-cpu-cdata.h: Delete.
26005 * config/arm/arm-cpu-data.h: Delete.
26007 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
26009 * config/arm/arm-cpus.in (cortex-a55): New.
26010 (cortex-a75): Likewise.
26011 (cortex-a75.cortex-a55): Likewise.
26012 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
26014 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
26015 * config/arm/arm-cpu-cdata.h: Regenerate.
26016 * config/arm/arm-cpu-data.h: Regenerate.
26017 * config/arm/arm-cpu.h: Regenerate.
26018 * config/arm/arm-tables.opt: Regenerate.
26019 * config/arm/arm-tune.md: Regenerate.
26021 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
26023 * haifa-sched.c (sched_create_recovery_edges): Update profile.
26025 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
26027 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
26030 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
26032 PR tree-optimization/81292
26033 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
26034 full_string_p, also call adjust_related_strinfos if the adjustment
26035 is simple, otherwise invalidate related strinfos.
26037 2017-07-04 Martin Liska <mliska@suse.cz>
26040 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
26041 newly created variable as DECL_IGNORED_P.
26043 2017-07-04 Martin Liska <mliska@suse.cz>
26046 * ipa-inline.c (inline_small_functions):
26047 Use xstrdup_for_dump.
26049 2017-07-04 Tom de Vries <tom@codesourcery.com>
26051 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
26053 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
26056 * config/darwin.c (darwin_function_switched_text_sections):
26059 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
26061 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
26063 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
26065 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
26067 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26069 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
26070 min_profitable_iters, and th as inclusive lower bounds.
26071 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
26072 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
26073 for min_profitable_iters and min_profitable_estimate.
26074 (vect_transform_loop): Treat th as an inclusive lower bound.
26075 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
26077 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
26080 * config/darwin.c (darwin_function_switched_text_sections):
26081 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
26082 in two pieces, and suppress the use of buf.
26084 2017-07-03 Nathan Sidwell <nathan@acm.org>
26086 * hash-table.h (hash_table_mod1): Fix indentation.
26088 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26090 PR middle-end/81290
26091 * predict.c (force_edge_cold): Be more careful about propagation
26093 * profile-count.h (profile_probability::guessed,
26094 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
26096 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
26098 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
26100 * doc/invoke.texi (rcpc architecture extension): Document it.
26102 2017-07-03 Richard Biener <rguenther@suse.de>
26104 PR tree-optimization/60510
26105 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
26106 the scalar reduction PHI and use it.
26107 (vectorizable_reduction): Properly guard the single_defuse_cycle
26108 path for non-SLP reduction chains where we cannot use it.
26109 Rework reduc_def/index and vector type deduction. Rework
26110 vector operand gathering during reduction op code-gen.
26111 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
26112 chains dissolve the chain and leave it to non-SLP reduction
26115 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26117 * tree-data-ref.h (dr_alignment): Declare.
26118 * tree-data-ref.c (dr_alignment): New function.
26119 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
26120 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
26122 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
26124 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26126 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
26127 and base_misalignment fields.
26128 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
26129 * tree-data-ref.c: Include builtins.h.
26130 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
26131 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
26132 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
26133 * tree-vect-data-refs.c: Include tree-cfg.h.
26134 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
26135 fields instead of calculating an alignment here.
26136 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
26137 innermost_loop_behavior fields.
26139 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26141 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
26143 (DR_STEP_ALIGNMENT): New macro.
26144 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
26145 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
26146 (create_data_ref): Print it.
26147 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
26148 to tell whether the step preserves vector (mis)alignment.
26149 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26150 Move the check for an integer step and generalise to all INTEGER_CST.
26151 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
26152 Print the outer step alignment.
26154 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26156 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
26157 with offset_alignment.
26158 (DR_ALIGNED_TO): Delete.
26159 (DR_OFFSET_ALIGNMENT): New macro.
26160 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
26161 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
26162 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
26163 (create_data_ref): Likewise.
26164 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26165 (vect_analyze_data_refs): Likewise.
26166 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
26167 creating dummy innermost behavior.
26169 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26171 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
26172 with a "innermost_loop_behavior *" and refeence tree.
26173 * tree-data-ref.c (dr_analyze_innermost): Likewise.
26174 (create_data_ref): Update call accordingly.
26175 * tree-predcom.c (find_looparound_phi): Likewise.
26177 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26179 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
26180 fields with dr_wrt_vec_loop.
26181 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
26182 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
26183 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
26184 (vect_dr_behavior): New function.
26185 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26186 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
26187 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
26188 track whether the step preserves the misalignment.
26189 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26190 Use vect_dr_behavior.
26191 (vect_setup_realignment): Update call accordingly.
26192 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
26193 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
26194 call to vect_create_addr_base_for_vector_ref.
26195 (vect_create_cond_for_align_checks): Likewise.
26196 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
26197 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
26198 (vect_recog_mask_conversion_pattern): Likewise.
26199 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
26200 (new_stmt_vec_info): Remove redundant zeroing.
26202 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
26204 * common/config/arm/arm-common.c (arm_be8_option): New function.
26205 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
26206 (ISA_ARMv6): Add isa_bit_be8.
26207 * config/arm/arm.h (arm_be8_option): Add prototype.
26208 (BE8_SPEC_FUNCTION): New define.
26209 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
26210 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
26211 (mlittle-endian): Similarly.
26212 (mbe8, mbe32): New options.
26213 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
26214 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
26216 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26218 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
26220 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26222 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
26223 (cleanup_tree_cfg_bb): Use it.
26224 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
26226 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
26228 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26231 * loop-doloop.c (add_test): Update profile.
26233 2017-07-03 Martin Liska <mliska@suse.cz>
26236 * sanopt.c (rewrite_usage_of_param): New function.
26237 (sanitize_rewrite_addressable_params): Likewise.
26238 (pass_sanopt::execute): Call rewrite_usage_of_param.
26240 2017-07-03 Richard Biener <rguenther@suse.de>
26242 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
26243 back to using VIEW_CONVERT_EXPR.
26245 2017-07-03 Martin Liska <mliska@suse.cz>
26248 * doc/extend.texi: Document when a resolver function is
26249 generated for target_clones.
26251 2017-07-03 Martin Liska <mliska@suse.cz>
26253 * asan.c (asan_emit_stack_protection): Unpoison just red zones
26254 and shadow memory of auto variables which are subject of
26255 use-after-scope sanitization.
26256 (asan_expand_mark_ifn): Add do set only when is_poison.
26258 2016-07-03 Richard Biener <rguenther@suse.de>
26260 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
26262 (vect_force_simple_reduction): Record reduction def -> phi mapping.
26263 (vectorizable_reduction): Perform reduction PHI creation when
26264 visiting a reduction PHI and adjust and simplify code generation
26265 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
26266 (vect_transform_loop): Visit reduction PHIs.
26267 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
26268 defs into the SLP tree.
26269 (vect_build_slp_tree): Reduction defs terminate the recursion.
26270 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
26272 (vect_get_vec_defs_for_stmt_copy): Export.
26273 (vect_get_vec_defs): Likewise.
26274 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
26276 (vect_get_vec_defs_for_stmt_copy): Declare.
26277 (vect_get_vec_defs): Likewise.
26279 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26281 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
26282 parameter with a "loop" parameter and use it instead of the
26283 loop containing DR_STMT. Don't check simple_iv when doing
26284 BB analysis. Describe the two analysis modes in the comment.
26286 2017-07-03 Tom de Vries <tom@codesourcery.com>
26288 PR tree-optimization/69468
26289 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
26290 (find_same_succ_bb): Handle ignore_edge_flags.
26292 2017-07-03 Tom de Vries <tom@codesourcery.com>
26294 PR tree-optimization/81192
26295 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
26297 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
26299 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
26301 2017-07-03 Tom de Vries <tom@codesourcery.com>
26303 PR tree-optimization/81192
26304 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
26305 BB_SAME_SUCC (bb) == NULL.
26307 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26309 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
26312 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26314 * dumpfile.c: Include profile-count.h
26315 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
26317 (insert_cond_bb): Update profile.
26318 * tree-cfg.h (insert_cond_bb): Update prototype.
26319 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
26320 * tree-dump.c: Do not include tree-cfg.
26322 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26324 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
26326 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26328 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
26331 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26333 * tree-complex.c (expand_complex_div_wide): update profile.
26335 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26336 Alan Hayward <alan.hayward@arm.com>
26337 David Sherwood <david.sherwood@arm.com>
26339 * Makefile.in (MACHMODE_H): Remove insn-modes.h
26340 (CORETYPES_H): New define.
26341 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
26342 (insn-modes-inline.h, s-modes-inline-h): New rules.
26343 (generated_files): Add insn-modes-inline.h.
26344 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
26345 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
26346 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
26347 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
26348 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
26349 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
26350 (build/gencodes.o, build/genconditions.o): Likewise.
26351 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
26352 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
26353 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
26354 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
26355 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
26356 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
26357 * coretypes.h: Include everything up to real.h for generators.
26358 Include insn-modes.h first. Include wide-int-print.h after
26359 wide-int.h. Include insn-modes-inline.h and then machmode.h.
26360 * machmode.h: Don't include insn-modes.h here.
26361 * function-tests.c: Remove includes of signop.h, machmode.h,
26362 double-int.h and wide-int.h.
26364 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
26366 * optc-save-gen.awk: Likewise.
26367 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
26368 * godump.c: Remove include of wide-int-print.h.
26369 * pretty-print.h: Likewise.
26370 * wide-int-print.cc: Likewise.
26371 * wide-int.cc: Likewise.
26372 * hash-map-tests.c: Remove include of signop.h.
26373 * hash-set-tests.c: Likewise.
26374 * rtl-tests.c: Likewise.
26375 * mkconfig.sh: Remove include of machmode.h.
26376 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
26378 (emit_insn_modes_inline_h): ...this new function. Emit the code
26379 into an insn-modes-inline.h header file, adding appropriate
26380 include guards and end comments.
26381 (emit_insn_modes_c_header): Remove include of machmode.h.
26382 (emit_min_insn_modes_c_header): Include coretypes.h rather than
26384 (main): Handle -i flag and call emit_insn_modes_inline_h when
26387 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26389 * tree-ssa-strlen.c (strinfo): Rename the length field to
26390 nonzero_chars. Add a full_string_p field.
26391 (compare_nonzero_chars, zero_length_string_p): New functions.
26392 (get_addr_stridx): Add an offset_out parameter.
26393 Use compare_nonzero_chars.
26394 (get_stridx): Update accordingly. Use compare_nonzero_chars.
26395 (new_strinfo): Update after above changes to strinfo.
26396 (set_endptr_and_length): Set full_string_p.
26397 (get_string_length): Update after above changes to strinfo.
26398 (unshare_strinfo): Update call to new_strinfo.
26399 (maybe_invalidate): Likewise.
26400 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
26401 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
26402 as a uhwi instead of an shwi. Update after above changes to
26403 strinfo and new_strinfo.
26404 (zero_length_string): Assert that chainsi contains full strings.
26405 Use zero_length_string_p. Update call to new_strinfo.
26406 (adjust_related_strinfos): Update after above changes to strinfo.
26407 Copy full_string_p from origsi.
26408 (adjust_last_stmt): Use zero_length_string_p.
26409 (handle_builtin_strlen): Update after above changes to strinfo and
26410 new_strinfo. Install the lhs as the string length if the previous
26411 entry didn't describe a full string.
26412 (handle_builtin_strchr): Update after above changes to strinfo
26414 (handle_builtin_strcpy): Likewise.
26415 (handle_builtin_strcat): Likewise.
26416 (handle_builtin_malloc): Likewise.
26417 (handle_pointer_plus): Likewise.
26418 (handle_builtin_memcpy): Likewise. Track nonzero characters
26419 that aren't necessarily followed by a nul terminator.
26420 (handle_char_store): Likewise.
26422 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26424 PR tree-optimization/80769
26425 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
26426 for malloc and calloc. Document the new invariant that all related
26427 strinfos have delayed lengths or none do.
26428 (verify_related_strinfos): Move earlier in file.
26429 (set_endptr_and_length): New function, split out from...
26430 (get_string_length): ...here. Also set the lengths of related
26432 (zero_length_string): Assert that chainsi has known (rather than
26434 (adjust_related_strinfos): Likewise.
26436 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26438 PR tree-optimization/81136
26439 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
26440 assert that two references with the same misalignment have the same
26441 compile-time misalignment if those compile-time misalignments
26444 2017-07-01 Andi Kleen <ak@linux.intel.com>
26446 * print-tree.c (print_node): Print all attributes.
26448 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26450 * cfg.c (scale_bbs_frequencies): New function.
26451 * cfg.h (scale_bbs_frequencies): Declare it.
26452 * cfgloopanal.c (single_likely_exit): Cleanup.
26453 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
26455 (scale_loop_profile): Likewise.
26456 (loop_version): Likewise.
26457 (create_empty_loop_on_edge): Update.
26458 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
26459 scale_loop_frequencies, scale_loop_profile, loopify,
26460 loop_version): Update prototypes.
26461 * modulo-sched.c (sms_schedule): Update.
26462 * predict.c (unlikely_executed_edge_p): Also check probability.
26463 (probably_never_executed_edge_p): Fix typo.
26464 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26465 * tree-parloops.c (gen_parallel_loop): Update.
26466 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
26467 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26468 * tree-ssa-loop-split.c (split_loop): Update.
26469 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26470 * tree-vect-loop-manip.c (vect_do_peeling): Update.
26471 (vect_loop_versioning): Update.
26472 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26474 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26476 * trans-mem.c (split_bb_make_tm_edge): Update profile.
26478 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26480 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
26481 to keep profile consistent.
26483 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26485 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
26486 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
26487 * profile-count.h (max_safe_multiplier): Make unsigned.
26488 (profile_count::guessed_zero): New.
26490 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26492 * bb-reorder.c (fix_up_crossing_landing_pad,
26493 fix_crossing_conditional_branches): Use make_single_succ_edge
26494 to keep profile consistent.
26496 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26498 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
26501 2017-07-01 Jakub Jelinek <jakub@redhat.com>
26504 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
26505 the right scopes, make sure cond_jump isn't preserved between multiple
26506 iterations. Search for fallthru edge whenever there are 3+ edges and
26507 use find_fallthru_edge for it.
26509 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26511 Patch by Alexander Monakov <amonakov@ispras.ru>
26512 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
26513 probabilities consistently.
26515 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26517 * pa.c (pa_expand_compare_and_swap_loop): Update call of
26518 emit_cmp_and_jump_insns.
26520 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26523 * tree-inline.c (expand_call_inline): Combine profile statuses.
26525 2017-06-30 Andrew Pinski <apinski@cavium.com>
26527 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
26528 fold_stmt returned true.
26530 2017-06-30 Nathan Sidwell <nathan@acm.org>
26532 * ggc.h (empty_string): Delete.
26533 * cfgexpand.c (expand_asm_stmt): Use plain "".
26534 * optabs.c (expand_asm_memory_barrier): Likewise.
26535 * stringpool.c (empty_string): Delete.
26536 (digit_vector, digit_string): Delete.
26537 (ggc_alloc_string): Use plain "", don't optimize single digit
26538 strings. Use ggc_alloc_atomic.
26540 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
26542 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
26543 comparison set and one other set, use the cost of the non-comparison
26546 2017-06-30 Nathan Sidwell <nathan@acm.org>
26548 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
26551 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
26553 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
26554 loops. Remove now unneeded calls to gimple_switch_set_label() that
26555 just set removed labels to NULL_TREE.
26557 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
26559 * tree-ssanames.c (set_range_info_raw): Abstract from ...
26560 (set_range_info): ...here. Only call set_range_info_raw if domain
26562 (set_nonzero_bits): Call set_range_info_raw.
26563 * tree-ssanames.h (set_range_info_raw): New.
26565 2017-06-30 Jakub Jelinek <jakub@redhat.com>
26568 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
26569 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
26570 of nonimmediate_operand and <store_mask_constraint> instead of m
26571 for the input operand. For V8FI iterator, always split if input
26572 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
26573 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
26574 <store_mask_predicate> instead of register_operand and
26575 <store_mask_constraint> instead of v for the input operand. Make
26576 sure both operands aren't MEMs for if not <mask_applied>.
26578 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
26580 * lto-wrapper.c (copy_file) Close both file descriptors before
26583 2017-06-30 Martin Liska <mliska@suse.cz>
26586 * multiple_target.c (create_dispatcher_calls): Make ifunc
26587 also for function that don't have calls or are not referenced.
26589 2017-06-30 Richard Biener <rguenther@suse.de>
26591 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
26592 analyze the first scalar stmt. Move vector type computation
26593 for the BB case here from ...
26594 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
26595 live operation processing in the SLP case properly.
26597 2017-06-30 Richard Biener <rguenther@suse.de>
26599 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
26601 2017-06-30 Martin Liska <mliska@suse.cz>
26604 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
26605 before BUILT_IN_UNWIND_RESUME when ASAN is used.
26607 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
26609 * doc/invoke.texi (AArch64): Add missing options and remove redundant
26612 2017-06-30 Richard Biener <rguenther@suse.de>
26614 PR tree-optimization/81249
26615 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
26616 condition reduction result to original scalar type.
26618 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26620 * profile-count.h (enum profile_quality): Fix typos and whitespace
26623 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26625 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
26626 type for branch probabilities.
26628 2017-06-29 Julian Brown <julian@codesourcery.com>
26629 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26631 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
26632 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
26633 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
26634 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
26636 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26638 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
26639 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
26640 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
26641 CC usage from generic code to here.
26642 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
26643 CC usage into the target macros.
26645 2017-06-29 Maya Rashish <coypu@sdf.org>
26647 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
26650 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26652 * arm/arm-builtins.c: Include profile-count.h
26653 * except.c (sjlj_emit_function_enter): Use
26654 profile_probability::unlikely.
26656 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26658 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
26659 and tocrel_offset be pointer args rather than implicitly using
26661 (legitimate_constant_pool_address_p, rs6000_emit_move,
26662 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
26663 tocrel_offset and use in toc_relative_expr_p call.
26664 (print_operand, print_operand_address): Use static tocrel_base_oac
26665 and tocrel_offset_oac.
26666 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
26669 2017-06-29 Maya Rashish <coypu@sdf.org>
26671 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
26673 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
26675 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
26676 objects, take into account only the alignment of 'op0' and 'mode1' if
26679 2017-06-29 Steve Ellcey <sellcey@cavium.com>
26681 * ccmp.c (ccmp_tree_comparison_p): New function.
26682 (ccmp_candidate_p): Update to use above function.
26683 (get_compare_parts): New function.
26684 (expand_ccmp_next): Update to use new functions.
26685 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
26687 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
26688 take mode as argument.
26689 * ccmp.h (expand_ccmp_expr): Add mode as argument.
26690 * expr.c (expand_expr_real_1): Pass mode as argument.
26692 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
26694 * combine.c (combine_instructions): Print insns to dump_file, together
26697 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26699 * asan.c (asan_emit_stack_protection): Update.
26700 (create_cond_insert_point): Update.
26701 * auto-profile.c (afdo_propagate_circuit): Update.
26702 * basic-block.h (struct edge_def): Turn probability to
26703 profile_probability.
26704 (EDGE_FREQUENCY): Update.
26705 * bb-reorder.c (find_traces_1_round): Update.
26706 (better_edge_p): Update.
26707 (sanitize_hot_paths): Update.
26708 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
26709 (make_single_succ_edge): Update.
26710 (check_bb_profile): Update.
26711 (dump_edge_info): Update.
26712 (update_bb_profile_for_threading): Update.
26713 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
26714 probabilitycount to 0.
26715 * cfgbuild.c (compute_outgoing_frequencies): Update.
26716 * cfgcleanup.c (try_forward_edges): Update.
26717 (outgoing_edges_match): Update.
26718 (try_crossjump_to_edge): Update.
26719 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
26720 (expand_gimple_tailcall): Update.
26721 (construct_init_block): Use make_single_succ_edge.
26722 (construct_exit_block): Use make_single_succ_edge.
26723 * cfghooks.c (verify_flow_info): Update.
26724 (redirect_edge_succ_nodup): Update.
26725 (split_edge): Update.
26726 (account_profile_record): Update.
26727 * cfgloopanal.c (single_likely_exit): Update.
26728 * cfgloopmanip.c (scale_loop_profile): Update.
26729 (set_zero_probability): Remove.
26730 (duplicate_loop_to_header_edge): Update.
26731 * cfgloopmanip.h (loop_version): Update prototype.
26732 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
26733 (force_nonfallthru_and_redirect): Update.
26734 (update_br_prob_note): Update.
26735 (rtl_verify_edges): Update.
26736 (purge_dead_edges): Update.
26737 (rtl_lv_add_condition_to_bb): Update.
26738 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
26739 * cgraphunit.c (init_lowered_empty_function): Update.
26740 (cgraph_node::expand_thunk): Update.
26741 * cilk-common.c: Include profile-count.h
26742 * dojump.c (inv): Remove.
26743 (jumpifnot): Update.
26744 (jumpifnot_1): Update.
26745 (do_jump_1): Update.
26747 (do_jump_by_parts_greater_rtx): Update.
26748 (do_compare_rtx_and_jump): Update.
26749 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
26750 do_jump_1. do_compare_rtx_and_jump): Update prototype.
26751 * dwarf2cfi.c: Include profile-count.h
26752 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
26753 (sjlj_emit_dispatch_table): Likewise.
26754 * explow.c: Include profile-count.h
26755 * expmed.c (emit_store_flag_force): Update.
26756 (do_cmp_and_jump): Update.
26757 * expr.c (compare_by_pieces_d::generate): Update.
26758 (compare_by_pieces_d::finish_mode): Update.
26759 (emit_block_move_via_loop): Update.
26760 (store_expr_with_bounds): Update.
26761 (store_constructor): Update.
26762 (expand_expr_real_2): Update.
26763 (expand_expr_real_1): Update.
26764 * expr.h (try_casesi, try_tablejump): Update prototypes.
26765 * gimple-pretty-print.c (dump_probability): Update.
26766 (dump_profile): New.
26767 (dump_gimple_label): Update.
26768 (dump_gimple_bb_header): Update.
26769 * graph.c (draw_cfg_node_succ_edges): Update.
26770 * hsa-gen.c (convert_switch_statements): Update.
26771 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
26772 (find_if_case_1): Update.
26773 (find_if_case_2): Update.
26774 * internal-fn.c (expand_arith_overflow_result_store): Update.
26775 (expand_addsub_overflow): Update.
26776 (expand_neg_overflow): Update.
26777 (expand_mul_overflow): Update.
26778 (expand_vector_ubsan_overflow): Update.
26779 * ipa-cp.c (good_cloning_opportunity_p): Update.
26780 * ipa-split.c (split_function): Use make_single_succ_edge.
26781 * ipa-utils.c (ipa_merge_profiles): Update.
26782 * loop-doloop.c (add_test): Update.
26783 (doloop_modify): Update.
26784 * loop-unroll.c (compare_and_jump_seq): Update.
26785 (unroll_loop_runtime_iterations): Update.
26786 * lra-constraints.c (lra_inheritance): Update.
26787 * lto-streamer-in.c (input_cfg): Update.
26788 * lto-streamer-out.c (output_cfg): Update.
26789 * mcf.c (adjust_cfg_counts): Update.
26790 * modulo-sched.c (sms_schedule): Update.
26791 * omp-expand.c (expand_omp_for_init_counts): Update.
26792 (extract_omp_for_update_vars): Update.
26793 (expand_omp_ordered_sink): Update.
26794 (expand_omp_for_ordered_loops): Update.
26795 (expand_omp_for_generic): Update.
26796 (expand_omp_for_static_nochunk): Update.
26797 (expand_omp_for_static_chunk): Update.
26798 (expand_cilk_for): Update.
26799 (expand_omp_simd): Update.
26800 (expand_omp_taskloop_for_outer): Update.
26801 (expand_omp_taskloop_for_inner): Update.
26802 * omp-simd-clone.c (simd_clone_adjust): Update.
26803 * optabs.c (expand_doubleword_shift): Update.
26804 (expand_abs): Update.
26805 (emit_cmp_and_jump_insn_1): Update.
26806 (expand_compare_and_swap_loop): Update.
26807 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26808 * predict.c (predictable_edge_p): Update.
26809 (edge_probability_reliable_p): Update.
26810 (set_even_probabilities): Update.
26811 (combine_predictions_for_insn): Update.
26812 (combine_predictions_for_bb): Update.
26813 (propagate_freq): Update.
26814 (estimate_bb_frequencies): Update.
26815 (force_edge_cold): Update.
26816 * profile-count.c (profile_count::dump): Add missing space into dump.
26817 (profile_count::debug): Add newline.
26818 (profile_count::differs_from_p): Explicitly convert to unsigned.
26819 (profile_count::stream_in): Update.
26820 (profile_probability::dump): New member function.
26821 (profile_probability::debug): New member function.
26822 (profile_probability::differs_from_p): New member function.
26823 (profile_probability::differs_lot_from_p): New member function.
26824 (profile_probability::stream_in): New member function.
26825 (profile_probability::stream_out): New member function.
26826 * profile-count.h (profile_count_quality): Rename to ...
26827 (profile_quality): ... this one.
26828 (profile_probability): New.
26829 (profile_count): Update.
26830 * profile.c (compute_branch_probabilities): Update.
26831 * recog.c (peep2_attempt): Update.
26832 * sched-ebb.c (schedule_ebbs): Update.
26833 * sched-rgn.c (find_single_block_region): Update.
26834 (compute_dom_prob_ps): Update.
26835 (schedule_region): Update.
26836 * sel-sched-ir.c (compute_succs_info): Update.
26837 * stmt.c (struct case_node): Update.
26838 (do_jump_if_equal): Update.
26839 (get_outgoing_edge_probs): Update.
26840 (conditional_probability): Update.
26841 (emit_case_dispatch_table): Update.
26842 (expand_case): Update.
26843 (expand_sjlj_dispatch_table): Update.
26844 (emit_case_nodes): Update.
26845 * targhooks.c: Update.
26846 * tracer.c (better_p): Update.
26847 (find_best_successor): Update.
26848 * trans-mem.c (expand_transaction): Update.
26849 * tree-call-cdce.c: Update.
26850 * tree-cfg.c (gimple_split_edge): Upate.
26851 (move_sese_region_to_fn): Upate.
26852 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26853 * tree-eh.c (lower_resx): Upate.
26854 (cleanup_empty_eh_move_lp): Upate.
26855 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26856 * tree-inline.c (copy_edges_for_bb): Update.
26857 (copy_cfg_body): Update.
26858 * tree-parloops.c (gen_parallel_loop): Update.
26859 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26860 (gimple_gen_time_profiler): Update.
26861 * tree-ssa-dce.c (remove_dead_stmt): Update.
26862 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26863 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26864 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26865 (unloop_loops): Update.
26866 (try_peel_loop): Update.
26867 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26868 * tree-ssa-loop-split.c (connect_loops): Update.
26869 (split_loop): Update.
26870 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26871 (hoist_guard): Update.
26872 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26873 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26874 (value_replacement): Update.
26875 * tree-ssa-reassoc.c (branch_fixup): Update.
26876 * tree-ssa-tail-merge.c (replace_block_by): Update.
26877 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26878 (create_edge_and_update_destination_phis): Update.
26879 (compute_path_counts): Update.
26880 (recompute_probabilities): Update.
26881 (update_joiner_offpath_counts): Update.
26882 (freqs_to_counts_path): Update.
26883 (duplicate_thread_path): Update.
26884 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26885 (struct switch_conv_info): Update.
26886 (gen_inbound_check): Update.
26887 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26888 (vect_do_peeling): Update.
26889 (vect_loop_versioning): Update.
26890 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26891 (optimize_mask_stores): Update.
26892 * ubsan.c (ubsan_expand_null_ifn): Update.
26893 * value-prof.c (gimple_divmod_fixed_value): Update.
26894 (gimple_divmod_fixed_value_transform): Update.
26895 (gimple_mod_pow2): Update.
26896 (gimple_mod_pow2_value_transform): Update.
26897 (gimple_mod_subtract): Update.
26898 (gimple_mod_subtract_transform): Update.
26899 (gimple_ic): Update.
26900 (gimple_stringop_fixed_value): Update.
26901 (gimple_stringops_transform): Update.
26902 * value-prof.h: Update.
26904 2017-06-29 Carl Love <cel@us.ibm.com>
26906 * config/rs6000/rs6000-c.c: Add support for built-in functions
26907 vector signed int vec_signed (vector float);
26908 vector signed long long vec_signed (vector double);
26909 vector signed int vec_signed2 (vector double, vector double);
26910 vector signed int vec_signede (vector double);
26911 vector signed int vec_signedo (vector double);
26912 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26913 instruction generator.
26914 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26915 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26916 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26918 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26919 vunsignede_v2df): Add define_expands.
26920 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26921 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26922 VEC_UNSIGNEDO): Add definitions.
26923 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26924 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26925 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26926 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26927 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26928 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26929 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26930 vec_unsignede, vec_unsignedo): Add builtin defines.
26931 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26933 * doc/extend.texi: Update the built-in documentation file for the
26934 new built-in functions.
26936 2017-06-29 Richard Biener <rguenther@suse.de>
26938 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26939 reduction chains to LOOP_VINFO_REDUCTIONS.
26940 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26941 SLP reductions after processing reduction chains.
26943 2017-06-29 Nathan Sidwell <nathan@acm.org>
26945 * builtins.c (fold_builtin_FUNCTION): Use
26946 lang_hooks.decl_printable_name.
26948 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26950 PR middle-end/81194
26951 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26952 with only one label.
26953 * stmt.c (expand_case): Assert NCASES is greater than one.
26955 2017-06-29 Richard Biener <rguenther@suse.de>
26957 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26959 (group_case_labels): Likewise.
26960 (find_taken_edge): Push sanity checking on val to workers...
26961 (find_taken_edge_cond_expr): ... here
26962 (find_taken_edge_switch_expr): ... and here, handle cases
26963 with just a default label.
26964 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26965 (group_case_labels): Likewise.
26966 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26967 group_case_labels does anything cleanup the CFG again.
26969 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26971 PR tree-optimization/81196
26972 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26973 exit condition comparing two IVs.
26975 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26977 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26978 profile to the dummy entry at the end of the list of architectures.
26979 * config/arm/arm-cpu-cdata.h: Regenerated.
26981 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26982 Michael Collison <michael.collison@arm.com>
26985 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26987 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26988 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26989 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26990 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26991 of shift when the shift amount is masked with constant equal to
26992 the size of the mode.
26993 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26996 2017-06-29 Martin Liska <mliska@suse.cz>
26998 * config/i386/i386.opt: Change range from [1,5] to [0,5].
27000 2017-06-29 Yury Gribov <tetra2005@gmail.com>
27003 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
27006 (edge_growth_cache_entry::edge_growth_cache_entry): New
27008 (reset_edge_growth_cache): Update to use constructor.
27010 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27012 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
27013 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
27014 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
27016 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
27018 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
27019 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
27021 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
27023 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
27024 (*-linux-uclibc*): Add t-uclibc tmake_file.
27025 * config/t-musl: New.
27026 * config/t-uclibc: New.
27028 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
27030 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
27032 (gen_comm_data): Emit architectural setting of arch_prof.
27033 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
27035 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
27036 (armv8-m.base, armv8-m.main): Likewise.
27037 * arm-protos.h (arm_build_target): Add profile field.
27038 (arch_option): Likewise.
27039 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
27041 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
27042 arm_active_target.profile.
27044 2017-06-28 Richard Biener <rguenther@suse.de>
27046 PR middle-end/81227
27047 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
27048 TYPE_OVERFLOW_WRAPS.
27049 * match.pd (negate_expr_p): Likewise.
27050 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
27051 fold_build2, not fold_binary.
27053 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
27055 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
27056 Convert memory address to Pmode.
27057 (aarch64_print_operand): Assert MEM operands are always Pmode.
27059 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
27062 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
27063 Remove redundant if.
27064 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
27065 * config/arm/aarch-common-protos.h
27066 (aarch_forward_to_shift_is_not_shifted_re): Remove.
27067 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
27069 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
27072 * multiple_target.c (create_dispatcher_calls): Set the default
27073 clone to be static, not public.
27075 2017-06-28 Richard Biener <rguenther@suse.de>
27077 * tree-vect-loop.c (vectorizable_reduction): Move special
27078 cond reduction IV var creation ...
27079 (vect_create_epilog_for_reduction): ... here. Remove induction_index
27080 parameter. Use STMT_VINFO_VECTYPE.
27081 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
27084 2017-06-28 Martin Liska <mliska@suse.cz>
27087 * ipa-visibility.c (non_local_p): Handle visibility.
27089 2017-06-28 Martin Liska <mliska@suse.cz>
27092 * common.opt: Add IntegerRange to various options.
27093 * opt-functions.awk (integer_range_info): New function.
27094 * optc-gen.awk: Add integer_range_info to cl_options struct.
27095 * opts-common.c (decode_cmdline_option): Handle
27096 CL_ERR_INT_RANGE_ARG.
27097 (cmdline_handle_error): Likewise.
27098 * opts.c (print_filtered_help): Show valid interval in
27099 when --help is provided.
27100 * opts.h (struct cl_option): Add range_min and range_max fields.
27101 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
27103 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
27105 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
27106 (x * C EQ/NE y * C): New transformation.
27108 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
27110 * genmultilib (combination_space): Accept '+' in option names.
27112 2017-06-28 Martin Liska <mliska@suse.cz>
27115 * asan.c (instrument_derefs): Bail out inner references
27116 that are hard register variables.
27118 2017-06-28 Jakub Jelinek <jakub@redhat.com>
27121 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
27122 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
27124 2017-06-28 Richard Biener <rguenther@suse.de>
27126 * tree-vectorizer.h (vect_get_vec_defs): Remove.
27127 (vect_get_slp_defs): Adjust.
27128 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
27130 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
27132 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
27133 get_initial_defs_for_reduction instead of vect_get_vec_defs.
27134 (vectorizable_reduction): Adjust.
27135 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
27137 (vect_get_slp_defs): Likewise.
27138 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
27139 (vectorizable_bswap): Adjust.
27140 (vectorizable_call): Likewise.
27141 (vectorizable_conversion): Likewise.
27142 (vectorizable_assignment): Likewise.
27143 (vectorizable_shift): Likewise.
27144 (vectorizable_operation): Likewise.
27145 (vectorizable_store): Likewise.
27146 (vectorizable_condition): Likewise.
27147 (vectorizable_comparison): Likewise.
27149 2017-06-28 Michael Collison <michael.collison@arm.com>
27152 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
27154 (arm_gen_movmemqi): Removed unused variable 'i'.
27155 Convert 'for' loop into 'while' loop.
27156 (arm_expand_prologue): Remove last unnecessary set of insn.
27157 (thumb_pop): Remove unused variable 'pushed_words'.
27158 (thumb_exit): Remove last unnecessary set of regs_to_pop.
27160 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27162 * config/s390/predicates.md: Use s390_rel_address_ok_p.
27163 * config/s390/s390-protos.h: Add prototype of
27164 s390_rel_address_ok_p.
27165 * config/s390/s390.c (s390_got_symbol): New function.
27166 (s390_rel_address_ok_p): New function.
27167 (legitimize_pic_address): Use s390_rel_address_ok_p.
27168 (s390_load_got): Use s390_got_symbol.
27169 (s390_option_override): Issue error if
27170 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
27171 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
27173 * config/s390/s390.opt: New option mpic-data-is-text-relative.
27175 2017-06-27 Andrew Pinski <apinski@cavium.com>
27177 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
27178 (X * copysign (1.0, X)): New pattern.
27179 (X * copysign (1.0, -X)): New pattern.
27180 (copysign (-1.0, CST)): New pattern.
27182 2017-06-27 Joseph Myers <joseph@codesourcery.com>
27184 * genmultilib (combination_space): Remove variable.
27185 Validate reuse rules against regular expression for any sequence
27186 of multilib options in any order.
27188 2017-06-27 Michael Collison <michael.collison@arm.com>
27190 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
27191 call aarch64_split_simd_combine.
27192 * (aarch64_combine_internal<mode>): Delete pattern.
27193 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
27194 Allow register and subreg operands.
27196 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27198 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
27199 specific need, just fallback on defaults.
27200 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
27202 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27203 Olivier Hainque <hainque@adacore.com>
27205 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
27207 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
27208 targets. Pick a default if no particular attempt applied.
27209 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
27212 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27214 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
27215 (x86_64-wrs-vxworks7): Likewise.
27217 2017-06-27 Marek Polacek <polacek@redhat.com>
27220 * ubsan.c (instrument_null): Check get_base_address's result for null.
27222 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27224 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
27226 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27228 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
27229 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
27230 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
27231 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
27232 New function types.
27233 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
27234 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
27235 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
27236 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
27237 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
27238 BUILT_IN_FEUPDATEENV): New builtins.
27239 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
27240 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
27241 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
27242 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
27244 (builtin_structptr_types): Adjust size.
27245 * tree.c (builtin_structptr_types): Add four entries.
27247 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27248 Olivier Hainque <hainque@adacore.com>
27250 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
27251 (TLS_SYM): New local macro, forcing reference to __tls__ on
27252 link command lines for VxWorks 7 RTPs, triggering initialization
27254 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
27255 OS features TLS support, true for RTPs on VxWorks 7.
27256 * config/vxworks.c (vxworks_override_options): Setup emutls
27259 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27261 * predict.c (test_prediction_value_range): Use -1U instead of -1
27262 to avoid narrowing conversion warning.
27263 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
27264 to avoid narrowing conversion warning.
27265 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
27267 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
27269 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27271 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
27272 64bit configurations.
27273 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
27274 (SIZE_TYPE): Likewise.
27275 * config/vxworks.c (vxworks_emutls_var_fields): Use
27276 long_unsigned_type_node instead of unsigned_type_node as the offset
27277 field type, which is "pointer" mode in emutls.c.
27279 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27282 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
27284 PR middle-end/81207
27285 * gimple-fold.c (replace_call_with_call_and_fold): Handle
27286 gimple_vuse copying separately from gimple_vdef copying.
27288 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27290 * value-prof.c (free_hist): Remove call to memset and the enclosing if
27293 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27294 Olivier Hainque <hainque@adacore.com>
27296 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
27297 for all vxworks7 targets.
27298 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
27299 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
27300 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
27301 variations for VX6/VX7 and 32/64bits later on in ...
27302 (VXWORKS_LIB_SPEC): Leverage new macros.
27303 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
27304 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
27306 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27308 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
27309 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
27311 2017-06-26 Carl Love <cel@us.ibm.com>
27313 * config/rs6000/rs6000-c.c: Add support for built-in functions
27314 vector bool char vec_reve (vector bool char);
27315 vector signed char vec_reve (vector signed char);
27316 vector unsigned char vec_reve (vector unsigned char);
27317 vector bool int vec_reve (vector bool int);
27318 vector signed int vec_reve (vector signed int);
27319 vector unsigned int vec_reve (vector unsigned int);
27320 vector bool long long vec_reve (vector bool long long);
27321 vector signed long long vec_reve (vector signed long long);
27322 vector unsigned long long vec_reve (vector unsigned long long);
27323 vector bool short vec_reve (vector bool short);
27324 vector signed short vec_reve (vector signed short);
27325 vector double vec_reve (vector double);
27326 vector float vec_reve (vector float);
27327 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
27328 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
27329 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
27330 (altivec_vreve): New pattern.
27331 * config/rs6000/altivec.h (vec_reve): New define.
27332 * doc/extend.texi (vec_rev): Update the built-in documentation file
27333 for the new built-in functions.
27335 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27337 PR tree-optimization/71815
27338 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
27340 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
27342 (slsr_process_phi): Likewise.
27343 (replace_uncond_cands_and_profitable_phis): Don't replace a
27344 multiply candidate with a stride of 1 (copy or cast).
27345 (phi_incr_cost): Call uses_consumed_by_stmt rather than
27347 (lowest_cost_path): Likewise.
27348 (total_savings): Likewise.
27350 2017-06-26 Richard Biener <rguenther@suse.de>
27353 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
27354 Use def_builtin_pure for all gather builtins.
27356 2017-06-26 Richard Biener <rguenther@suse.de>
27358 PR tree-optimization/81203
27359 * tree-tailcall.c (find_tail_calls): Do not move stmts into
27360 non-dominating BBs.
27362 2017-06-26 Marek Polacek <polacek@redhat.com>
27365 * doc/invoke.texi: Document -Wmultistatement-macros.
27367 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
27369 * doc/sourcebuild.texi (ARM-specific attributes): Document new
27370 arm_neon_ok_no_float_abi effective target.
27372 2017-06-26 Richard Biener <rguenther@suse.de>
27374 PR tree-optimization/80928
27375 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
27376 (copy_bbs): Set BB_DUPLICATED flag early.
27377 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
27379 (execute_on_shrinking_pred): Likewise.
27380 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
27381 BB_DUPLICATED blocks.
27382 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
27383 iterate over all PHIs considering removal of *gsi.
27385 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
27387 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
27390 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27392 * config/rs6000/rs6000-string.c: (expand_block_clear,
27393 do_load_for_compare, select_block_compare_mode,
27394 compute_current_alignment, expand_block_compare,
27395 expand_strncmp_align_check, expand_strn_compare,
27396 expand_block_move, rs6000_output_load_multiple)
27397 Move functions related to string/block move/compare
27398 to a separate file.
27399 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
27400 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
27401 for this function which is now used in two files.
27402 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
27403 * config.gcc: Add rs6000-string.o to extra_objs for
27404 targets powerpc*-*-* and rs6000*-*-*.
27406 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
27409 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
27410 32-bit, since indexed is not valid for DImode.
27411 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
27412 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
27413 (define_peephole2 for Altivec d-form load): Add 32-bit support.
27414 (define_peephole2 for Altivec d-form store): Likewise.
27417 * multiple_target.c (create_dispatcher_calls): Only create the
27418 dispatcher call if the function is the default clone of a
27419 versioned function.
27421 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
27423 PR middle-end/80902
27424 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
27425 a call, force the call to not be a tail call.
27427 2017-06-23 Jeff Law <law@redhat.com>
27429 * doc/contrib.texi: Add entry for Steven Pemberton's work on
27432 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
27434 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
27435 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
27436 handling for early expansion of vector shifts (sl,sr,sra,rl).
27437 (builtin_function_type): Add vector shift right instructions
27438 to the unsigned argument list.
27440 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
27442 rtl-optimizatoin/79286
27443 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
27444 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
27445 trap. PIC register plus a const unspec without offset can never trap.
27447 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
27449 * tree.h (builtin_structptr_type): New type.
27450 (builtin_structptr_types): Declare new array.
27451 * tree.c (builtin_structptr_types): New array.
27452 (free_lang_data, build_common_tree_nodes): Use it.
27454 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
27457 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
27460 2017-06-22 Matt Turner <mattst88@gmail.com>
27462 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
27463 Lake models to skylake case. Assume skylake for unknown
27464 models with clflushopt.
27466 2017-06-22 Jeff Law <law@redhat.com>
27468 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
27469 frame sizes that do not satisfy aarch64_uimm12_shift.
27471 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
27473 * profile-count.h (apply_probability,
27474 apply_scale, probability_in): Fix checks for zero.
27476 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27478 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
27479 * doc/cppdiropts.texi (-I @var{dir}): Document it.
27481 2016-06-22 Richard Biener <rguenther@suse.de>
27483 * tree-vect-loop.c (vect_model_reduction_cost): Handle
27484 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
27485 REDUC_MAX_EXPR support.
27486 (vectorizable_reduction): Likewise.
27487 (vect_create_epilog_for_reduction): Likewise.
27489 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
27491 * match.pd (A / (1 << B) -> A >> B): New.
27492 * generic-match-head.c: Include optabs-tree.h.
27493 * gimple-match-head.c: Likewise.
27494 * optabs-tree.h (target_supports_op_p): New.
27495 * optabs-tree.c (target_supports_op_p): New.
27497 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27499 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
27500 $gcc_cv_ld --help output.
27501 (gcc_cv_ld_demangle): Likewise.
27502 (gcc_cv_ld_eh_frame_hdr): Likewise.
27503 (gcc_cv_ld_pie): Likewise.
27504 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
27505 (gcc_cv_ld_buildid): Likewise.
27506 (gcc_cv_ld_sysroot): Likewise.
27507 (ld_bndplt_support): Likewise.
27508 (ld_pushpopstate_support): Likewise.
27509 * configure: Regenerate.
27510 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
27512 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27515 * config/i386/sse.md (round<mode>2): Renumber match_dup and
27516 operands indexes to avoid gap between operands and match_dups.
27518 2017-06-21 Andrew Pinski <apinski@cavium.com>
27520 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
27521 Increment Arith_shift and Arith_shift_reg by 1.
27522 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
27524 * config/aarch64/aarch64.c (thunderx_tunings): Enable
27525 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
27526 (aarch64_strip_extend): Add new argument and test for it.
27527 (aarch64_cheap_mult_shift_p): New function.
27528 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
27529 add a cost if it is true.
27530 Update calls to aarch64_strip_extend.
27531 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
27533 2017-06-21 Andrew Pinski <apinski@cavium.com>
27535 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
27537 (thunderxt88): Likewise.
27538 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
27539 (thunderx_prefetch_tune): New variable.
27540 (thunderx2t99_prefetch_tune): Update for the correct values.
27541 (thunderxt88_tunings): New variable.
27542 (thunderx_tunings): Use thunderx_prefetch_tune instead of
27543 generic_prefetch_tune.
27544 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
27546 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27548 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
27549 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
27550 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
27551 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
27552 (aarch64_atomic_cas<mode>, GPI): Likewise.
27554 2017-06-21 Martin Liska <mliska@suse.cz>
27556 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
27557 statements on cold and hot labels.
27558 * predict.c (tree_estimate_probability_bb): Remove the
27559 prediction from this place.
27561 2017-06-21 Martin Liska <mliska@suse.cz>
27563 PR tree-optimization/79489
27564 * gimplify.c (maybe_add_early_return_predict_stmt): New
27566 (gimplify_return_expr): Call the function.
27567 * predict.c (tree_estimate_probability_bb): Remove handling
27569 * predict.def: Update comment about early return predictor.
27570 * gimple-predict.h (is_gimple_predict): New function.
27571 * predict.def: Change default value of early return to 66.
27572 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
27574 * passes.def: Put pass_strip_predict_hints to the beginning of
27577 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
27579 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
27580 FUNCTION_DECL declarations.
27581 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
27583 (dwaf2out_decl): Likewise.
27584 * godump.c (go_early_global_decl): Skip call to the real debug hook
27585 for FUNCTION_DECL declarations.
27586 * passes.c (rest_of_decl_compilation): Skip call to the
27587 early_global_decl debug hook for FUNCTION_DECL declarations, unless
27588 -fdump-go-spec is passed.
27590 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27592 * config/i386/i386.c (struct builtin_isa): New field pure_p.
27593 Reorder for compactness.
27594 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
27595 (def_builtin_pure, def_builtin_pure2): New functions.
27596 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
27598 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27600 * match.pd (nop_convert): New predicate.
27601 ((A +- CST1) +- CST2): Allow some NOP conversions.
27603 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27606 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
27607 with ctors/dtors if GOVD_SHARED is set.
27609 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27611 * config/aarch64/aarch64.md (movti_aarch64):
27612 Emit mov rather than orr.
27613 (movtf_aarch64): Likewise.
27614 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
27615 Emit mov rather than orr.
27617 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27619 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
27620 Swap alternatives, make integer dup more expensive.
27622 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27624 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
27625 Return true for non-tls symbols.
27627 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
27629 * config/aarch64/aarch64-cores.def (cortex-a55): New.
27630 (cortex-a75): Likewise.
27631 (cortex-a75.cortex-a55): Likewise.
27632 * config/aarch64/aarch64-tune.md: Regenerate.
27633 * doc/invoke.texi (-mtune): Document new values for -mtune.
27635 2017-06-21 Tom de Vries <tom@codesourcery.com>
27637 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
27638 stack_size feature.
27639 (Effective-Target Keywords, Other attributes): Suggest using
27640 dg-add-options stack_size feature to get stack limit in stack_size
27641 effective target documentation.
27643 2017-06-21 Julian Brown <julian@codesourcery.com>
27644 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27646 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
27647 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
27648 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
27650 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
27651 attribute type list for neon_multiply.
27652 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
27653 attribute type list for neon_multiply.
27654 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
27655 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
27656 attribute type list for neon_multiply.
27657 * config/arm/types.md (crypto_pmull): Add.
27658 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
27659 attribute type list.
27661 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
27663 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
27666 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27668 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
27669 to make sure not to dereference a NULL cost_classes_ptr pointer.
27671 2017-06-20 Carl Love <cel@us.ibm.com>
27673 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27674 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
27675 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
27676 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
27677 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
27678 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
27679 VMULOSW): New enum "unspec" values.
27680 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
27681 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
27682 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
27683 altivec_vmulosw): New patterns.
27684 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
27685 VMULOSW): Add definitions.
27687 2017-06-20 Julia Koval <julia.koval@intel.com>
27689 * config/i386/i386.c: Fix rounding expand for new pattern.
27690 * config/i386/subst.md: Fix pattern (parallel -> unspec).
27692 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27694 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
27695 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
27697 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27699 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
27702 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27704 * config/aarch64/aarch64-cores.def: Rearrange to sort by
27705 architecture, then by implementer ID.
27706 * config/aarch64/aarch64-tune.md: Regenerate.
27708 2017-06-20 Richard Biener <rguenther@suse.de>
27710 PR middle-end/81097
27711 * fold-const.c (split_tree): Fold to type before negating.
27713 2017-06-20 David Malcolm <dmalcolm@redhat.com>
27715 * diagnostic-show-locus.c
27716 (selftest::test_fixit_deletion_affecting_newline): New function.
27717 (selftest::diagnostic_show_locus_c_tests): Call it.
27719 2017-06-20 Andreas Schwab <schwab@suse.de>
27722 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
27725 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27727 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
27728 __ARM_FEATURE_COPROC according to support.
27730 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27732 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27733 Rewritten to avoid overflow for > 32-bit pointers.
27736 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
27737 by removing enum keyword.
27738 (ubsan_type_descriptor): Likewise. Formatting fix.
27741 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
27742 splitter): Require TARGET_SSE2 in the condition.
27744 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
27747 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
27748 for doing vector set of SFmode on ISA 3.0.
27749 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
27750 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
27752 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
27753 SFmode value into a V4SF variable that was extracted from another
27754 V4SF variable without converting the element to double precision
27755 and back to single precision vector format.
27756 (vsx_insert_extract_v4sf_p9_2): Likewise.
27758 2017-06-19 Jakub Jelinek <jakub@redhat.com>
27760 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
27761 in UWHI to avoid undefined overflow.
27764 * ubsan.h (enum ubsan_encode_value_phase): New.
27765 (ubsan_encode_value): Change second argument to
27766 enum ubsan_encode_value_phase with default value of
27767 UBSAN_ENCODE_VALUE_GENERIC.
27768 * ubsan.c (ubsan_encode_value): Change second argument to
27769 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
27770 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
27771 create_tmp_var_raw instead of create_tmp_var and use a
27773 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
27774 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27775 ubsan_encode_value callers.
27778 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27779 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27780 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27782 2017-06-19 Richard Biener <rguenther@suse.de>
27784 PR middle-end/81118
27785 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27786 estimates if we changed anything.
27788 2017-06-19 Richard Biener <rguenther@suse.de>
27790 PR tree-optimization/80887
27791 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27792 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27793 simplified lookups, then reset mprts_hook.
27794 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27796 (try_to_simplify): Likewise.
27798 2017-06-19 Martin Liska <mliska@suse.cz>
27801 * gimplify.c (gimplify_switch_expr):
27802 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27804 2017-06-19 Martin Liska <mliska@suse.cz>
27806 * doc/install.texi: Document that PGO runs in 4 stages.
27808 2017-06-19 Martin Liska <mliska@suse.cz>
27811 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27812 to dispatcher function name.
27813 * multiple_target.c (replace_function_decl): New function.
27814 (create_dispatcher_calls): Redirect both edges and references.
27816 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27818 * profile-count.c (profile_count::dump): Dump quality.
27819 (profile_count::differs_from_p): Update for unsigned val.
27820 * profile-count.h (profile_count_quality): New enum.
27821 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27823 2017-06-19 Richard Biener <rguenther@suse.de>
27825 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27826 struct function as arg.
27827 (estimate_numbers_of_iterations): Export overload with loop arg.
27828 (free_numbers_of_iterations_estimates_loop): Use an overload of
27829 free_numbers_of_iterations_estimates instead.
27830 * tree-cfg.c (remove_bb): Adjust.
27831 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27832 * tree-parloops.c (gen_parallel_loop): Likewise.
27833 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27835 (tree_unroll_loops_completely): Likewise.
27836 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27837 Use an overload instead and export.
27838 (estimated_loop_iterations): Adjust.
27839 (max_loop_iterations): Likewise.
27840 (likely_max_loop_iterations): Likewise.
27841 (estimate_numbers_of_iterations): Take struct function as arg
27843 (loop_exits_before_overflow): Adjust.
27844 (free_numbers_of_iterations_estimates_loop): Use an overload.
27845 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27846 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27848 2017-06-19 Richard Biener <rguenther@suse.de>
27851 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27852 RANGE_EXPR conservatively.
27854 2017-06-16 Carl Love <cel@us.ibm.com>
27856 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27857 definitions for vec_float, vec_float2, vec_floato,
27858 vec_floate built-ins.
27859 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27860 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27862 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27863 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27864 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27865 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27866 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27867 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27868 vec_floato): Add builtin defines.
27869 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27870 Update the built-in documentation file for the new built-in
27873 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27875 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27876 (mthumb): Mark as the negative of -marm.
27878 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27880 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27882 (ARM Options, -mtune): Document that this accepts the same
27883 extension options as -mcpu.
27884 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27886 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27888 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27889 permitted extensions.
27891 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27893 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27894 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27895 (armv8-m.main): Add option +nodsp.
27896 * config/arm/arm-cpu-cdata.h: Regenerated.
27898 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27900 * config/arm/t-fuchsia: New file.
27901 * config.gcc (arm*-*-fuchsia*): Use it.
27903 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27905 * config/arm/t-symbian: Rewrite for new option infrastructure.
27907 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27909 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27910 (MULTILIB_REQUIRED): Likewise.
27912 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27914 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27915 (MULTILIB_RESUE): Likewise.
27916 (MULTILIB_MATCHES): Likewise.
27917 (MULTLIB_REQUIRED): Likewise.
27919 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27921 * config/arm/t-rtems: Rewrite for new option framework.
27923 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27925 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27926 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27927 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27928 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27929 * config/arm/t-multilib: ... here.
27930 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27931 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27932 armv7-a and armv8*-a when A-profile libraries have not been built.
27933 * config/arm/t-rmprofile: Rewrite.
27935 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27937 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27938 with a backslash. Remove the backslash after substituting unescaped
27940 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27942 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27944 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27945 the driver to pass through the default setting of -mfloat-abi.
27946 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27948 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27949 (all_feat_combs): New rule.
27950 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27952 * config/arm/t-aprofile: Rewrite.
27954 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27956 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27957 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27958 fpu is not specified by the user/command-line.
27959 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27960 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27961 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27962 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27963 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27964 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27966 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27968 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27969 * config/arm/t-arm-elf: Rewritten.
27971 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27973 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27974 have some floating-point instructions.
27975 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27976 (TARGET_MAYBE_HARD_FLOAT): New macro.
27977 * config/arm/arm-builtins.c (arm_init_builtins): Use
27978 TARGET_MAYBE_HARD_FLOAT.
27979 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27981 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27983 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27984 (configargs.h): Include it.
27985 (arm_print_hint_for_fpu_option): New function.
27986 (arm_parse_fpu_option): New function.
27987 (candidate_extension): New class.
27988 (arm_canon_for_multilib): New function.
27989 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27990 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27991 (ARCH_CANONICAL_SPECS): New macro.
27992 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27994 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27996 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27997 are set after handling multilib fragments. Set target_cpu_default2
28000 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28002 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
28004 (arm*-*-*): Set target_cpu_default2 to a quoted string.
28005 * config/arm/parsecpu.awk (check_cpu): Validate any extension
28007 (check_arch): Likewise.
28008 * config/arm/arm.c (arm_configure_build_target): Handle
28009 TARGET_CPU_DEFAULT being a string constant. Scan any feature
28010 options in the default.
28012 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28014 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
28015 when an option is an alias of another.
28016 * config/arm/parsecpu.awk (optalias): New parser token.
28017 (gen_comm_data): Mark non-alias options as such. Emit entries
28018 for extension aliases.
28019 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
28020 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
28021 (armv6kz, armv6zk, armv6t2): Likewise.
28022 (armv7): Make vfpv3-d16 an alias.
28023 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
28025 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
28026 Sort in canonical order.
28027 (armv8-a): Sort in canonical order.
28028 (armv8.1-a, armv8.2-a): Likewise.
28029 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
28031 (cortex-a9): Sort in canonical order.
28032 * config/arm/arm.c (selftests.h): Include it.
28033 (arm_test_cpu_arch_data): New function.
28034 (arm_run_self_tests): New function.
28035 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
28036 (targetm): Move declaration to the end of the file.
28037 * arm-cpu-cdata.h: Regenerated.
28039 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28041 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
28042 call to target_mode_check describing the type of option passed.
28043 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
28044 (arm_target_thumb_only): Use arm_parse_arch_option_name or
28045 arm_parse_cpu_option_name to match parameters against list of
28047 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
28048 arm_arch_core_flags data structure.
28049 * config/arm/arm-cpu_cdata.h: Regenerated.
28051 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28053 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
28055 (arm_print_hint_for_cpu_option): Likewise.
28056 (arm_print_hint_for_arch_option): Likewise.
28057 (arm_parse_cpu_option_name): Likewise.
28058 (arm_parse_arch_option_name): Likewise.
28059 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
28060 of entries in the all_fpus list.
28061 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
28062 (arm_parse_cpu_option_name): Declare.
28063 (arm_parse_arch_option_name): Declare.
28064 (arm_parse_option_features): Declare.
28065 (arm_intialize_isa): Declare.
28066 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
28068 (gen_comm_data): ... here. Make definitions non-static.
28069 * config/arm/arm-cpu-data.h: Regenerated.
28070 * config/arm/arm-cpu-cdata.h: Regenerated.
28072 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28074 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
28075 (cpu_arch_extension): New structure.
28076 (cpu_arch_option, arch_option, cpu_option): New structures.
28077 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
28078 architecture types.
28079 (gen_data): Generate new format data tables.
28080 * config/arm/arm.c (cpu_tune): New structure.
28081 (cpu_option, processors): Delete.
28082 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
28083 (arm_print_hint_for_cpu_option): ... this and ...
28084 (arm_print_hint_for_arch_option): ... this.
28085 (arm_parse_arch_cpu_name): Delete. Replace with ...
28086 (arm_parse_cpu_option_name): ... this and ...
28087 (arm_parse_arch_option_name): ... this.
28088 (arm_unrecognized_feature): Change type of target parameter to
28090 (arm_parse_arch_cpu_features): Delete. Replace with ...
28091 (arm_parse_option_features): ... this.
28092 (arm_configure_build_target): Rework to use new configuration data
28094 (arm_print_tune_info): Rework for new configuration data tables.
28095 * config/arm/arm-cpu-data.h: Regenerated.
28096 * config/arm/arm-cpu.h: Regenerated.
28098 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28100 * Makefile.in (OBJS): Move sbitmap.o from here ...
28101 (OBJS-libcommon): ... to here.
28103 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28105 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
28106 (ISA_ALL_CRYPTO): New macro.
28107 (ISA_ALL_SIMD): New macro
28108 (ISA_ALL_FP): New macro.
28109 * config/arm/arm.c (fpu_bitlist): Update initializer.
28110 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
28112 (arm9e): Add fpu. Add option for nofp
28113 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
28114 (arm926ej-s, arm1026ej-s): Likewise.
28115 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
28116 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
28117 neon-fp16, neon-vfpv4, nofp and nosimd.
28118 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
28119 (cortex-a8): Add fpu. Add option for nofp.
28120 (cortex-a9): Add fpu. Add options for nosimd and nofp.
28121 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
28122 (cortex-r4f): Add fpu.
28123 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
28124 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
28126 (cortex-r8): Likewise.
28127 (cortex-m4): Add fpu. Add option for nofp.
28128 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
28129 (cortex-a17.cortex-a7): Likewise.
28130 (cortex-a32): Add fpu. Add options for crypto and nofp.
28131 (cortex-a35, cortex-a53): Likewise.
28132 (cortex-a57): Add fpu. Add option for crypto.
28133 (cortex-a72, cortex-a73): Likewise.
28134 (exynos-m1): Likewise.
28135 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
28136 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
28137 (cortex-m33): Add fpu. Add option for nofp.
28138 * config/arm/arm-cpu-cdata.h: Regenerated
28139 * config/arm/arm-cpu-data.h: Regenerated.
28141 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28143 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
28144 (armv5te, armv5tej): Likewise.
28145 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
28146 (armv7): Add options fp and vfpv3-d16.
28147 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
28148 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
28150 (armv7ve): Likewise.
28151 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
28152 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
28153 (armv8-a): Add nocrypto option.
28154 (armv8.1-a, armv8.2-a): Likewise.
28155 (armv8-m.main): add options fp, fp.dp and nofp.
28157 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28159 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
28161 (armv8-a+crc): Delete.
28162 (armv8.1-a): Add options simd, crypto and nofp.
28163 (armv8.2-a): Add options fp16, simd, crypto and nofp.
28164 (armv8.2-a+fp16): Delete.
28165 (armv8-m.main): Add option dsp.
28166 (armv8-m.main+dsp): Delete.
28167 (cortex-a8): Add fpu. Add nofp option.
28168 (cortex-a9): Add fpu. Add nofp and nosimd options.
28169 * config/arm/parsecpu.awk (gen_data): Generate option tables and
28170 link to main cpu and architecture data structures.
28171 (gen_comm_data): Only put isa attributes from the main architecture
28173 (option): New statement for architecture and CPU entries.
28174 * arm.c (struct cpu_option): New structure.
28175 (struct processors): Add entry for options.
28176 (arm_unrecognized_feature): New function.
28177 (arm_parse_arch_cpu_name): Ignore any characters after the first
28179 (arm_parse_arch_cpu_feature): New function.
28180 (arm_configure_build_target): Separate out any CPU and architecture
28181 features and parse separately. Don't error out if -mfpu=auto is
28182 used with only an architecture string.
28183 (arm_print_asm_arch_directives): New function.
28184 (arm_file_start): Call it.
28185 * config/arm/arm-cpu-cdata.h: Regenerated.
28186 * config/arm/arm-cpu-data.h: Likewise.
28187 * config/arm/arm-tables.opt: Likewise.
28189 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28191 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
28192 assembler when it is not -mfpu=auto.
28194 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28196 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
28197 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
28198 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
28199 (ASM_CPU_SPEC): Rewrite.
28200 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
28201 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
28202 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
28203 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
28204 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
28205 copied string is NUL-terminated. Also strip any characters prefixed
28207 (arm_rewrite_selected_arch): New function.
28208 (arm_rewrite_march): New function.
28210 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28212 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
28213 (x_arm_cpu_string, x_arm_tune_string): Likewise.
28214 (march, mcpu, mtune): Convert to string-based options.
28215 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
28216 (arm_parse_arch_cpu_name): New function.
28217 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
28218 identify selected architecture or CPU.
28219 (arm_option_save): New function.
28220 (TARGET_OPTION_SAVE): Redefine.
28221 (arm_option_restore): Restore string options.
28222 (arm_option_print): Print string options.
28224 2017-06-16 Martin Sebor <msebor@redhat.com>
28226 PR tree-optimization/80933
28227 PR tree-optimization/80934
28228 * builtins.c (fold_builtin_3): Do not handle bcmp here.
28229 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
28230 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
28231 (gimple_fold_builtin): Call them.
28233 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28235 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
28236 as unlikely; update profile.
28238 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28240 * predict.c (force_edge_cold): Handle declaring edges impossible
28243 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28245 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
28247 (try_unroll_loop_completely): Fix reporting.
28249 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28251 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
28253 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
28256 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
28257 if given a non-constant argument for an intrinsic which requires a
28260 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28262 * profile.c (compare_freqs): New function.
28263 (branch_prob): Sort edge list.
28264 (find_spanning_tree): Assume that the list is priority sorted.
28266 2017-06-16 Richard Biener <rguenther@suse.de>
28268 PR tree-optimization/81090
28269 * passes.def (pass_record_bounds): Remove.
28270 * tree-pass.h (make_pass_record_bounds): Likewise.
28271 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
28272 make_pass_record_bounds): Likewise.
28273 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
28274 not free niter estimates at the beginning but at the end.
28275 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
28277 2017-06-16 Richard Biener <rguenther@suse.de>
28279 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
28280 initializer to workaround ICE in host GCC 4.8.
28282 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28284 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
28286 (clone_inlined_nodes): Update.
28288 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28290 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
28291 prefetch settings, and enable prefetching by default at -O3.
28293 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28295 * config/aarch64/aarch64.c (aarch64_override_options_internal):
28296 Set flag_prefetch_loop_arrays according to tuning data.
28298 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28300 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
28301 New tune structure.
28302 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
28303 [Unrelated to main purpose of the patch] Place the pointer field last
28304 to enable type checking errors when tune structure are wrongly merged.
28305 * config/aarch64/aarch64.c (generic_prefetch_tune,)
28306 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
28307 (thunderx2t99_prefetch_tune): New tune constants.
28308 (tune_params *_tunings): Update all tunings (no functional change).
28309 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
28310 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
28311 from tunings structures.
28313 2017-06-16 Jakub Jelinek <jakub@redhat.com>
28316 * ubsan.c (instrument_null): Add T argument, use it instead
28317 of computing it based on IS_LHS.
28318 (instrument_object_size): Likewise.
28319 (pass_ubsan::execute): Adjust instrument_null and
28320 instrument_object_size callers to pass gimple_get_lhs or
28321 gimple_assign_rhs1 result to it. Use instrument_null instead of
28322 calling get_base_address and instrument_mem_ref. Handle
28323 aggregate call arguments for object-size sanitization.
28325 2017-06-16 Yury Gribov <tetra2005@gmail.com>
28327 PR tree-optimization/81089
28328 * tree-vrp.c (is_masked_range_test): Validate operands of
28331 2017-06-15 Martin Sebor <msebor@redhat.com>
28334 * dumpfile.c (dump_register): Avoid calling memset to initialize
28335 a class with a default ctor.
28336 * gcc.c (struct compiler): Remove const qualification.
28337 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
28338 * hash-table.h: Ditto.
28339 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
28341 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
28342 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
28344 * params.h (struct param_info): Make struct members non-const.
28345 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
28346 with default initialization.
28347 * vec.h (vec_copy_construct, vec_default_construct): New helper
28349 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
28350 with vec_copy_construct.
28351 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
28352 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
28353 * doc/invoke.texi (-Wclass-memaccess): Document.
28355 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28357 * emit-rtl.h (is_leaf): Update comment about local
28358 register allocator.
28360 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28363 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
28364 for a variable to have a section before checking if the section has a
28366 Set section to.persistent if persistent attribute is set.
28367 Warn if .persistent attribute is used on an automatic variable.
28369 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
28371 PR rtl-optimization/80474
28372 * reorg.c (update_block): Do not ignore instructions in a delay slot.
28374 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
28376 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
28379 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
28381 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
28382 (casesi): Emit bounds checking as RTL.
28383 (casesi_internal_mips16_<mode>): Remove bounds checking.
28385 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
28387 * config/xtensa/xtensa.c (xtensa_option_override): Append
28388 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
28389 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
28390 xtensa_doloop_hooks): Define unconditionally.
28391 (xtensa_reorg_loops): Only call reorg_loops in the presence of
28393 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
28394 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
28395 for it in xtensa_option_override.
28396 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
28397 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
28399 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
28401 * doc/cppopts.texi: Document '-' special value to -MF.
28403 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
28405 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
28406 (cortex_a53_fconst): Likewise.
28407 (cortex_a53_fpmul): Likewise.
28408 (cortex_a53_f_load_64): Likewise.
28409 (cortex_a53_f_load_many): Likewise.
28410 (cortex_a53_advsimd_alu): Likewise.
28411 (cortex_a53_advsimd_alu_q): Likewise.
28412 (cortex_a53_advsimd_mul): Likewise.
28413 (cortex_a53_advsimd_mul_q): Likewise.
28414 (fpmac bypass): Add new bypass for fpmac-fpmac case.
28415 Add missing fmul, r2f_cvt and fconst cases.
28417 2017-06-14 Richard Biener <rguenther@suse.de>
28419 PR middle-end/81088
28420 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
28422 (fold_binary_loc): When associating do not treat pre-existing
28423 TREE_OVERFLOW on literal constants as a reason to allow
28424 TREE_OVERFLOW on associated literal constants.
28426 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28428 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
28429 (MASK_FEATURES): New macro.
28430 * config/sparc/sparc.c (sparc_option_override): Remove the special
28431 handling of -mfpu and generalize it to all MASK_FEATURES switches.
28433 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28435 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
28436 a division of 0 if non-call exceptions are enabled.
28438 2017-06-14 Andrew Pinski <apinski@cavium.com>
28439 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28442 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
28443 Improve vector initialization code gen for only variable case.
28445 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28447 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
28449 2017-06-14 Richard Biener <rguenther@suse.de>
28451 PR tree-optimization/81083
28452 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
28455 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28457 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
28458 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
28459 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
28460 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
28461 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
28462 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
28464 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28466 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
28467 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
28469 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28471 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
28473 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28475 * config/rs6000/t-rtems: Don't handle SPE.
28477 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28479 * config/rs6000/t-linux: Don't handle SPE.
28481 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28483 * config/rs6000/eabispe.h: Delete file.
28485 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28487 * config/rs6000/t-spe: Delete file.
28489 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28491 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
28492 (rs6000_legitimate_offset_address_p): Return false for anything in
28493 V2SImode or V2SFmode.
28495 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28497 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
28498 except V2SF and V2SI. Rearrange the vector modes, and add comments.
28499 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
28501 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
28502 (rs6000_legitimate_offset_address_p): Ditto.
28503 (rs6000_emit_move): Ditto.
28504 (rs6000_init_builtins): Remove V4HI_type_node.
28506 2017-06-13 Martin Liska <mliska@suse.cz>
28509 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
28510 (gate_asan): Likewise.
28511 * asan.h (asan_no_sanitize_address_p): Remove the function.
28512 (sanitize_flags_p): New function.
28513 * builtins.def: Fix coding style.
28514 * common.opt: Use renamed enum value.
28515 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
28516 * doc/extend.texi: Document no_sanitize attribute.
28517 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
28518 to SANITIZE_UNDEFINED_NONDEFAULT.
28519 * gcc.c (sanitize_spec_function): Use the renamed enum value.
28520 * gimple-fold.c (optimize_atomic_compare_exchange_p):
28521 Use sanitize_flags_p.
28522 * gimplify.c (gimplify_function_tree): Likewise.
28523 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
28524 * opts.c (parse_no_sanitize_attribute): New function.
28525 (common_handle_option): Use renamed enum value.
28526 * opts.h (parse_no_sanitize_attribute): Declare.
28527 * tree.c (sanitize_flags_p): New function.
28528 * tree.h: Declared here.
28529 * tsan.c: Use sanitize_flags_p.
28530 * ubsan.c (ubsan_expand_null_ifn): Likewise.
28531 (instrument_mem_ref): Likewise.
28532 (instrument_bool_enum_load): Likewise.
28533 (do_ubsan_in_current_function): Remove the function.
28534 (pass_ubsan::execute): Use sanitize_flags_p.
28535 * ubsan.h: Remove do_ubsan_in_current_function
28536 * tree-cfg.c (print_no_sanitize_attr_value): New function.
28537 (dump_function_to_file): Use it here.
28539 2017-06-13 Martin Jambor <mjambor@suse.cz>
28541 PR tree-optimization/80803
28542 PR tree-optimization/81063
28543 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
28544 (propagate_subaccesses_across_link): Enqueue subtree whenever
28545 necessary instead of relying on the caller.
28547 2017-06-13 Martin Jambor <mjambor@suse.cz>
28549 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
28550 that have a first_link.
28551 (sort_and_splice_var_accesses): Do not check first_link before
28553 (subtree_mark_written_and_enqueue): Likewise.
28554 (propagate_all_subaccesses): Likewise and do not stop at first
28555 parent with a first_link.
28557 2017-06-13 Martin Jambor <mjambor@suse.cz>
28559 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
28562 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28564 * match.pd: New pattern.
28566 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28568 * tree-vrp.c (is_masked_range_test): New function.
28569 (register_edge_assert_for): Determine ranges for
28572 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28574 PR tree-optimization/67328
28575 * fold-const.c (maskable_range_p): New function.
28576 (build_range_check): Generate bittests if possible.
28578 2017-06-13 Martin Liska <mliska@suse.cz>
28580 * gimple-pretty-print.c (dump_probability): Add new argument.
28581 (dump_edge_probability): Dump both probability and count.
28582 (dump_gimple_label): Likewise.
28583 (dump_gimple_bb_header): Likewise.
28585 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
28588 * config/avr/avr-devices.c: Fix indentation.
28589 * config/avr/gen-avr-mmcu-specs.c: Dito.
28591 2017-06-13 Richard Biener <rguenther@suse.de>
28593 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
28594 instead get vector type from stmt_info.
28595 (vectorizable_reduction): Adjust. Remove dead code.
28597 2017-06-13 Richard Biener <rguenther@suse.de>
28599 PR middle-end/81065
28600 * fold-const.c (extract_muldiv_1): Remove bogus distribution
28601 case of C * (x * C2 + C3).
28602 (fold_addr_of_array_ref_difference): Properly fold index difference.
28604 2017-06-12 David S. Miller <davem@davemloft.net>
28607 * config/sparc/sparc.md (return expander): Emit frame blockage if
28608 function uses alloca.
28610 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
28612 * combine.c (make_field_assignment): Check len rather than the mode
28613 precision when calling force_to_mode.
28615 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
28617 Support multilibs and devices that see flash in RAM address range.
28620 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
28621 (avr_mcu_t) <flash_pm_offset>: New field.
28622 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
28623 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
28624 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
28625 (AVR_TINY_PM_OFFSET): Remove macro.
28626 * config/avr/avr.opt (-mshort-calls): New option.
28627 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
28628 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
28629 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
28630 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
28631 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
28632 instead of avr_arch->have_jmp_call.
28633 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
28634 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
28635 avr_arch->flash_pm_offset to define.
28636 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
28637 new field flash_pm_offset. Add entry for avrxmega3.
28638 (avr_texinfo): Add entry for avrxmega3.
28639 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
28640 attiny212, attiny214,
28641 attiny412, attiny414, attiny416, attiny417,
28642 attiny814, attiny816, attiny817,
28643 attiny1614, attiny1616, attiny1617,
28644 attiny3214, attiny3216, attiny3217.
28645 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
28646 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
28647 (avr_print_operand_address) [AVR_TINY]: Same.
28648 (avr_asm_init_sections) <readonly_data_section>: Only patch
28649 callback if avr_arch->flash_pm_offset = 0.
28650 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
28651 for rodata if avr_arch->flash_pm_offset != 0.
28652 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
28653 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
28654 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
28655 (m_options): Append opt_rcall.
28656 (m_dirnames): Append dir_rcall.
28657 * config/avr/t-multilib: Regenerate.
28659 * configure.ac [target=avr]: Check whether avrxmega3 default
28660 linker description file works as needed.
28661 * configure: Regenerate.
28662 * doc/avr-mmcu.texi: Regenerate.
28663 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
28664 <__AVR_ARCH__>: Document avrxmega3 and 103.
28665 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
28666 <__AVR_SHORT_CALLS__>: Document it.
28667 <__AVR_PM_BASE_ADDRESS__>: Document it.
28668 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
28669 (AVR Variable Attributes) <progmem>: Document this is
28670 not needed for avrxmega3.
28671 (AVR Named Address Spaces) <__flash>: Dito.
28673 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
28675 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
28677 2017-06-12 Doug Rupp <rupp@adacore.com>
28679 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
28680 Append vxworks-stdint.h to the tm_file list.
28681 * config/vxworks-stdint.h: New file.
28683 2017-06-12 Martin Liska <mliska@suse.cz>
28685 PR tree-optimization/81041
28686 * tree-profile.c (gimple_gen_ic_func_profiler):
28687 Create an extra BB in profile-generate
28688 (gimple_gen_time_profiler): Likewise.
28690 2017-06-12 Jakub Jelinek <jakub@redhat.com>
28692 PR tree-optimization/81003
28693 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
28694 (update_range_test): Use it instead of force_gimple_operand_gsi.
28696 2017-06-12 Richard Biener <rguenther@suse.de>
28698 PR tree-optimization/81053
28699 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
28700 with backedge value not defined in loop. Simplify def stmt
28703 2017-06-11 Tom de Vries <tom@codesourcery.com>
28706 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
28708 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
28709 nvptx_cannot_force_const_mem.
28711 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28713 * opts.c (finish_options): Move test for flag_split_stack after
28714 it has been initialized.
28716 2017-06-11 Jason Merrill <jason@redhat.com>
28718 * tree.h (id_equal): New.
28719 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
28720 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
28721 instead of strcmp of IDENTIFIER_POINTER.
28723 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28725 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
28726 (mark_all_inlined_calls_cdtor): Fix formating.
28727 (inline_transform): Rescale profile before inlining.
28729 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28731 * cgraph.h (cgraph_edge::clone): Update prototype.
28732 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
28733 (cgraph_node::create_clone): Update.
28734 (cgraph_node::create_version_clone): Update.
28735 * tree-inline.c (copy_bb): Update.
28736 (expand_call_inline): Update.
28738 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28740 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
28741 factored out from ...
28742 (rs6000_emit_prologue): ... here.
28744 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28746 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
28747 factored out from ...
28748 (rs6000_emit_prologue): ... here.
28750 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28752 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
28754 (handle_missing_profiles): Fix computation of tp_first_run.
28755 (counts_to_freqs): Do not touch freqs when count is 0.
28757 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28759 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
28762 2017-06-10 Tom de Vries <tom@codesourcery.com>
28764 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28765 attributes): Document signal effective target.
28767 2017-06-10 Tom de Vries <tom@codesourcery.com>
28769 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28770 Document effective target stack_size.
28772 2017-06-09 David Malcolm <dmalcolm@redhat.com>
28774 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28775 to the edit_context if they can be auto-applied.
28777 2017-06-9 Ian Lance Taylor <iant@golang.org>
28779 * opts.c (finish_options): If -fsplit-stack, disable implicit
28780 -forder-blocks-and-partition.
28781 * doc/invoke.texi (Optimize Options): Document that when using
28782 -fsplit-stack -forder-blocks-and-partition is not implicitly
28785 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28787 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28788 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28789 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28790 * builtins.def (abort, trap, unreachable): Declare cold.
28791 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28792 * tree-core.h (ECF_COLD): New.
28793 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28794 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28796 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28798 * predict.c (unlikely_executed_stmt_p): Cleanup.
28800 2017-06-09 Richard Biener <rguenther@suse.de>
28802 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28803 model if the ref is always written to.
28805 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28807 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28809 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28811 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28814 2017-06-09 Tom de Vries <tom@codesourcery.com>
28817 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28818 "target cannot support label values" when encountering LABEL_REF.
28820 2017-06-09 Martin Liska <mliska@suse.cz>
28822 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28823 (gimple_gen_ic_func_profiler): Emit direct comparison
28824 of __gcov_indirect_call_callee with NULL.
28825 (gimple_gen_time_profiler): Change probability from
28826 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28828 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28830 * profile.c (edge_gcov_counts): Turn to pointer.
28831 (compute_branch_probabilities, compute_branch_probabilities): Update.
28832 (branch_prob): Do not clear edge_gcov_count.
28833 * profile.h (edge_gcov_counts): Turn to pointer.
28834 (edge_gcov_count): Update.
28836 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28838 * gimple.h (gimple_check_failed): Mark cold.
28840 2017-06-09 Richard Biener <rguenther@suse.de>
28842 PR tree-optimization/66623
28843 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28844 refactor check_reduction into two parts, properly computing
28845 whether we have to check reduction validity for outer loop
28848 2017-06-09 Richard Biener <rguenther@suse.de>
28850 PR tree-optimization/79483
28851 * graphite-scop-detection.c (order): New global.
28852 (get_order): Compute bb to order mapping that satisfies code
28853 generation constraints.
28854 (cmp_pbbs): New helper.
28855 (build_scops): Start domwalk at entry block, sort generated
28858 2017-06-09 Richard Biener <rguenther@suse.de>
28860 PR middle-end/81007
28861 * ipa-polymorphic-call.c
28862 (ipa_polymorphic_call_context::restrict_to_inner_class):
28863 Skip FIELD_DECLs with error_mark_node type.
28864 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28867 2017-06-09 Martin Liska <mliska@suse.cz>
28869 * predict.c (struct branch_predictor): New struct.
28870 (test_prediction_value_range): New test.
28871 (predict_c_tests): New function.
28872 * selftest-run-tests.c (selftest::run_tests): Run the function.
28873 * selftest.h: Declare new tests.
28875 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28878 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28879 gen_add3_insn did not fail.
28880 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28881 r0, construct that number in a temporary reg and add that reg to r0.
28882 If asked to put the result in r0 as well, fail.
28884 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28886 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28887 for early expansion of vec_eqv.
28889 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28891 PR middle-end/81005
28892 * ubsan.c (instrument_null): Avoid pointless code temporary.
28893 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28895 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28899 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28902 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28905 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28907 * predict.c (unlikely_executed_edge_p): Move ahead.
28908 (probably_never_executed_edge_p): Use it.
28910 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28912 PR middle-end/79988
28913 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28914 gimple_call_builtin_p call.
28916 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28918 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28919 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28920 rtl_check_failed_type2, rtl_check_failed_code1,
28921 rtl_check_failed_code2, rtl_check_failed_code_mode,
28922 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28923 rtvec_check_failed_bounds, rtl_check_failed_flag,
28924 _fatal_insn_not_found, _fatal_insn): Likewise.
28925 * tree.h (tree_contains_struct_check_failed,
28926 tree_check_failed, tree_not_check_failed,
28927 tree_class_check_failed, tree_range_check_failed,
28928 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28929 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28930 tree_operand_check_failed, omp_clause_check_failed,
28931 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28934 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28936 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28937 flag_branch_probabilities.
28938 * ipa-inline.c (edge_badness): Likewise.
28939 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28940 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28941 * predict.c (maybe_hot_frequency_p): Likewise.
28942 (probably_never_executed): Likewise.
28943 * sched-ebb.c (schedule_ebbs): Likewise.
28944 * sched-rgn.c (find_single_block_region): Likewise.
28945 * tracer.c (tail_duplicate): Likewise.
28947 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28949 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28950 longer requires x_flag_profile_use.
28952 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28954 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28955 instead of flag_reorder_blocks_and_partition.
28956 * dbxout.c (dbxout_function_end): Likewise.
28957 * dwarf2out.c (gen_subprogram_die): Likewise.
28958 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28959 * hw-doloop.c (reorg_loops): Likewise.
28960 * varasm.c (assemble_start_function,
28961 assemble_end_function): Likewise.
28962 (decide_function_section): Do not check for
28963 flag_reorder_blocks_and_partition.
28965 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28967 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28969 (chkp_get_hard_register_fake_addr_expr): Ditto.
28970 (chkp_build_addr_expr): Add check for hard reg case.
28971 (chkp_parse_array_and_component_ref): Ditto.
28972 (chkp_find_bounds_1): Ditto.
28973 (chkp_process_stmt): Don't generate bounds store for
28976 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28978 * predict.c (maybe_hot_bb_p): Do not check profile status.
28979 (maybe_hot_edge_p): Likewise.
28980 (probably_never_executed): Check for zero counts even if profile
28982 (unlikely_executed_edge_p): New function.
28983 (unlikely_executed_stmt_p): New function.
28984 (unlikely_executed_bb_p): New function.
28985 (set_even_probabilities): Use unlikely predicates.
28986 (combine_predictions_for_bb): Likewise.
28987 (predict_paths_for_bb): Likewise.
28988 (predict_paths_leading_to_edge): Likewise.
28989 (determine_unlikely_bbs): New function.
28990 (estimate_bb_frequencies): Use it.
28991 (compute_function_frequency): Use zero counts even if profile is
28993 * profile-count.h: Fix typo.
28995 2017-08-08 Julia Koval <julia.koval@intel.com>
28997 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28998 _mm512_mask_cvtsepi16_storeu_epi8,
28999 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
29000 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
29001 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
29002 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
29003 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
29004 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
29005 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
29006 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
29007 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
29008 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
29009 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
29010 __builtin_ia32_pmovuswb256mem_mask,
29011 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
29012 __builtin_ia32_pmovwb512mem_mask): New builtins.
29014 2017-08-08 Julia Koval <julia.koval@intel.com>
29016 PR target/73350,80862
29017 * config/i386/subst.md (round): Fix round pattern.
29018 * config/i386/i386.c (ix86_erase_embedded_rounding):
29019 Fix erasing rounding for the fixed pattern.
29021 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
29023 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
29025 2017-06-08 Martin Liska <mliska@suse.cz>
29027 PR gcov-profile/80911
29028 * gcov.c (block_info::block_info): New constructor.
29030 2017-06-07 Carl Love <cel@us.ibm.com>
29032 * config/rs6000/rs6000-c: The return type of the following
29033 built-in functions was implemented as int not long long. Fix sign
29034 of return value for the unsigned version of vec_mulo and vec_mule.
29035 vector unsigned long long vec_bperm (vector unsigned long long,
29036 vector unsigned char)
29037 vector signed long long vec_mule (vector signed int,
29039 vector unsigned long long vec_mule (vector unsigned int,
29040 vector unsigned int)
29041 vector signed long long vec_mulo (vector signed int,
29043 vector unsigned long long vec_mulo (vector unsigned int,
29044 vector unsigned int)
29045 * doc/extend.texi: Fix the documentation for the built-in
29048 2017-06-07 Carl Love <cel@us.ibm.com>
29051 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
29054 2017-06-07 Carl Love <cel@us.ibm.com>
29056 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
29057 support, Generate doublehv for signed int/float for BE case only.
29059 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
29061 * doc/invoke.texi (mcx16): Rewrite.
29063 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29065 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
29066 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
29067 *mov<mode>_softfloat, and an anonymous splitter): Use
29068 nonimmediate_operand instead of rs6000_nonimmediate_operand.
29070 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29072 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
29074 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
29075 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
29076 (rs6000_debug_reg_global): Adjust.
29077 (rs6000_init_hard_regno_mode_ok): Adjust.
29078 (rs6000_dbx_register_number): Adjust.
29079 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
29080 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
29081 Remove SPE_ACC and SPEFSCR.
29082 (REG_ALLOC_ORDER): Ditto.
29083 (FRAME_POINTER_REGNUM): Change to 111.
29084 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
29085 (REG_CLASS_NAMES): Ditto.
29086 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
29087 (REGISTER_NAMES): Ditto.
29088 (ADDITIONAL_REG_NAMES): Ditto.
29089 (rs6000_reg_names): Ditto.
29090 * config/rs6000/rs6000.md: Renumber some register number
29093 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29095 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
29097 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
29098 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
29100 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
29101 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
29102 Delete the SPE high registers.
29103 (REG_ALLOC_ORDER): Ditto.
29104 (enum reg_class): Remove SPE_HIGH_REGS.
29105 (REG_CLASS_NAMES): Ditto.
29106 (REG_CLASS_CONTENTS): Delete the SPE high registers.
29107 (REGISTER_NAMES): Ditto.
29108 (rs6000_reg_names): Ditto.
29109 * doc/tm.texi.in: Remove SPE as example.
29110 * doc/tm.texi: Regenerate.
29112 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29114 * config/rs6000/8540.md (ppc8540_brinc): Delete.
29115 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
29116 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
29117 * config/rs6000/rs6000.md (type): Remove "brinc".
29119 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29121 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
29122 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
29123 * config/rs6000/linuxspe.h: Delete file.
29124 * config/rs6000/rs6000.md: Don't include spe.md.
29125 * config/rs6000/spe.h: Delete file.
29126 * config/rs6000/spe.md: Delete file.
29127 * config/rs6000/t-rs6000: Remove spe.md.
29129 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29131 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
29132 (reg_or_none500mem_operand): Delete.
29133 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
29134 instead of reg_or_none500mem_operand.
29136 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29138 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29139 handling of SPE flags.
29140 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
29142 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29144 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
29146 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
29147 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
29148 paired_divv2sf3): Similar.
29149 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
29150 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
29151 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
29153 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
29154 Rename the paired_* instruction patterns.
29155 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
29157 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
29158 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
29159 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
29160 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
29161 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
29162 PAIRED_VECTOR_MODE.
29163 (struct machine_function): Delete field spe_insn_chain_scanned_p.
29164 (spe_func_has_64bit_regs_p): Delete.
29165 (spe_expand_predicate_builtin): Delete.
29166 (spe_expand_evsel_builtin): Delete.
29167 (TARGET_DWARF_REGISTER_SPAN): Do not define.
29168 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
29169 (invalid_e500_subreg): Delete.
29170 (rs6000_legitimize_address): Always force_reg op2 as well, for
29171 paired single memory accesses.
29172 (rs6000_member_type_forces_blk): Delete.
29173 (rs6000_spe_function_arg): Delete.
29174 (rs6000_expand_unop_builtin): Delete SPE handling.
29175 (rs6000_expand_binop_builtin): Ditto.
29176 (spe_expand_stv_builtin): Delete.
29177 (bdesc_2arg_spe): Delete.
29178 (spe_expand_builtin): Delete.
29179 (spe_expand_predicate_builtin): Delete.
29180 (spe_expand_evsel_builtin): Delete.
29181 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
29182 (spe_init_builtins): Delete.
29183 (spe_func_has_64bit_regs_p): Delete.
29184 (savres_routine_name): Delete "info" parameter. Adjust callers.
29185 (rs6000_emit_stack_reset): Ditto.
29186 (rs6000_dwarf_register_span): Delete.
29187 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
29188 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
29189 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
29191 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
29193 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
29194 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
29195 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
29196 mulv2sf3, divv2sf3): Delete expanders.
29198 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29200 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
29202 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29204 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
29205 * config/rs6000/rs6000.c: Ditto.
29207 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29209 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
29210 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
29211 comparison_operator.
29213 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29215 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
29216 * config/rs6000/rs6000.opt: Ditto.
29217 * config/rs6000/t-rtems: Ditto.
29219 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29221 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
29222 TARGET_E500_SINGLE by 0, simplify.
29223 * config/rs6000/rs6000.c: Ditto.
29224 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
29225 (spe_build_register_parallel): Delete.
29226 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
29227 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
29228 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
29229 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
29230 (E500_CONVERT): Delete.
29231 * config/rs6000/spe.md: Remove many patterns and all define_constants.
29233 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29235 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
29236 * config/rs6000/dfp.md: Ditto.
29237 (negdd2, *negdd2_fpr): Merge.
29238 (absdd2, *absdd2_fpr): Merge.
29239 (negtd2, *negtd2_fpr): Merge.
29240 (abstd2, *abstd2_fpr): Merge.
29241 * config/rs6000/e500.h: Delete file.
29242 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
29243 TARGET_FPRS by 1 and simplify.
29244 * config/rs6000/rs6000-c.c: Ditto.
29245 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
29246 TARGET_DF_SPE by 0.
29247 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
29249 * config/rs6000/rs6000.md: Ditto.
29250 (floatdidf2, *floatdidf2_fpr): Merge.
29251 (move_from_CR_gt_bit): Delete.
29252 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
29253 (E500_CR_IOR_COMPARE): Delete.
29254 (All patterns that require !TARGET_FPRS): Delete.
29255 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
29257 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29259 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
29261 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29263 * graphds.c (add_edge): Intitialize edge's attached data.
29264 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
29265 pointer parameter. Call pointed function on each edge during
29266 graph traversing. Skip traversing the edge when the function
29268 (graphds_dfs, graphds_scc): Ditto.
29269 (for_each_edge): New parameter. Pass the new parameter to callback
29271 * graphds.h (skip_edge_callback): New function pointer type.
29272 (graphds_dfs, graphds_scc): New function pointer parameter.
29273 (graphds_edge_callback, for_each_edge): New parameter.
29275 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29277 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
29278 out code checking if runtime alias check is possible to below ...
29279 Call the new function.
29280 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
29281 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
29283 2017-06-07 Marek Polacek <polacek@redhat.com>
29286 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
29287 TYPE_OVERFLOW_WRAPS check.
29289 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29291 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
29292 if versioning is required.
29293 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
29294 peeling with the check for versioning.
29296 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29298 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
29299 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
29300 Set true to new parameter if new ssa variable is defined.
29301 (vect_gen_vector_loop_niters): Refactor. Set range information
29302 for the new vector loop bound variable.
29303 (vect_do_peeling): Ditto.
29305 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29307 * tree-affine.c (ssa.h): Include header file.
29308 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
29309 has wrapping overflow behavior.
29311 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29313 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
29315 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29317 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
29318 (tree_to_aff_combination): ... here.
29320 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29322 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
29323 reg_pressure model function.
29324 (ivopts_global_cost_for_size): Delete.
29325 (determine_set_costs, iv_ca_recount_cost): Call new model function
29326 ivopts_estimate_reg_pressure.
29328 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29330 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
29331 expensive than udiv. Remove floating point cases from mod.
29333 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29335 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
29336 Increase idiv cost.
29338 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29340 * config/aarch64/aarch64.md
29341 (copysignsf3): Fix mask generation.
29343 2017-06-07 Jakub Jelinek <jakub@redhat.com>
29345 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
29347 (class dump_manager): Add register_dumps method.
29348 * dumpfile.c: Include langhooks.h.
29349 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
29350 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
29351 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
29352 (dump_manager::dump_register): Start with 512 entries instead of 32.
29353 (dump_manager::register_dumps): New method.
29354 * toplev.c (general_init): Instead of invoking register_dumps
29355 langhook, invoke register_dumps method on the dump manager.
29356 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
29359 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
29361 * doc/md.texi: Clarify the restrictions on a define_insn condition.
29362 Say that # requires an associated define_split to exist, and that
29363 the define_split must be suitable for use after register allocation.
29365 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29367 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
29368 (compute_outgoing_frequencies): Also initialize zero counts.
29369 (find_many_sub_basic_blocks): Do not produce uninitialized profile
29370 around loops; preserve more of profile when nothing changes.
29372 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
29374 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
29376 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
29377 * config/arm/arm-cpu-cdata.h: Regenerate.
29378 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
29379 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
29380 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
29381 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
29382 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
29384 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
29386 * config/arm/t-rmprofile: Likewise.
29387 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
29389 2017-06-06 David S. Miller <davem@davemloft.net>
29392 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
29393 blockage if function uses alloca.
29395 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29397 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
29398 New "uid" fields to hold pretty-print IDs of group and ref.
29399 Memory references are now identified as <group_id>:<ref_id>
29400 instead of using [random] addresses.
29401 (dump_mem_details): Simplify, no functional change.
29402 (dump_mem_ref): Simplify and make output more concise.
29403 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
29404 (find_or_create_group): Initialize group uid.
29405 (record_ref): Initialize ref uid. Improve debug output.
29406 (prune_group_by_reuse, should_issue_prefetch_p,)
29407 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
29408 (mark_nontemporal_store, determine_loop_nest_reuse):
29409 Improve debug output.
29411 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29413 * dbgcnt.def (prefetch): New debug counter.
29414 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
29415 (schedule_prefetches): Stop issueing prefetches if debug counter
29418 2017-06-06 Tom de Vries <tom@codesourcery.com>
29420 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
29421 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
29423 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29425 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
29426 Use aarch64_reg_or_zero predicate for operand 4.
29427 (aarch64_compare_and_swap<mode> define_insn_and_split):
29428 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
29429 (aarch64_store_exclusive<mode>): Likewise for operand 2.
29431 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
29433 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
29434 (arm_compute_save_core_reg_mask): This.
29435 (thumb1_compute_save_reg_mask): Rename into ...
29436 (thumb1_compute_save_core_reg_mask): This.
29437 (arm_compute_save_reg0_reg12_mask): Adapt comment.
29438 (arm_compute_frame_layout): Likewise.
29440 2017-06-06 Richard Biener <rguenther@suse.de>
29442 PR tree-optimization/80974
29443 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
29444 keep or clear leaders SSA info.
29446 2017-06-06 Tom de Vries <tom@codesourcery.com>
29448 * config/nvptx/nvptx.c (split_mode_p): New function.
29449 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
29451 2017-06-06 Tom de Vries <tom@codesourcery.com>
29453 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
29455 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29458 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
29461 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29463 * shrink-wrap.c (handle_simple_exit): Update profile.
29464 (try_shrink_wrapping): Upate profile.
29466 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29468 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
29469 (tree_guess_outgoing_edge_probabilities): New.
29470 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
29471 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
29473 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29475 * ipa-split.c (split_function): Initialize return bb profile.
29477 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29479 * profile.c (compute_branch_probabilities): Also initialize
29480 EXIT_BLOCK profile.
29482 2017-06-06 Richard Biener <rguenther@suse.de>
29484 PR tree-optimization/80928
29485 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
29486 (vect_analyze_loop_operations): Properly guard analysis for
29488 (vect_transform_loop): Likewise.
29489 (vect_analyze_loop_2): Also reset SLP type on PHIs.
29490 (vect_model_induction_cost): Do not cost for pure SLP.
29491 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
29492 of induction in inner loop vectorization.
29493 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
29494 (vect_get_and_check_slp_defs): Handle vect_induction_def.
29495 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
29497 (vect_analyze_slp_cost_1): Cost induction.
29498 (vect_detect_hybrid_slp_stmts): Handle PHIs.
29499 (vect_get_slp_vect_defs): Likewise.
29500 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
29501 (vect_transform_stmt): Handle SLP reductions.
29502 * tree-vectorizer.h (vectorizable_induction): Adjust.
29504 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29506 * config/rs6000/rs6000.c (make_resolver_func): Update
29507 init_lowered_empty_function call.
29509 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
29511 * doc/invoke.texi: Document the -fprofile-abs-path option.
29512 * common.opt (fprofile-abs-path): New option.
29513 * gcov-io.h (gcov_write_filename): Declare.
29514 * gcov-io.c (gcov_write_filename): New function.
29515 * coverage.c (coverage_begin_function): Use gcov_write_filename.
29516 * profile.c (output_location): Likewise.
29518 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29520 * shring-wrap.c: Revert accidental commit.
29522 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
29524 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
29526 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29528 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
29530 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
29531 profile in callgraph edge.
29532 * profile-count.h (apply_probability): If THIS is 0, then result is 0
29533 (apply_scale): Likewise.
29534 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
29535 Also scale profile when inlining function with zero profile.
29536 (initialize_cfun): Update exit block profile even when it is zero.
29537 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
29538 when profile is read.
29540 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29542 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
29543 (CLONE_*): New constants to define the processors we can generate
29544 code for with the target_clone attribute.
29545 (rs6000_clone_map): New array to identify which clone processors
29546 the current program is running on.
29547 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
29548 target_clone attribute.
29549 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
29550 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
29551 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
29552 (cpu_expand_builtin): Add support for target_clone attribute.
29553 (rs6000_valid_attribute_p): Allow "default" attribute.
29554 (get_decl_name): New debug function to simplify printing the
29555 current function name in debugging statements.
29556 (rs6000_clone_priority): New functions to support the target_clone
29557 attribute, and be able to generate code to switch between ISA 2.05
29558 through ISA 3.0 (power6 through power9).
29559 (rs6000_compare_version_priority): Likewise.
29560 (rs6000_get_function_versions_dispatcher): Likewise.
29561 (make_resolver_func): Likewise.
29562 (add_condition_to_bb): Likewise.
29563 (dispatch_function_versions): Likewise.
29564 (rs6000_generate_version_dispatcher_body): Likewise.
29565 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
29566 (fusion_gpr_load_p): Fix a spacing issue.
29567 * doc/extend.texi (Common Function Attributes): Document that the
29568 PowerPC supports the target_clone attribute.
29570 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29572 * config/arm/arm.h: explain F symbol found in description of ARM
29573 register allocation in its legend.
29575 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29577 * config/mips/frame-header-opt.c: Include profile-count.h.
29578 * config/riscv/riscv.c: Include profile-count.h
29580 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29582 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
29584 (sm_set_flag_if_changed): Add bbs field.
29585 (execute_sm_if_changed_flag_set): Pass BBS.
29586 (execute_sm): Update.
29588 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29590 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
29593 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29595 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
29596 (peephole2): New peephole2 to emit the above.
29597 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
29599 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29601 * config/aarch64/aarch64.c (define_peephole2 above
29602 *sub_<shift>_<mode>): New peephole.
29604 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29606 * config/i386/i386.c (make_resolver_func): Update.
29607 * Makefile.in: Add profile-count.h and profile-count.o
29608 * auto-profile.c (afdo_indirect_call): Update to new API.
29609 (afdo_set_bb_count): Update.
29610 (afdo_propagate_edge): Update.
29611 (afdo_propagate_circuit): Update.
29612 (afdo_calculate_branch_prob): Update.
29613 (afdo_annotate_cfg): Update.
29614 * basic-block.h: Include profile-count.h
29615 (struct edge_def): Turn count to profile_count.
29616 (struct basic_block_def): Likewie.
29617 (REG_BR_PROB_BASE): Move to profile-count.h
29618 (RDIV): Move to profile-count.h
29619 * bb-reorder.c (max_entry_count): Turn to profile_count.
29620 (find_traces): Update.
29621 (rotate_loop):Update.
29622 (connect_traces):Update.
29623 (sanitize_hot_paths):Update.
29624 * bt-load.c (migrate_btr_defs): Update.
29625 * cfg.c (RDIV): Remove.
29626 (init_flow): Use alloc_block.
29627 (alloc_block): Uninitialize count.
29628 (unchecked_make_edge): Uninitialize count.
29629 (check_bb_profile): Update.
29630 (dump_edge_info): Update.
29631 (dump_bb_info): Update.
29632 (update_bb_profile_for_threading): Update.
29633 (scale_bbs_frequencies_int): Update.
29634 (scale_bbs_frequencies_gcov_type): Update.
29635 (scale_bbs_frequencies_profile_count): New.
29636 * cfg.h (update_bb_profile_for_threading): Update.
29637 (scale_bbs_frequencies_profile_count): Declare.
29638 * cfgbuild.c (compute_outgoing_frequencies): Update.
29639 (find_many_sub_basic_blocks): Update.
29640 * cfgcleanup.c (try_forward_edges): Update.
29641 (try_crossjump_to_edge): Update.
29642 * cfgexpand.c (expand_gimple_tailcall): Update.
29643 (construct_exit_block): Update.
29644 * cfghooks.c (verify_flow_info): Update.
29645 (dump_bb_for_graph): Update.
29646 (split_edge): Update.
29647 (make_forwarder_block): Update.
29648 (duplicate_block): Update.
29649 (account_profile_record): Update.
29650 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
29651 (get_estimated_loop_iterations): Update.
29652 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
29653 (single_likely_exit): Update.
29654 * cfgloopmanip.c (scale_loop_profile): Update.
29656 (set_zero_probability): Update.
29657 (lv_adjust_loop_entry_edge): Update.
29658 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
29659 (purge_dead_edges): Update.
29660 (rtl_account_profile_record): Update.
29661 * cgraph.c (cgraph_node::create): Uninitialize count.
29662 (symbol_table::create_edge): Uninitialize count.
29663 (cgraph_update_edges_for_call_stmt_node): Update.
29664 (cgraph_edge::dump_edge_flags): Update.
29665 (cgraph_node::dump): Update.
29666 (cgraph_edge::maybe_hot_p): Update.
29667 * cgraph.h: Include profile-count.h
29668 (create_clone), create_edge, create_indirect_edge): Update.
29669 (cgraph_node): Turn count to profile_count.
29670 (cgraph_edge0: Likewise.
29671 (make_speculative, clone): Update.
29672 (create_edge): Update.
29673 (init_lowered_empty_function): Update.
29674 * cgraphclones.c (cgraph_edge::clone): Update.
29675 (duplicate_thunk_for_node): Update.
29676 (cgraph_node::create_clone): Update.
29677 * cgraphunit.c (cgraph_node::analyze): Update.
29678 (cgraph_node::expand_thunk): Update.
29679 * final.c (dump_basic_block_info): Update.
29680 * gimple-streamer-in.c (input_bb): Update.
29681 * gimple-streamer-out.c (output_bb): Update.
29682 * graphite.c (print_global_statistics): Update.
29683 (print_graphite_scop_statistics): Update.
29684 * hsa-brig.c: Include basic-block.h.
29685 * hsa-dump.c: Include basic-block.h.
29686 * hsa-gen.c (T sum_slice): Update.
29687 (convert_switch_statements):Update.
29688 * hsa-regalloc.c: Include basic-block.h.
29689 * ipa-chkp.c (chkp_produce_thunks): Update.
29690 * ipa-cp.c (struct caller_statistics): Update.
29691 (init_caller_stats): Update.
29692 (gather_caller_stats): Update.
29693 (ipcp_cloning_candidate_p): Update.
29694 (good_cloning_opportunity_p): Update.
29695 (get_info_about_necessary_edges): Update.
29696 (dump_profile_updates): Update.
29697 (update_profiling_info): Update.
29698 (update_specialized_profile): Update.
29699 (perhaps_add_new_callers): Update.
29700 (decide_about_value): Update.
29701 (ipa_cp_c_finalize): Update.
29702 * ipa-devirt.c (struct odr_type_warn_count): Update.
29703 (struct decl_warn_count): Update.
29704 (struct final_warning_record): Update.
29705 (possible_polymorphic_call_targets): Update.
29706 (ipa_devirt): Update.
29707 * ipa-fnsummary.c (redirect_to_unreachable): Update.
29708 * ipa-icf.c (sem_function::merge): Update.
29709 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
29710 * ipa-inline.c (compute_uninlined_call_time): Update.
29711 (compute_inlined_call_time): Update.
29712 (want_inline_small_function_p): Update.
29713 (want_inline_self_recursive_call_p): Update.
29714 (edge_badness): Update.
29715 (lookup_recursive_calls): Update.
29716 (recursive_inlining): Update.
29717 (inline_small_functions): Update.
29718 (dump_overall_stats): Update.
29719 (dump_inline_stats): Update.
29720 * ipa-profile.c (ipa_profile_generate_summary): Update.
29721 (ipa_propagate_frequency): Update.
29722 (ipa_profile): Update.
29723 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29724 * ipa-utils.c (ipa_merge_profiles): Update.
29725 * loop-doloop.c (doloop_modify): Update.
29726 * loop-unroll.c (report_unroll): Update.
29727 (unroll_loop_runtime_iterations): Update.
29728 * lto-cgraph.c (lto_output_edge): Update.
29729 (lto_output_node): Update.
29730 (input_node): Update.
29731 (input_edge): Update.
29732 (merge_profile_summaries): Update.
29733 * lto-streamer-in.c (input_cfg): Update.
29734 * lto-streamer-out.c (output_cfg): Update.
29735 * mcf.c (create_fixup_graph): Update.
29736 (adjust_cfg_counts): Update.
29737 (sum_edge_counts): Update.
29738 * modulo-sched.c (sms_schedule): Update.
29739 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
29740 * predict.c (maybe_hot_count_p): Update.
29741 (probably_never_executed): Update.
29742 (dump_prediction): Update.
29743 (combine_predictions_for_bb): Update.
29744 (propagate_freq): Update.
29745 (handle_missing_profiles): Update.
29746 (counts_to_freqs): Update.
29747 (rebuild_frequencies): Update.
29748 (force_edge_cold): Update.
29749 * predict.h: Include profile-count.h
29750 (maybe_hot_count_p, counts_to_freqs): UPdate.
29751 * print-rtl-function.c: Do not include cfg.h
29752 * print-rtl.c: Include basic-block.h
29753 * profile-count.c: New file.
29754 * profile-count.h: New file.
29755 * profile.c (is_edge_inconsistent): Update.
29756 (correct_negative_edge_counts): Update.
29757 (is_inconsistent): Update.
29758 (set_bb_counts): Update.
29759 (read_profile_edge_counts): Update.
29760 (compute_frequency_overlap): Update.
29761 (compute_branch_probabilities): Update; Initialize and deinitialize
29763 (branch_prob): Update.
29764 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
29765 (edge_gcov_count): New.
29766 (bb_gcov_count): New.
29767 * shrink-wrap.c (try_shrink_wrapping): Update.
29768 * tracer.c (better_p): Update.
29769 * trans-mem.c (expand_transaction): Update.
29770 (ipa_tm_insert_irr_call): Update.
29771 (ipa_tm_insert_gettmclone_call): Update.
29772 * tree-call-cdce.c: Update.
29773 * tree-cfg.c (gimple_duplicate_sese_region): Update.
29774 (gimple_duplicate_sese_tail): Update.
29775 (gimple_account_profile_record): Update.
29776 (execute_fixup_cfg): Update.
29777 * tree-inline.c (copy_bb): Update.
29778 (copy_edges_for_bb): Update.
29779 (initialize_cfun): Update.
29780 (freqs_to_counts): Update.
29781 (copy_cfg_body): Update.
29782 (expand_call_inline): Update.
29783 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29784 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29785 (try_unroll_loop_completely): Update.
29786 (try_peel_loop): Update.
29787 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29788 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29789 * tree-ssa-loop-split.c (connect_loops): Update.
29790 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29791 * tree-ssa-reassoc.c (branch_fixup): Update.
29792 * tree-ssa-tail-merge.c (replace_block_by): Update.
29793 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29794 (compute_path_counts): Update.
29795 (update_profile): Update.
29796 (recompute_probabilities): Update.
29797 (update_joiner_offpath_counts): Update.
29798 (estimated_freqs_path): Update.
29799 (freqs_to_counts_path): Update.
29800 (clear_counts_path): Update.
29801 (ssa_fix_duplicate_block_edges): Update.
29802 (duplicate_thread_path): Update.
29803 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29804 (struct switch_conv_info): Update.
29805 * tree-tailcall.c (decrease_profile): Update.
29806 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29807 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29808 * value-prof.c (check_counter): Update.
29809 (gimple_divmod_fixed_value): Update.
29810 (gimple_mod_pow2): Update.
29811 (gimple_mod_subtract): Update.
29812 (gimple_ic_transform): Update.
29813 (gimple_stringop_fixed_value): Update.
29814 * value-prof.h (gimple_ic): Update.
29816 2017-06-02 Carl Love <cel@us.ibm.com>
29818 * config/rs6000/rs6000-c: Add support for built-in functions
29819 vector double vec_doublee (vector signed int);
29820 vector double vec_doublee (vector unsigned int);
29821 vector double vec_doublee (vector float);
29822 vector double vec_doubleh (vector signed int);
29823 vector double vec_doubleh (vector unsigned int);
29824 vector double vec_doubleh (vector float);
29825 vector double vec_doublel (vector signed int);
29826 vector double vec_doublel (vector unsigned int);
29827 vector double vec_doublel (vector float);
29828 vector double vec_doubleo (vector signed int);
29829 vector double vec_doubleo (vector unsigned int);
29830 vector double vec_doubleo (vector float);.
29831 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29832 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29834 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29835 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29836 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29838 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29839 vec_doublel, vec_doubleh.
29840 * doc/extend.texi: Update the built-in documentation file for the
29841 new built-in functions.
29843 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29846 * ipa-inline-analysis.c (free_growth_caches): Set
29847 edge_removal_hook_holder to NULL after removing it.
29849 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29851 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29852 comparision with zero.
29854 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29855 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29856 for early expansion of vec_min and vec_max builtins.
29857 (builtin_function_type): Add min/max unsigned variants to those
29858 identified as having unsigned arguments.
29860 2017-06-02 Olivier Hainque <hainque@adacore.com>
29862 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29864 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29866 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29867 Use VALL_F16 iterator rather than VALL.
29869 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29871 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29872 Emit CBNZ inside loop when doing a strong exchange and comparing
29873 against zero. Generate the CC flags after the loop.
29875 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29877 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29878 (dl_section_ref): New.
29879 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29880 On AIX, append an expression to subtract the size of the
29881 section length to dl_section_ref.
29883 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29885 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29886 for early expansion of vector absolute builtins.
29888 2017-06-02 Richard Biener <rguenther@suse.de>
29890 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29891 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29893 2017-06-02 Richard Biener <rguenther@suse.de>
29895 PR tree-optimization/80948
29896 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29897 stmt order as well.
29899 2017-06-02 Richard Biener <rguenther@suse.de>
29901 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29903 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29904 for inductions as relevant.
29906 2017-06-02 Richard Biener <rguenther@suse.de>
29908 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29909 (vectorizable_induction): ... this. Remove dead code.
29911 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29913 * builtins. (expand_builtin_alloca): Remove second parameter and
29914 infer its value from the first parameter instead.
29915 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29917 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29919 PR rtl-optimization/80903
29920 * loop-doloop.c (add_test): Unshare sequence.
29922 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29924 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29926 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29928 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29930 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29931 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29932 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29933 (xlogue_layout::get_stub_rtx): Make static.
29934 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29935 (xlogue_layout::compute_stub_managed_regs): Rename to...
29936 (xlogue_layout::count_stub_managed_regs): ...this.
29937 (xlogue_layout::is_stub_managed_reg): New function.
29938 (xlogue_layout::m_stub_names): Rename to...
29939 (xlogue_layout::s_stub_names): ...this, make static.
29940 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29941 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29942 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29943 xlogue_layout::s_stub_names): Instantiate statics.
29944 (stub_managed_regs): Remove.
29945 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29946 (disable_call_ms2sysv_xlogues): Rename to...
29947 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29948 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29950 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29951 change after reload_completed.
29952 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29954 (ix86_expand_prologue): Likewise.
29955 (ix86_expand_epilogue): Likewise.
29956 (ix86_expand_split_stack_prologue): Likewise.
29957 (ix86_compute_frame_layout): Remove frame parameter ...
29958 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29959 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29961 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29962 (ix86_frame): Move from here ...
29963 * config/i386/i386.h (ix86_frame): ... to here.
29964 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29965 complete ix86_frame data structure instead. Remove some_ld_name.
29967 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29969 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29970 symbols that hold a DECL_VALUE_EXPR.
29972 2017-06-01 Martin Jambor <mjambor@suse.cz>
29974 PR tree-optimization/80898
29975 * tree-sra.c (process_subtree_disqualification): Removed.
29976 (disqualify_candidate): Do not acll
29977 process_subtree_disqualification.
29978 (subtree_mark_written_and_enqueue): New function.
29979 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29980 RHS has been disqualified and re-queue LHS if necessary. Apart
29981 from that, ignore disqualified RHS.
29983 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29985 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29986 address fetch for z10 or later.
29988 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29990 * config/arc/arc.md (tst_movb): Add guard when splitting.
29992 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29994 * config/arc/arc.c (arc_can_eliminate): Test against
29995 arc_frame_pointer_needed.
29997 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29999 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
30000 to prevent store reordering.
30001 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
30002 (type): Add block type.
30003 (stack_tie): Define special instruction to be used in
30006 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30008 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
30009 constraint. It is not valid for the pattern.
30010 (noncommutative_binary_comparison): Likewise.
30012 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30014 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
30017 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30019 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
30020 be used by the reg-alloc.
30022 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30024 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
30025 reg-alloc when having mul64 or mul32x16 instructions.
30026 (mulsidi3): Likewise.
30027 (umulsidi3): Likewise.
30028 (mulsi32x16): New pattern.
30029 (mulsi64): Likewise.
30030 (mulsidi64): Likewise.
30031 (umulsidi64): Likewise.
30032 (MUL32x16_REG): Define.
30033 (mul64_600): Use MUL32x16_REG.
30034 (mac64_600): Likewise.
30035 (umul64_600): Likewise.
30036 (umac64_600): Likewise.
30038 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
30040 * config/arc/arc.md (mulsi3_700): Make it commutative.
30042 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
30044 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
30046 (*sign_extendsidi2_insn): Likewise for movstosw.
30048 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
30050 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
30051 the type of the input discriminant value. Convert the
30052 discriminant value of signedness vary.
30054 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
30056 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
30057 Add to -Wall section.
30059 2017-06-01 Richard Biener <rguenther@suse.de>
30061 PR middle-end/66313
30062 * fold-const.c (fold_plusminus_mult_expr): If the factored
30063 factor may be zero use a wrapping type for the inner operation.
30064 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
30065 and handle moved defs.
30066 (process_assignment): Properly guard the unary op case. Return a
30067 tri-state indicating that moving the stmt before the call may allow
30068 to continue. Pass through to_move.
30069 (find_tail_calls): Handle moving unrelated defs before
30072 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
30075 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
30076 splitter result in the canonical way.
30078 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
30080 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
30081 also for 32bit target. Update insn attributes.
30082 (zero-extendsidi2 splitter): Allow all registers for operand 1.
30084 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
30086 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
30087 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
30088 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
30089 (_mm_maskz_min_ss): New intrinsics.
30091 2017-05-31 Martin Liska <mliska@suse.cz>
30093 * tree-vect-loop.c (vect_create_epilog_for_reduction):
30094 Change comment style to one we normally use.
30095 (vectorizable_reduction): Likewise.
30096 (vectorizable_induction): Likewise.
30097 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
30098 (vectorizable_call): Likewise.
30099 (vectorizable_simd_clone_call): Likewise.
30100 (vectorizable_conversion): Likewise.
30101 (vectorizable_assignment): Likewise.
30102 (vectorizable_shift): Likewise.
30103 (vectorizable_operation): Likewise.
30104 (vectorizable_store): Likewise.
30105 (vectorizable_load): Likewise.
30106 * tree-vectorizer.h: Likewise.
30108 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
30110 * passes.c (emergency_dump_function): New.
30111 * tree-pass.h (emergency_dump_function): Declare.
30112 * plugin.c (plugins_internal_error_function): Remove.
30113 * plugin.h (plugins_internal_error_function): Remove declaration.
30114 * toplev.c (internal_error_function): New static function. Use it...
30115 (general_init): ...here.
30117 2017-05-31 Graham Markall <graham.markall@embecosm.com>
30119 * config/arc/arc.c (arc_print_operand): Handle constant operands.
30120 (arc_rtx_costs): Add costs for new patterns.
30121 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
30122 * config/arc/predicates.md: Add _1_2_3_operand predicate.
30124 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
30126 * tree-ssa-strlen.c (get_next_strinfo): New function.
30127 (get_stridx_plus_constant): Use it.
30128 (zero_length_string): Likewise.
30129 (adjust_related_strinfos): Likewise.
30130 (adjust_last_stmt): Likewise.
30132 2017-05-31 Richard Biener <rguenther@suse.de>
30135 * config/i386/i386.c (ix86_expand_builtin): Remove assert
30136 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
30138 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
30140 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
30141 loop_vinfo argument and use of dependence distance vectors.
30142 Check instead whether the two references differ only in their
30143 initial value and assume that they have the same alignment if the
30144 difference is a multiple of the vector alignment.
30145 (vect_analyze_data_refs_alignment): Update call accordingly.
30147 2017-05-31 Martin Liska <mliska@suse.cz>
30150 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
30152 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30154 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
30155 (create_intersect_range_checks): Move from ...
30156 * tree-data-ref.c (create_intersect_range_checks_index)
30157 (create_intersect_range_checks): ... to here.
30158 (create_runtime_alias_checks): New function factored from ...
30159 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
30160 here. Call above function.
30161 * tree-data-ref.h (create_runtime_alias_checks): New function.
30163 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30165 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
30166 segment length for dr_b and compute it in wide_int.
30168 2017-05-31 Richard Biener <rguenther@suse.de>
30170 PR tree-optimization/80906
30171 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
30172 and pass through iv_map.
30173 (copy_bb_and_scalar_dependences): Adjust.
30174 (translate_pending_phi_nodes): Likewise.
30175 (copy_loop_close_phi_args): Handle code-generating IVs instead
30178 2017-05-30 David Malcolm <dmalcolm@redhat.com>
30180 * diagnostic-color.c (color_dict): Add "type-diff".
30181 (parse_gcc_colors): Update comment.
30182 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
30183 -fdiagnostics-show-template-tree and -fno-elide-type.
30184 (GCC_COLORS): Add type-diff to example.
30186 (-fdiagnostics-show-template-tree): New.
30187 (-fno-elide-type): New.
30188 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
30189 the pp_format_decoder callback. Call any m_format_postprocessor's
30191 (pretty_printer::pretty_printer): Initialize
30192 m_format_postprocessor.
30193 (pretty_printer::~pretty_printer): Delete any
30194 m_format_postprocessor.
30195 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
30196 (class format_postprocessor): New class.
30197 (struct pretty_printer::format_decoder): Document the new parameters.
30198 (struct pretty_printer::m_format_postprocessor): New field.
30199 * tree-diagnostic.c (default_tree_printer): Update for new
30200 bool and const char ** params.
30201 * tree-diagnostic.h (default_tree_printer): Likewise.
30203 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
30205 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
30206 (lwa_operand): Delete rs6000_gen_cell_microcode test.
30207 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
30208 rs6000_gen_cell_microcode code.
30209 (rs6000_final_prescan_insn): Delete.
30210 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
30211 "warn-cell-microcode" entries.
30212 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
30213 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
30214 throughout. Change cc_reg_not_micro_cr0_operand to
30215 cc_reg_not_cr0_operand throughout.
30216 (*extendhi<mode>2_noload): Delete.
30217 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
30218 (mwarn-cell-microcode): Delete.
30219 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
30220 -mgen-cell-microcode and -mwarn-cell-microcode.
30222 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
30225 * config/i386/constraints.md (Yd): New constraint.
30227 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
30228 and (?Yd, r) alternatives. Update insn attributes.
30229 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
30230 and (?*Yd, r) alternatives. Update insn attributes.
30231 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
30233 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
30235 * gimplify.c (gimplify_modify_expr): Don't create a
30236 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
30239 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
30241 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
30243 2017-05-30 Richard Biener <rguenther@suse.de>
30245 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
30246 and reduc_def fields.
30247 (STMT_VINFO_REDUC_TYPE): New define.
30248 (STMT_VINFO_REDUC_DEF): Likewise.
30249 (vect_force_simple_reduction): Adjust prototype.
30250 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
30251 (vect_is_simple_reduction): Remove check_reduction argument.
30252 (vect_force_simple_reduction): Adjust and set
30253 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30254 (vectorizable_reduction): Do not re-do reduction analysis
30255 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30256 * tree-parloops.c (gather_scalar_reductions): Adjust.
30258 2017-05-30 Richard Biener <rguenther@suse.de>
30260 PR middle-end/80901
30261 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
30264 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
30266 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
30267 Introduce unknown_misalignment parameter and remove vf.
30268 (vect_peeling_hash_get_lowest_cost):
30269 Pass unknown_misalignment parameter.
30270 (vect_enhance_data_refs_alignment):
30271 Fix unsupportable data ref treatment.
30273 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30275 * tree-vect-data-refs.c (vect_get_data_access_cost):
30276 Workaround for SLP handling.
30277 (vect_enhance_data_refs_alignment):
30278 Compute costs for doing no peeling at all, compare to the best
30279 peeling costs so far and avoid peeling if cheaper.
30281 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30283 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
30284 Return peeling info and set costs to zero for unlimited cost
30286 (vect_enhance_data_refs_alignment): Also inspect all datarefs
30287 with unknown misalignment. Compute and costs for unknown
30288 misalignment, compare them to the costs for known misalignment
30289 and choose the cheapest for peeling.
30291 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30293 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
30294 (vect_get_peeling_costs_all_drs): Create function.
30295 (vect_peeling_hash_get_lowest_cost):
30296 Use vect_get_peeling_costs_all_drs.
30297 (vect_peeling_supportable): Create function.
30298 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
30300 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30302 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
30303 DR_HAS_NEGATIVE_STEP.
30304 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
30305 (vect_enhance_data_refs_alignment): Use.
30306 (vect_duplicate_ssa_name_ptr_info): Use.
30307 * tree-vectorizer.h (dr_misalignment): Use.
30308 (known_alignment_for_access_p): Use.
30310 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
30313 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
30315 (has_section_name): New function.
30317 2017-05-30 Martin Liska <mliska@suse.cz>
30320 * auto-profile.c (get_function_decl_from_block): Fix
30323 2017-05-30 Richard Biener <rguenther@suse.de>
30325 PR middle-end/80876
30326 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
30328 2017-05-30 Martin Liska <mliska@suse.cz>
30330 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
30331 * dumpfile.h (struct dump_file_info): Remove ctors.
30333 2017-05-30 Martin Liska <mliska@suse.cz>
30335 * predict.def: Fix GNU coding style.
30337 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30339 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30340 Mark 'to' argument with ATTRIBUTE_UNUSED.
30342 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30344 * config/xtensa/xtensa.c (xtensa_emit_call): Use
30345 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
30346 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
30349 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
30351 * doc/install.texi (Options specification): Restore entry of
30352 --enable-sjlj-exceptions.
30354 2017-05-27 Michael Eager <eager@eagercon.com>
30357 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
30359 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
30361 * config/microblaze/microblaze.h
30362 (FIXED_REGISTERS): Update in macro.
30363 (CALL_USED_REGISTERS): Update in macro.
30365 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
30367 * doc/install.texi: Add links to macOS binary distributions.
30369 2017-05-27 Jakub Jelinek <jakub@redhat.com>
30373 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30375 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30377 2017-05-26 Martin Liska <mliska@suse.cz>
30379 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
30381 2017-05-26 Martin Liska <mliska@suse.cz>
30383 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
30384 always leading ';; '.
30385 (dump_bb_info): Likewise.
30386 (brief_dump_cfg): Likewise.
30387 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
30388 * dumpfile.c: Remove usage of TDF_VERBOSE.
30389 * dumpfile.h (enum dump_kind): Likewise.
30390 (dump_gimple_bb_header): Do not use TDF_COMMENT.
30391 * print-tree.c (debug_verbose): Remove.
30392 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
30393 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
30394 * tree-diagnostic.c (default_tree_printer): Replace
30395 TDF_DIAGNOSTIC with TDF_SLIM.
30397 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30399 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
30400 in parameter loop, rather than loop_vinfo.
30401 (create_intersect_range_checks): Ditto.
30402 (vect_create_cond_for_alias_checks): Update call to above functions.
30404 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30406 PR tree-optimization/80815
30407 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
30408 for merging runtime alias checks. Handle negative DR_STEPs.
30410 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30412 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
30414 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
30415 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
30416 out code pruning runtime alias checks.
30417 * tree-data-ref.c (prune_runtime_alias_test_list): New function
30418 factored out from above.
30419 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30421 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30423 (prune_runtime_alias_test_list): New decalaration.
30425 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30427 * tree-vect-data-refs.c (compare_tree): Rename and move ...
30428 * tree-data-ref.c (data_ref_compare_tree): ... to here.
30429 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
30430 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
30431 (operator==, comp_dr_with_seg_len_pair): Ditto.
30432 (vect_prune_runtime_alias_test_list): Ditto.
30434 2017-05-26 Martin Liska <mliska@suse.cz>
30437 * params.def: Bound partial-inlining-entry-probability param.
30439 2017-05-26 Marek Polacek <polacek@redhat.com>
30442 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
30445 2017-05-26 Richard Biener <rguenther@suse.de>
30447 PR tree-optimization/80842
30448 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
30451 2017-05-26 Richard Biener <rguenther@suse.de>
30453 PR tree-optimization/80844
30454 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
30456 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
30458 * doc/md.texi (Machine Constraints): Update x86 family
30459 machine constraints section to match 'config/i386/constraints.md'.
30461 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
30463 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
30465 2017-05-25 Nathan Sidwell <nathan@acm.org>
30467 * doc/invoke.texi (--enable-languages): Update documentation.
30469 2017-05-25 Martin Liska <mliska@suse.cz>
30471 * dumpfile.c: Add TDF_FOLDING.
30472 * dumpfile.h (enum dump_kind): Likewise.
30473 * genmatch.c (dt_simplify::gen_1): Use it.
30475 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30477 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
30479 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30481 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30482 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
30484 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30486 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
30487 * match.pd (X == C): Rewrite it here.
30488 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
30489 with_certain_nonzero_bits2): New predicates.
30490 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
30492 2017-05-24 Nathan Sidwell <nathan@acm.org>
30494 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
30497 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
30500 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30502 * config/powerpcspe: New port. Files are copied from the rs6000
30503 port, with "rs6000" in filenames replaced by "powerpcspe".
30505 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
30507 PR rtl-optimization/80754
30508 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
30510 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
30512 * config/sparc/sparc.md (length): Return the correct value for -mflat
30513 sibcalls to match output_sibcall.
30515 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30519 * config/rs6000/rs6000.c (struct machine_function): Add new field
30521 (rs6000_get_separate_components): Init that field, use it.
30522 (rs6000_components_for_bb): Use the field.
30524 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
30526 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
30528 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
30530 PR middle-end/80823
30531 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
30533 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30536 * config/s390/s390.c (s390_check_qrst_address): Check incoming
30537 address against address_operand predicate.
30538 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
30540 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
30542 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
30543 parameters passed indirectly.
30545 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
30547 * config/i386/i386.md (*movdi_internal): Remove SSE4
30548 alternative 18 (?r, *v). Update insn attributes.
30549 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
30550 Update insn attributes.
30551 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
30552 Update insn attributes.
30553 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
30554 alternative 1 (r, v). Remove isa attribute.
30555 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
30556 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
30557 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
30559 2017-05-23 Tom de Vries <tom@codesourcery.com>
30561 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
30564 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30566 * cgraphunit.c (symbol_table::process_new_functions): Update.
30567 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
30568 (inline_generate_summary): Rename to ...
30569 (ipa_fn_summary_generate): ... this one.
30570 (inline_read_summary): Rename to ...
30571 (ipa_fn_summary_read): ... this one.
30572 (inline_write_summary): Rename to ...
30573 (ipa_fn_summary_write): ... this one.
30574 (inline_free_summary): Rename to ...
30575 (ipa_free_fn_summary): ... this one.
30576 (pass_data_local_fn_summary, pass_local_fn_summary,
30577 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
30578 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
30579 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
30580 make_pass_ipa_fn_summary): New.
30581 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
30582 inline_write_summary, inline_free_summary): Remove.
30583 (ipa_free_fn_summary) : New.
30584 * ipa-inline.c (ipa_inline): Update.
30585 (pass_ipa_inline): Do not generate summaries.
30586 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
30588 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
30589 and add pass_ipa_fn_summary.
30590 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
30592 (make_pass_inline_parameters): Remove.
30594 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
30596 * omp-low.c (struct omp_context): Remove "default_kind" member.
30599 * omp-offload.c (execute_oacc_device_lower): Remove the
30600 parallelism dimensions function attributes for unparallelized
30601 OpenACC kernels constructs.
30603 2017-05-23 Martin Liska <mliska@suse.cz>
30605 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
30607 (cgraph_edge::make_speculative): Likewise.
30608 (cgraph_edge::resolve_speculation): Likewise.
30609 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30610 (cgraph_node::dump): Likewise.
30611 * cgraph.h: Likewise.
30612 * cgraphunit.c (analyze_functions): Likewise.
30613 (symbol_table::compile): Likewise.
30614 * ipa-cp.c (print_all_lattices): Likewise.
30615 (determine_versionability): Likewise.
30616 (initialize_node_lattices): Likewise.
30617 (ipcp_verify_propagated_values): Likewise.
30618 (estimate_local_effects): Likewise.
30619 (update_profiling_info): Likewise.
30620 (create_specialized_node): Likewise.
30621 (perhaps_add_new_callers): Likewise.
30622 (decide_about_value): Likewise.
30623 (decide_whether_version_node): Likewise.
30624 (identify_dead_nodes): Likewise.
30625 (ipcp_store_bits_results): Likewise.
30626 * ipa-devirt.c (dump_targets): Likewise.
30627 (ipa_devirt): Likewise.
30628 * ipa-icf.c (sem_item::dump): Likewise.
30629 (sem_function::equals): Likewise.
30630 (sem_variable::equals): Likewise.
30631 (sem_item_optimizer::read_section): Likewise.
30632 (sem_item_optimizer::execute): Likewise.
30633 (congruence_class::dump): Likewise.
30634 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
30635 (dump_inline_summary): Likewise.
30636 (estimate_node_size_and_time): Likewise.
30637 (inline_analyze_function): Likewise.
30638 * ipa-inline-transform.c (inline_call): Likewise.
30639 * ipa-inline.c (report_inline_failed_reason): Likewise.
30640 (want_early_inline_function_p): Likewise.
30641 (edge_badness): Likewise.
30642 (update_edge_key): Likewise.
30643 (inline_small_functions): Likewise.
30644 * ipa-profile.c (ipa_profile): Likewise.
30645 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
30646 (ipa_make_edge_direct_to_target): Likewise.
30647 (remove_described_reference): Likewise.
30648 (ipa_impossible_devirt_target): Likewise.
30649 (propagate_controlled_uses): Likewise.
30650 (ipa_print_node_params): Likewise.
30651 (ipcp_transform_function): Likewise.
30652 * ipa-pure-const.c (pure_const_read_summary): Likewise.
30653 (propagate_pure_const): Likewise.
30654 * ipa-reference.c (generate_summary): Likewise.
30655 (read_write_all_from_decl): Likewise.
30656 (propagate): Likewise.
30657 (ipa_reference_read_optimization_summary): Likewise.
30658 * ipa-utils.c (ipa_merge_profiles): Likewise.
30659 * ipa.c (walk_polymorphic_call_targets): Likewise.
30660 (symbol_table::remove_unreachable_nodes): Likewise.
30661 (ipa_single_use): Likewise.
30662 * passes.c (execute_todo): Likewise.
30663 * predict.c (drop_profile): Likewise.
30664 * symtab.c (symtab_node::get_dump_name): New function.
30665 (symtab_node::dump_name): Likewise.
30666 (symtab_node::dump_asm_name): Likewise.
30667 (symtab_node::dump_references): Likewise.
30668 (symtab_node::dump_referring): Likewise.
30669 (symtab_node::dump_base): Likewise.
30670 (symtab_node::debug_symtab): Likewise.
30671 * tree-sra.c (convert_callers_for_node): Likewise.
30672 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
30673 * value-prof.c (init_node_map): Likewise.
30675 2017-05-23 Martin Liska <mliska@suse.cz>
30677 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
30678 and symtab_node::debug_symtab to symbol_table::debug.
30679 * cgraphunit.c (analyze_functions): Use the renamed function.
30680 (symbol_table::compile): Likewise.
30681 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
30682 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
30683 * passes.c (execute_todo): Likewise.
30684 * symtab.c (symbol_table::dump): New function.
30685 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
30687 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30689 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
30690 that nonconst implies exec.
30692 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30694 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30695 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30696 (inline_edge_summary_vec): Turn into ...
30697 (ipa_call_summaries): ... this one.
30698 (redirect_to_unreachable, edge_set_predicate,
30699 evaluate_properties_for_edge, inline_summary_alloc,
30700 reset_ipa_call_summary, reset_inline_summary,
30701 inline_summary_t::duplicate): Update.
30702 (inline_edge_duplication_hook): Turn to ...
30703 (ipa_call_summary_t::duplicate): ... this one.
30704 (inline_edge_removal_hook): Turn to ...
30705 (ipa_call_summary_t::remove): ... this one.
30706 (dump_inline_edge_summary): Turn to ...
30707 (dump_ipa_call_summary): ... this one.
30708 (estimate_function_body_sizes): Update.
30709 (inline_update_callee_summaries): Update.
30710 (remap_edge_change_prob): Update.
30711 (remap_edge_summaries): Update.
30712 (inline_merge_summary): Update.
30713 (do_estimate_edge_time): Update.
30714 (inline_generate_summary): Update.
30715 (inline_read_section): Update.
30716 (inline_read_summary): Update.
30717 (inline_free_summary): Update.
30718 * ipa-inline.c (can_inline_edge_p): Update.
30719 (compute_inlined_call_time): Update.
30720 (want_inline_small_function_p): Update.
30721 (edge_badness): Update.
30722 (early_inliner): Update.
30723 * ipa-inline.h (inline_edge_summary): Turn to ...
30724 (ipa_call_summary): ... this one.
30725 (ipa_call_summary_t): New class.
30726 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30727 (ipa_call_summaries): New.
30728 (inline_edge_summary): Remove.
30729 (estimate_edge_growth): Update.
30730 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30731 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30732 * ipa-split.c (execute_split_functions): Update.
30733 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30735 2017-05-23 Tom de Vries <tom@codesourcery.com>
30737 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30738 attributes): Document rdrand effective target.
30740 2017-05-23 Tom de Vries <tom@codesourcery.com>
30742 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30743 attributes): Sort alphabetically.
30745 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
30747 * config/avr/genmultilib.awk: Use gsub instead of gensub.
30749 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
30752 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
30753 V2DF/V2DI splat into two separate patterns, one that handles
30754 registers, and the other that only handles memory. Drop support
30755 for splatting from a GPR on ISA 2.07 and then splitting the
30756 splat into direct move and splat.
30757 (vsx_splat_<mode>_reg): Likewise.
30758 (vsx_splat_<mode>_mem): Likewise.
30760 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
30762 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
30764 2017-05-22 Jakub Jelinek <jakub@redhat.com>
30766 PR middle-end/80809
30767 * omp-low.c (finish_taskreg_remap): New function.
30768 (finish_taskreg_scan): If unit size of ctx->record_type
30769 is non-constant, unshare the size expression and replace
30770 decls in it with possible outer var refs.
30772 PR middle-end/80809
30773 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
30774 GOVD_SHARED rather than GOVD_PRIVATE with it.
30775 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30776 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30778 PR middle-end/80853
30779 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30780 as last argument to build_outer_var_ref for pointer bases of array
30781 section reductions.
30783 2017-05-19 Martin Sebor <msebor@redhat.com>
30785 * print-tree.c (print_node): Print DECL_READ_P flag.
30787 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30789 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30790 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30791 * cgraph.c: Likewise.
30792 * cgraphunit.c: Likewise.
30793 * gengtype.c: Likewise.
30794 * ipa-cp.c: Likewise.
30795 * ipa-devirt.c: Likewise.
30796 * ipa-icf.c: Likewise.
30797 * ipa-predicate.c: Likewise.
30798 * ipa-profile.c: Likewise.
30799 * ipa-prop.c: Likewise.
30800 * ipa-split.c: Likewise.
30802 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30803 edge_predicate_pool, dump_inline_hints,
30804 inline_summary::account_size_time, redirect_to_unreachable,
30805 edge_set_predicate, set_hint_predicate,
30806 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30807 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30808 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30809 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30810 ipa_call_summary_t::remove, initialize_growth_caches,
30811 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30812 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30813 mark_modified, unmodified_parm_1, unmodified_parm,
30814 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30815 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30816 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30817 will_be_nonconstant_predicate, record_modified_bb_info,
30818 get_minimal_bb, record_modified, param_change_prob,
30819 phi_result_unknown_predicate, predicate_for_phi_result,
30820 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30821 estimate_function_body_sizes, compute_inline_parameters,
30822 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30823 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30824 inline_update_callee_summaries, remap_edge_change_prob,
30825 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30826 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30827 inline_analyze_function, inline_summary_t::insert,
30828 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30829 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30830 inline_free_summary): Move to ipa-fnsummary.h
30831 (predicate_t): Remove.
30832 * ipa-fnsummary.c: New file.
30833 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30834 (enum inline_hints_vals, inline_hints, agg_position_info,
30835 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30836 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30837 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30838 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30839 inline_read_summary, inline_write_summary, inline_free_summary,
30840 inline_analyze_function, initialize_inline_failed,
30841 inline_merge_summary, inline_update_overall_summary,
30842 compute_inline_parameters): Move to ipa-fnsummary.h
30843 * ipa-fnsummary.h: New file.
30844 * ipa-inline-transform.h: Include ipa-inline.h.
30845 * ipa-inline.c: LIkewise.
30847 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30849 * ipa-inline.c (edge_badness): Use inlined_time instead of
30850 inline_summaries->get.
30852 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30854 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30856 2017-05-22 Nathan Sidwell <nathan@acm.org>
30858 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30859 (fdump-lang): Document 'raw' option.
30860 * dumpfile.h (TDI_tu): Delete.
30861 * dumpfile.c (dump_files): Remove translation-unit.
30862 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30864 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30866 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30867 command option from $(AWK) call.
30868 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30869 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30870 [FORMAT]: Remove handling of variable.
30871 * config/avr/t-multilib: Regenerate.
30873 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30875 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30877 (dump_inline_summary): Do not print self_time.
30878 (estimate_function_body_sizes): Do not set self_time.
30879 (compute_inline_parameters): Likewise.
30880 (inline_read_section, inline_write_summary): Do not stream self_time.
30881 * ipa-inline.h (inline_summary): Drop self_time.
30883 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30885 * ipa-inline-analysis.c (account_size_time): Rename to ...
30886 (inline_summary::account_size_time): ... this one.
30887 (reset_ipa_call_summary): Turn to ...
30888 (ipa_call_summary::reset): ... this one.
30889 (reset_inline_summary): Turn to ...
30890 (inline_summary::reset): ... this one.
30891 (inline_summary_t::remove): Update.
30892 (inline_summary_t::duplicate): Update.
30893 (ipa_call_summary_t::remove): Update.
30894 (dump_inline_summary): Update.
30895 (estimate_function_body_sizes): Update.
30896 (compute_inline_parameters): Update.
30897 (estimate_node_size_and_time): Update.
30898 (inline_merge_summary): Update.
30899 (inline_update_overall_summary): Update.
30900 (inline_read_section): Update.
30901 (inline_write_summary): Update.
30902 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30903 add account_size_time and reset member functions.
30904 (ipa_call_summary): Add reset function.
30905 * ipa-predicate.h (predicate::operator &): Constify.
30907 2017-05-22 Richard Biener <rguenther@suse.de>
30909 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30911 2017-05-19 Jason Merrill <jason@redhat.com>
30913 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30915 2017-05-19 Marek Polacek <polacek@redhat.com>
30918 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30919 TYPE_OVERFLOW_WRAPS checks.
30921 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30923 * tree-core.h (enum omp_clause_default_kind): Add
30924 "OMP_CLAUSE_DEFAULT_PRESENT".
30925 * tree-pretty-print.c (dump_omp_clause): Handle it.
30926 * gimplify.c (enum gimplify_omp_var_data): Add
30927 "GOVD_MAP_FORCE_PRESENT".
30928 (gimplify_adjust_omp_clauses_1): Map it to
30929 "GOMP_MAP_FORCE_PRESENT".
30930 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30932 * gimplify.c (oacc_default_clause): Clarify.
30934 2017-05-19 Nathan Sidwell <nathan@acm.org>
30936 LANG_HOOK_REGISTER_DUMPS
30937 * toplev.c (general_init): Call register dump lang hook.
30938 * doc/invoke.texi: Document -fdump-lang option family.
30939 * dumpfile.c (dump_files): Remove class dump here.
30940 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30941 * dumpfile.h (tree_dump_index): Remove TDI_class.
30942 * langhooks-def.h (lhd_register_dumps): Declare.
30943 (LANG_HOOKS_REGISTER_DUMPS): Define.
30944 (LANG_HOOKS_INITIALIZER): Add it.
30945 * langhooks.c (lhd_register_dumps): Define.
30946 * langhooks.h (struct lang_hooks): Add register_dumps.
30948 2017-05-19 Nathan Sidwell <nathan@acm.org>
30950 * context.h (context::set_passes): New.
30951 * context.c (context::context): Do not create pass manager.
30952 * toplev.c (general_init): Create pass manager here.
30954 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30956 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30957 use this splitter if two add or or instructions would also work for
30958 the constant we want to generate.
30960 2017-05-19 Richard Biener <rguenther@suse.de>
30963 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30964 predicate evaluation.
30966 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30968 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30970 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30973 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30975 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30976 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30977 (inline_edge_summary_vec): Turn into ...
30978 (ipa_call_summaries): ... this one.
30979 (redirect_to_unreachable, edge_set_predicate,
30980 evaluate_properties_for_edge, inline_summary_alloc,
30981 reset_ipa_call_summary, reset_inline_summary,
30982 inline_summary_t::duplicate): Update.
30983 (inline_edge_duplication_hook): Turn to ...
30984 (ipa_call_summary_t::duplicate): ... this one.
30985 (inline_edge_removal_hook): Turn to ...
30986 (ipa_call_summary_t::remove): ... this one.
30987 (dump_inline_edge_summary): Turn to ...
30988 (dump_ipa_call_summary): ... this one.
30989 (estimate_function_body_sizes): Update.
30990 (inline_update_callee_summaries): Update.
30991 (remap_edge_change_prob): Update.
30992 (remap_edge_summaries): Update.
30993 (inline_merge_summary): Update.
30994 (do_estimate_edge_time): Update.
30995 (inline_generate_summary): Update.
30996 (inline_read_section): Update.
30997 (inline_read_summary): Update.
30998 (inline_free_summary): Update.
30999 * ipa-inline.c (can_inline_edge_p): Update.
31000 (compute_inlined_call_time): Update.
31001 (want_inline_small_function_p): Update.
31002 (edge_badness): Update.
31003 (early_inliner): Update.
31004 * ipa-inline.h (inline_edge_summary): Turn to ...
31005 (ipa_call_summary): ... this one.
31006 (ipa_call_summary_t): New class.
31007 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
31008 (ipa_call_summaries): New.
31009 (inline_edge_summary): Remove.
31010 (estimate_edge_growth): Update.
31011 * ipa-profile.c (ipa_propagate_frequency_1): Update.
31012 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
31013 * ipa-split.c (execute_split_functions): Update.
31014 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
31016 2017-05-19 Richard Biener <rguenther@suse.de>
31018 PR middle-end/80764
31019 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
31021 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
31023 * config/rs6000/rs6000.c (struct machine_function): Add field
31024 fpr_is_wrapped_separately.
31025 (rs6000_get_separate_components): Use 64 components. Handle the
31026 new FPR components.
31027 (rs6000_components_for_bb): Handle the FPR components.
31028 (rs6000_emit_prologue_components): Handle the FPR components.
31029 (rs6000_emit_epilogue_components): Handle the FPR components.
31030 (rs6000_set_handled_components): Handle the FPR components.
31031 (rs6000_emit_prologue): Don't output prologue code for those FPRs
31032 that are already separately shrink-wrapped.
31033 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
31034 that are already separately shrink-wrapped.
31036 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31039 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
31042 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
31043 (define_peephole2 for Altivec d-form load): Add peepholes to catch
31044 cases where the register allocator uses a move and an offsettable
31045 memory operation to/from a FPR register on ISA 2.06/2.07.
31046 (define_peephole2 for Altivec d-form store): Likewise.
31048 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
31051 * config/i386/mmx.md (*mov<mode>_internal): Enable
31052 alternatives 11, 12, 13 and 14 also for 32bit targets.
31053 Remove alternatives 15, 16, 17 and 18.
31054 * config/i386/sse.md (vec_concatv2di): Change
31055 alternative (!x, *y) to (x, ?!*Yn).
31057 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
31059 * dumpfile.h (enum dump_kind): Remove stray comma.
31061 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
31063 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
31064 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
31065 predicate::num_conditions
31066 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
31067 (CHANGED): turn into predicate::changed.
31068 (agg_position_info): Move to ipa-predicate.h
31069 (add_condition, predicate::add_clause, predicate::operator &=,
31070 predicate::or_with, predicate::evaluate, predicate::probability,
31071 dump_condition, dump_clause, predicate::dump,
31072 predicate::remap_after_duplication, predicate::remap_after_inlining,
31073 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
31074 (evaluate_conditions_for_known_args): Update.
31075 (set_cond_stmt_execution_predicate): Update.
31076 * ipa-inline.h: Include ipa-predicate.h
31077 (condition, inline_param_summary, conditions, agg_position_info,
31078 predicate): Move to ipa-predicate.h
31079 * ipa-predicate.c: New file.
31080 * ipa-predicate.h: New file.
31082 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
31084 * final.c (leaf_function_p): Check we are not in a sequence.
31086 2017-05-18 Martin Liska <mliska@suse.cz>
31088 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
31089 * dumpfile.c (dump_register): Use new enum dump_kind.
31090 (get_dump_file_name): Likewise.
31091 (dump_enable_all): Likewise.
31092 (dump_switch_p_1): Likewise.
31093 (enable_rtl_dump_file): Remove usage of TDF_RTL.
31094 * dumpfile.h (enum dump_kind): New enum type.
31095 (struct dump_file_info): Create constructor and
31096 format fields and comments.
31097 * passes.c (pass_manager::register_one_dump_file):
31099 * statistics.c (statistics_early_init): Likewise.
31100 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
31101 TDF_TREE with TDF_SLIM.
31102 (gather_memory_references_ref): Likewise.
31104 2017-05-18 Martin Liska <mliska@suse.cz>
31106 * vec.h (struct vnull): Use it.
31108 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
31110 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
31111 (true_predicate, false_predicate, true_predicate_p,
31112 false_predicate_p): Remove.
31113 (single_cond_predicate, not_inlined_predicate): Turn to member function
31115 (add_condition): Update.
31116 (add_clause): Turn to...
31117 (predicate::add_clause): ... this one; update; allow passing NULL
31119 (and_predicates): Turn to ...
31120 (predicate::operator &=): ... this one.
31121 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
31122 (or_predicates): Turn to ...
31123 (predicate::or_with): ... this one.
31124 (evaluate_predicate): Turn to ...
31125 (predicate::evaluate): ... this one.
31126 (predicate_probability): Turn to ...
31127 (predicate::probability): ... this one.
31128 (dump_condition): Update.
31129 (dump_predicate): Turn to ...
31130 (predicate::dump): ... this one.
31131 (account_size_time): Update.
31132 (edge_set_predicate): Update.
31133 (set_hint_predicate): UPdate.
31134 (evaluate_conditions_for_known_args): Update.
31135 (evaluate_properties_for_edge): Update.
31136 (remap_predicate_after_duplication): Turn to...
31137 (predicate::remap_after_duplication): ... this one.
31138 (remap_hint_predicate_after_duplication): Update.
31139 (inline_summary_t::duplicate): UPdate.
31140 (dump_inline_edge_summary): Update.
31141 (dump_inline_summary): Update.
31142 (set_cond_stmt_execution_predicate): Update.
31143 (set_switch_stmt_execution_predicate): Update.
31144 (compute_bb_predicates): Update.
31145 (will_be_nonconstant_expr_predicate): Update.
31146 (will_be_nonconstant_predicate): Update.
31147 (phi_result_unknown_predicate): Update.
31148 (predicate_for_phi_result): Update.
31149 (array_index_predicate): Update.
31150 (estimate_function_body_sizes): Update.
31151 (estimate_node_size_and_time): Update.
31152 (estimate_ipcp_clone_size_and_time): Update.
31153 (remap_predicate): Rename to ...
31154 (predicate::remap_after_inlining): ... this one.
31155 (remap_hint_predicate): Update.
31156 (inline_merge_summary): Update.
31157 (inline_update_overall_summary): Update.
31158 (estimate_size_after_inlining): Update.
31159 (read_predicate): Rename to ...
31160 (predicate::stream_in): ... this one.
31161 (read_inline_edge_summary): Update.
31162 (write_predicate): Rename to ...
31163 (predicate::stream_out): ... this one.
31164 (write_inline_edge_summary): Update.
31165 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
31166 (clause_t): Turn to uint32_t
31167 (predicate): Turn to class; implement constructor and operators
31169 (size_time_entry): Update.
31170 (inline_summary): Update.
31171 (inline_edge_summary): Update.
31173 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
31175 * fold-const.c (fold_binary_loc): Move transformation...
31176 * match.pd (C - X CMP X): ... here.
31178 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
31180 * config/sparc/sparc.c (sparc_option_override): Set function
31181 alignment for -mcpu=niagara7 to 64 to match the I$ line.
31182 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
31184 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
31186 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
31188 2017-05-18 Marek Polacek <polacek@redhat.com>
31191 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
31192 (pass_ubsan::execute): Call gimple_assign_single_p instead of
31193 gimple_assign_load_p.
31195 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
31197 PR middle-end/80692
31198 * real.c (do_compare): Give decimal_do_compare preference over
31199 comparing just the signs.
31201 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
31203 * doc/md.texi (Canonicalization of Instructions): Describe the
31204 canonical form of instructions that inherently set a condition
31207 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
31209 PR middle-end/80775
31210 * tree-cfg.c: Move deletion of unreachable case statements to after
31211 the merging of consecutive case labels.
31213 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
31215 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
31216 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
31217 restoring of callee-saved registers.
31219 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
31221 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
31222 * config/visium/visium.c (single_set_and_flags): Likewise.
31223 * config/visium/visium.md (Substitutions): Likewise.
31225 2017-05-17 Martin Liska <mliska@suse.cz>
31227 * cfg.c: Introduce dump_flags_t type and
31228 use it instead of int type.
31230 * cfghooks.c: Likewise.
31231 * cfghooks.h (struct cfg_hooks): Likewise.
31232 * cfgrtl.c: Likewise.
31233 * cfgrtl.h: Likewise.
31234 * cgraph.c (cgraph_node::get_body): Likewise.
31235 * coretypes.h: Likewise.
31236 * domwalk.c: Likewise.
31237 * domwalk.h: Likewise.
31238 * dumpfile.c (struct dump_option_value_info): Likewise.
31239 (dump_enable_all): Likewise.
31240 (dump_switch_p_1): Likewise.
31241 (opt_info_switch_p): Likewise.
31242 * dumpfile.h (enum tree_dump_index): Likewise.
31243 (struct dump_file_info): Likewise.
31244 * genemit.c: Likewise.
31245 * generic-match-head.c: Likewise.
31246 * gengtype.c (open_base_files): Likewise.
31247 * gimple-pretty-print.c: Likewise.
31248 * gimple-pretty-print.h: Likewise.
31249 * graph.c (print_graph_cfg): Likewise.
31250 * graphite-scop-detection.c (dot_all_sese): Likewise.
31251 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
31252 * loop-unroll.c (report_unroll): Likewise.
31253 * passes.c (pass_manager::register_one_dump_file): Likewise.
31254 * print-tree.c: Likewise.
31255 * statistics.c: Likewise.
31256 * tree-cfg.c: Likewise.
31257 * tree-cfg.h: Likewise.
31258 * tree-dfa.c: Likewise.
31259 * tree-dfa.h: Likewise.
31260 * tree-dump.c (dump_function): Likewise.
31261 * tree-dump.h (struct dump_info): Likewise.
31262 * tree-pretty-print.c: Likewise.
31263 * tree-pretty-print.h: Likewise.
31264 * tree-ssa-live.c: Likewise.
31265 * tree-ssa-live.h: Likewise.
31266 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
31267 * tree-vect-loop.c: Likewise.
31268 * tree-vect-slp.c: Likewise.
31270 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
31271 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31273 PR tree-optimization/80457
31274 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
31275 of all arguments to a statement as scalar_to_vec operations.
31276 (vectorizable_call): Adjust call to vect_model_simple_cost for
31278 (vectorizable_conversion): Likewise.
31279 (vectorizable_assignment): Likewise.
31280 (vectorizable_shift): Likewise.
31281 (vectorizable_operation): Likewise.
31282 (vectorizable_comparison): Likewise.
31283 (vect_is_simple_cond): Record the def types for operands.
31284 (vectorizable_condition): Likewise, call vect_model_simple_cost.
31285 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
31286 for statement argument count.
31288 2017-05-16 Carl Love <cel@us.ibm.com>
31290 * config/rs6000/rs6000-c: Add support for built-in functions
31291 vector unsigned long long vec_bperm (vector unsigned long long,
31292 vector unsigned char)
31293 vector signed long long vec_mule (vector signed int,
31295 vector unsigned long long vec_mule (vector unsigned int,
31296 vector unsigned int)
31297 vector signed long long vec_mulo (vector signed int,
31299 vector unsigned long long vec_mulo (vector unsigned int,
31300 vector unsigned int)
31301 vector signed char vec_sldw (vector signed char,
31302 vector signed char,
31304 vector unsigned char vec_sldw (vector unsigned char,
31305 vector unsigned char,
31307 vector signed short vec_sldw (vector signed short,
31308 vector signed short,
31310 vector unsigned short vec_sldw (vector unsigned short,
31311 vector unsigned short,
31313 vector signed int vec_sldw (vector signed int,
31316 vector unsigned int vec_sldw (vector unsigned int,
31317 vector unsigned int,
31319 vector signed long long vec_sldw (vector signed long long,
31320 vector signed long long,
31322 vector unsigned long long vec_sldw (vector unsigned long long,
31323 vector unsigned long long,
31325 * config/rs6000/rs6000-c: Add support for built-in functions
31326 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
31327 * config/rs6000/altivec.h: Add defintion for vec_sldw.
31328 * doc/extend.texi: Update the built-in documentation for the
31329 new built-in functions.
31331 2017-05-16 Marek Polacek <polacek@redhat.com>
31335 * tree.c (save_expr): Don't fold the expression.
31337 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
31339 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
31340 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
31341 and (?*y,m). Update insn attributes.
31343 2017-05-16 Martin Liska <mliska@suse.cz>
31345 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
31346 flags argument of print_gimple_stmt, print_gimple_expr,
31347 print_generic_stmt and print_generic_expr.
31348 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
31349 * coretypes.h: Likewise.
31350 * except.c (dump_eh_tree): Likewise.
31351 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
31352 * gimple-pretty-print.h: Likewise.
31353 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
31354 (backprop::push_to_worklist): Likewise.
31355 (backprop::pop_from_worklist): Likewise.
31356 (backprop::process_use): Likewise.
31357 (backprop::intersect_uses): Likewise.
31358 (note_replacement): Likewise.
31359 * gimple-ssa-store-merging.c
31360 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
31361 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
31362 (pass_store_merging::execute): Likewise.
31363 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
31364 (ssa_base_cand_dump_callback): Likewise.
31365 (dump_incr_vec): Likewise.
31366 (replace_refs): Likewise.
31367 (replace_mult_candidate): Likewise.
31368 (create_add_on_incoming_edge): Likewise.
31369 (create_phi_basis): Likewise.
31370 (insert_initializers): Likewise.
31371 (all_phi_incrs_profitable): Likewise.
31372 (introduce_cast_before_cand): Likewise.
31373 (replace_one_candidate): Likewise.
31374 * gimplify.c (gimplify_expr): Likewise.
31375 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
31376 (set_rename): Likewise.
31377 (rename_uses): Likewise.
31378 (copy_loop_phi_nodes): Likewise.
31379 (add_close_phis_to_merge_points): Likewise.
31380 (copy_loop_close_phi_args): Likewise.
31381 (copy_cond_phi_args): Likewise.
31382 (graphite_copy_stmts_from_block): Likewise.
31383 (translate_pending_phi_nodes): Likewise.
31384 * graphite-poly.c (print_pdr): Likewise.
31385 (dump_gbb_cases): Likewise.
31386 (dump_gbb_conditions): Likewise.
31387 (print_scop_params): Likewise.
31388 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
31389 (build_cross_bb_scalars_use): Likewise.
31390 (gather_bbs::before_dom_children): Likewise.
31391 * hsa-dump.c (dump_hsa_immed): Likewise.
31392 * ipa-cp.c (print_ipcp_constant_value): Likewise.
31393 (get_replacement_map): Likewise.
31394 * ipa-inline-analysis.c (dump_condition): Likewise.
31395 (estimate_function_body_sizes): Likewise.
31396 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
31397 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
31398 * ipa-prop.c (ipa_dump_param): Likewise.
31399 (ipa_print_node_jump_functions_for_edge): Likewise.
31400 (ipa_modify_call_arguments): Likewise.
31401 (ipa_modify_expr): Likewise.
31402 (ipa_dump_param_adjustments): Likewise.
31403 (ipa_dump_agg_replacement_values): Likewise.
31404 (ipcp_modif_dom_walker::before_dom_children): Likewise.
31405 * ipa-pure-const.c (check_stmt): Likewise.
31406 (pass_nothrow::execute): Likewise.
31407 * ipa-split.c (execute_split_functions): Likewise.
31408 * omp-offload.c (dump_oacc_loop_part): Likewise.
31409 (dump_oacc_loop): Likewise.
31410 * trans-mem.c (tm_log_emit): Likewise.
31411 (tm_memopt_accumulate_memops): Likewise.
31412 (dump_tm_memopt_set): Likewise.
31413 (dump_tm_memopt_transform): Likewise.
31414 * tree-cfg.c (gimple_verify_flow_info): Likewise.
31415 (print_loop): Likewise.
31416 * tree-chkp-opt.c (chkp_print_addr): Likewise.
31417 (chkp_gather_checks_info): Likewise.
31418 (chkp_get_check_result): Likewise.
31419 (chkp_remove_check_if_pass): Likewise.
31420 (chkp_use_outer_bounds_if_possible): Likewise.
31421 (chkp_reduce_bounds_lifetime): Likewise.
31422 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
31423 (chkp_mark_completed_bounds): Likewise.
31424 (chkp_register_incomplete_bounds): Likewise.
31425 (chkp_mark_invalid_bounds): Likewise.
31426 (chkp_maybe_copy_and_register_bounds): Likewise.
31427 (chkp_build_returned_bound): Likewise.
31428 (chkp_get_bound_for_parm): Likewise.
31429 (chkp_build_bndldx): Likewise.
31430 (chkp_get_bounds_by_definition): Likewise.
31431 (chkp_generate_extern_var_bounds): Likewise.
31432 (chkp_get_bounds_for_decl_addr): Likewise.
31433 * tree-chrec.c (chrec_apply): Likewise.
31434 * tree-data-ref.c (dump_data_reference): Likewise.
31435 (dump_subscript): Likewise.
31436 (dump_data_dependence_relation): Likewise.
31437 (analyze_overlapping_iterations): Likewise.
31438 * tree-inline.c (expand_call_inline): Likewise.
31439 (tree_function_versioning): Likewise.
31440 * tree-into-ssa.c (dump_defs_stack): Likewise.
31441 (dump_currdefs): Likewise.
31442 (dump_names_replaced_by): Likewise.
31443 (dump_update_ssa): Likewise.
31444 (update_ssa): Likewise.
31445 * tree-object-size.c (pass_object_sizes::execute): Likewise.
31446 * tree-parloops.c (build_new_reduction): Likewise.
31447 (try_create_reduction_list): Likewise.
31448 (ref_conflicts_with_region): Likewise.
31449 (oacc_entry_exit_ok_1): Likewise.
31450 (oacc_entry_exit_single_gang): Likewise.
31451 * tree-pretty-print.h: Likewise.
31452 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
31453 (get_scalar_evolution): Likewise.
31454 (add_to_evolution): Likewise.
31455 (get_loop_exit_condition): Likewise.
31456 (analyze_evolution_in_loop): Likewise.
31457 (analyze_initial_condition): Likewise.
31458 (analyze_scalar_evolution): Likewise.
31459 (instantiate_scev): Likewise.
31460 (number_of_latch_executions): Likewise.
31461 (gather_chrec_stats): Likewise.
31462 (final_value_replacement_loop): Likewise.
31463 (scev_const_prop): Likewise.
31464 * tree-sra.c (dump_access): Likewise.
31465 (disqualify_candidate): Likewise.
31466 (create_access): Likewise.
31467 (reject): Likewise.
31468 (maybe_add_sra_candidate): Likewise.
31469 (create_access_replacement): Likewise.
31470 (analyze_access_subtree): Likewise.
31471 (analyze_all_variable_accesses): Likewise.
31472 (sra_modify_assign): Likewise.
31473 (initialize_constant_pool_replacements): Likewise.
31474 (find_param_candidates): Likewise.
31475 (decide_one_param_reduction): Likewise.
31476 (replace_removed_params_ssa_names): Likewise.
31477 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
31478 * tree-ssa-copy.c (dump_copy_of): Likewise.
31479 (copy_prop_visit_cond_stmt): Likewise.
31480 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
31481 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
31482 (record_equivalences_from_stmt): Likewise.
31483 * tree-ssa-dse.c (compute_trims): Likewise.
31484 (delete_dead_call): Likewise.
31485 (delete_dead_assignment): Likewise.
31486 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
31487 (forward_propagate_into_cond): Likewise.
31488 (pass_forwprop::execute): Likewise.
31489 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
31490 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
31492 (move_computations_worker): Likewise.
31493 (execute_sm): Likewise.
31494 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
31495 (remove_exits_and_undefined_stmts): Likewise.
31496 (remove_redundant_iv_tests): Likewise.
31497 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
31498 (adjust_iv_update_pos): Likewise.
31499 * tree-ssa-math-opts.c (bswap_replace): Likewise.
31500 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
31501 (value_replacement): Likewise.
31502 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
31503 * tree-ssa-pre.c (print_pre_expr): Likewise.
31504 (get_representative_for): Likewise.
31505 (create_expression_by_pieces): Likewise.
31506 (insert_into_preds_of_block): Likewise.
31507 (eliminate_insert): Likewise.
31508 (eliminate_dom_walker::before_dom_children): Likewise.
31509 (eliminate): Likewise.
31510 (remove_dead_inserted_code): Likewise.
31511 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
31512 * tree-ssa-reassoc.c (get_rank): Likewise.
31513 (eliminate_duplicate_pair): Likewise.
31514 (eliminate_plus_minus_pair): Likewise.
31515 (eliminate_not_pairs): Likewise.
31516 (undistribute_ops_list): Likewise.
31517 (eliminate_redundant_comparison): Likewise.
31518 (update_range_test): Likewise.
31519 (optimize_range_tests_var_bound): Likewise.
31520 (optimize_vec_cond_expr): Likewise.
31521 (rewrite_expr_tree): Likewise.
31522 (rewrite_expr_tree_parallel): Likewise.
31523 (linearize_expr): Likewise.
31524 (break_up_subtract): Likewise.
31525 (linearize_expr_tree): Likewise.
31526 (attempt_builtin_powi): Likewise.
31527 (attempt_builtin_copysign): Likewise.
31528 (transform_stmt_to_copy): Likewise.
31529 (transform_stmt_to_multiply): Likewise.
31530 (dump_ops_vector): Likewise.
31531 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
31532 (print_scc): Likewise.
31533 (set_ssa_val_to): Likewise.
31534 (visit_reference_op_store): Likewise.
31535 (visit_use): Likewise.
31536 (sccvn_dom_walker::before_dom_children): Likewise.
31537 (run_scc_vn): Likewise.
31538 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
31540 (expr_hash_elt::print): Likewise.
31541 (const_and_copies::pop_to_marker): Likewise.
31542 (const_and_copies::record_const_or_copy_raw): Likewise.
31543 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
31544 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
31545 (dump_predicates): Likewise.
31546 (find_uninit_use): Likewise.
31547 (warn_uninitialized_phi): Likewise.
31548 (pass_late_warn_uninitialized::execute): Likewise.
31549 * tree-ssa.c (verify_vssa): Likewise.
31550 (verify_ssa): Likewise.
31551 (maybe_optimize_var): Likewise.
31552 * tree-vrp.c (dump_value_range): Likewise.
31553 (dump_all_value_ranges): Likewise.
31554 (dump_asserts_for): Likewise.
31555 (register_edge_assert_for_2): Likewise.
31556 (vrp_visit_cond_stmt): Likewise.
31557 (vrp_visit_switch_stmt): Likewise.
31558 (vrp_visit_stmt): Likewise.
31559 (vrp_visit_phi_node): Likewise.
31560 (simplify_cond_using_ranges_1): Likewise.
31561 (fold_predicate_in): Likewise.
31562 (evrp_dom_walker::before_dom_children): Likewise.
31563 (evrp_dom_walker::push_value_range): Likewise.
31564 (evrp_dom_walker::pop_value_range): Likewise.
31565 (execute_early_vrp): Likewise.
31567 2017-05-16 Richard Biener <rguenther@suse.de>
31569 * dwarf2out.c (loc_list_from_tree_1): Do not create
31570 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
31572 2017-05-16 Richard Biener <rguenther@suse.de>
31574 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
31576 (note_variable_value_in_expr): If we resolved the decl ref
31577 do not push to the stack.
31579 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
31581 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
31582 operations in fast-math mode.
31583 (vaddq_f16): Likewise.
31584 (vmul_f16): Likewise.
31585 (vmulq_f16): Likewise.
31586 (vsub_f16): Likewise.
31587 (vsubq_f16): Likewise.
31588 * config/arm/neon.md (add<mode>3): New.
31590 (fma:<VH:mode>3): New. Also remove outdated comment.
31593 2017-05-16 Martin Liska <mliska@suse.cz>
31597 * ipa-devirt.c (warn_types_mismatch): Fix typo.
31598 (odr_types_equivalent_p): Likewise.
31600 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
31602 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
31604 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
31607 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
31608 non-interunit SSE move alternatives with '?'.
31609 (zero-extendsidi peephole2): New peephole to skip intermediate
31610 general register in SSE zero-extend sequence.
31612 2017-05-15 Jeff Law <law@redhat.com>
31614 * reorg.c (relax_delay_slots): Create a new variable to hold
31615 the temporary target rather than clobbering TARGET_LABEL.
31617 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
31618 missing argument to extract_bit_field call.
31619 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
31621 2017-05-15 Martin Liska <mliska@suse.cz>
31624 * gcc.c (process_command): Do not allow empty argument of -o option.
31626 2017-05-15 Renlin Li <renlin.li@arm.com>
31628 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
31629 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
31630 * config/aarch64/constraints.md (Usf): Add long call check.
31631 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
31632 (call_value): Likewise.
31633 (sibcall): Likewise.
31634 (sibcall_value): Likewise.
31636 (call_value_insn): New.
31637 (sibcall_insn): Update rtx pattern.
31638 (sibcall_value_insn): Likewise.
31639 (call_internal): Remove.
31640 (call_value_internal): Likewise.
31641 (sibcall_internal): Likewise.
31642 (sibcall_value_internal): Likewise.
31643 (call_reg): Likewise.
31644 (call_symbol): Likewise.
31645 (call_value_reg): Likewise.
31646 (call_value_symbol): Likewise.
31648 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
31651 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
31653 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
31655 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
31656 compatible with CCGOCmode and with CCZmode.
31658 2017-05-14 Martin Sebor <msebor@redhat.com>
31660 PR middle-end/77671
31661 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
31662 (gimple_fold_builtin_snprintf): Same.
31663 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
31664 (gimple_fold_builtin_snprintf): Same.
31665 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
31666 of character types.
31667 (is_call_safe): New function.
31668 (try_substitute_return_value): Call it.
31669 (try_simplify_call): New function.
31670 (pass_sprintf_length::handle_gimple_call): Call it.
31672 2017-05-14 Martin Sebor <msebor@redhat.com>
31674 PR middle-end/80669
31675 * builtins.c (expand_builtin_stpncpy): Simplify.
31677 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
31679 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
31680 * config/i386/i386.h
31681 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
31682 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
31683 (struct machine_function): Add new members call_ms2sysv,
31684 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
31685 (struct machine_frame_state): New fields sp_realigned and
31686 sp_realigned_offset.
31687 * config/i386/i386.c
31688 (enum xlogue_stub): New enum.
31689 (enum xlogue_stub_sets): New enum.
31690 (class xlogue_layout): New class.
31691 (struct ix86_frame): New fields stack_realign_allocate_offset,
31692 stack_realign_offset and outlined_save_offset. Modify comments to
31693 detail stack layout when using out-of-line stubs.
31694 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
31695 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
31696 -mcall-ms2sysv-xlogues.
31697 (stub_managed_regs): New static variable.
31698 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
31699 registers managed by out-of-line stub.
31700 (disable_call_ms2sysv_xlogues): New function.
31701 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
31702 m->call_ms2sysv when appropriate and compute frame layout for
31704 (sp_valid_at, fp_valid_at): New inline functions.
31705 (choose_basereg): New function.
31706 (choose_baseaddr): Add align parameter, use choose_basereg and modify
31708 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
31709 Use align parameter of choose_baseaddr to generated aligned SSE movs
31711 (pro_epilogue_adjust_stack): Modify to track
31712 machine_frame_state::sp_realigned.
31713 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
31714 (ix86_nsaved_sseregs): Likewise.
31715 (ix86_emit_save_regs): Likewise.
31716 (ix86_emit_save_regs_using_mov): Likewise.
31717 (ix86_emit_save_sse_regs_using_mov): Likewise.
31718 (get_scratch_register_on_entry): Likewise.
31719 (gen_frame_set): New function.
31720 (gen_frame_load): Likewise.
31721 (gen_frame_store): Likewise.
31722 (emit_outlined_ms2sysv_save): Likewise.
31723 (emit_outlined_ms2sysv_restore): Likewise.
31724 (ix86_expand_prologue): Modify stack re-alignment code and call
31725 emit_outlined_ms2sysv_save when appropriate.
31726 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
31727 parameter rtx_insn *insn, which allows the function to be used to only
31728 generate the notes.
31729 (ix86_expand_epilogue): Modify validity checks of frame and stack
31730 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
31731 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
31732 * config/i386/predicates.md
31733 (save_multiple): New predicate.
31734 (restore_multiple): Likewise.
31735 * config/i386/sse.md
31736 (save_multiple<mode>): New pattern.
31737 (save_multiple_realign<mode>): Likewise.
31738 (restore_multiple<mode>): Likewise.
31739 (restore_multiple_and_return<mode>): Likewise.
31740 (restore_multiple_leave_return<mode>): Likewise.
31741 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
31743 2017-05-14 Julia Koval <julia.koval@intel.com>
31745 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
31746 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
31747 (__builtin_ia32_xsetbv): New builtins.
31748 * config/i386/i386.c (ix86_expand_special_args_builtin):
31750 (ix86_expand_builtin): Special expand for new intrinsics.
31751 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
31752 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
31753 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
31755 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31757 * cfganal.c (inverted_post_order_compute): Change argument type
31759 * cfganal.h (inverted_post_order_compute): Adjust prototype.
31760 * df-core.c (rest_of_handle_df_initialize): Adjust.
31761 (rest_of_handle_df_finish): Likewise.
31762 (df_analyze_1): Likewise.
31763 (df_analyze): Likewise.
31764 (loop_inverted_post_order_compute): Change argument to be a vec *.
31765 (df_analyze_loop): Adjust.
31766 (df_get_n_blocks): Likewise.
31767 (df_get_postorder): Likewise.
31768 * df.h (struct df_d): Change field to be a vec.
31769 * lcm.c (compute_laterin): Adjust.
31770 (compute_available): Likewise.
31771 * lra-lives.c (lra_create_live_ranges_1): Likewise.
31772 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
31773 * tree-ssa-pre.c (compute_antic): Likewise.
31775 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31777 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31778 (depth_first_search::depth_first_search): Change structure init
31779 function to this constructor.
31780 (depth_first_search::add_bb): Rename function to this member.
31781 (depth_first_search::execute): Likewise.
31782 (flow_dfs_compute_reverse_finish): Adjust.
31784 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31786 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31787 (longest_simple_path): Likewise.
31788 * shrink-wrap.c (spread_components): Likewise.
31789 (disqualify_problematic_components): Likewise.
31790 (emit_common_heads_for_components): Likewise.
31791 (emit_common_tails_for_components): Likewise.
31792 (insert_prologue_epilogue_for_components): Likewise.
31794 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31796 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31799 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31801 * df-core.c (df_set_blocks): Start using auto_bitmap.
31802 (df_compact_blocks): Likewise.
31803 * df-problems.c (df_rd_confluence_n): Likewise.
31804 * df-scan.c (df_insn_rescan_all): Likewise.
31805 (df_process_deferred_rescans): Likewise.
31806 (df_update_entry_block_defs): Likewise.
31807 (df_update_exit_block_uses): Likewise.
31808 (df_entry_block_bitmap_verify): Likewise.
31809 (df_exit_block_bitmap_verify): Likewise.
31810 (df_scan_verify): Likewise.
31811 * lra-constraints.c (lra_constraints): Likewise.
31812 (undo_optional_reloads): Likewise.
31813 (lra_undo_inheritance): Likewise.
31814 * lra-remat.c (calculate_gen_cands): Likewise.
31815 (do_remat): Likewise.
31816 * lra-spills.c (assign_spill_hard_regs): Likewise.
31817 (spill_pseudos): Likewise.
31818 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31819 (bitmap_set_subtract_values): Likewise.
31821 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31823 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31824 management with auto_bitmap.
31825 (fix_inter_tick): Likewise.
31826 (fix_recovery_deps): Likewise.
31827 * ira.c (add_store_equivs): Likewise.
31828 (find_moveable_pseudos): Likewise.
31829 (split_live_ranges_for_shrink_wrap): Likewise.
31830 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31831 (rtx_reuse_manager::seen_def_p): Likewise.
31832 (rtx_reuse_manager::set_seen_def): Likewise.
31833 * print-rtl.h (class rtx_reuse_manager): Likewise.
31835 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31837 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31839 (migrate_btr_def): Likewise.
31840 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31841 * df-core.c (loop_post_order_compute): Likewise.
31842 (loop_inverted_post_order_compute): Likewise.
31843 * hsa-common.h: Likewise.
31844 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31845 * init-regs.c (initialize_uninitialized_regs): Likewise.
31846 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31847 (inline_small_functions): Likewise.
31848 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31849 * ira.c (combine_and_move_insns): Likewise.
31850 (build_insn_chain): Likewise.
31851 * loop-invariant.c (find_invariants): Likewise.
31852 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31853 * predict.c (tree_predict_by_opcode): Likewise.
31854 (predict_paths_leading_to): Likewise.
31855 (predict_paths_leading_to_edge): Likewise.
31856 (estimate_loops_at_level): Likewise.
31857 (estimate_loops): Likewise.
31858 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31859 (spread_components): Likewise.
31860 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31861 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31862 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31863 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31864 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31865 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31866 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31867 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31868 (mark_threaded_blocks): Likewise.
31869 (thread_through_all_blocks): Likewise.
31870 * tree-ssa.c (verify_ssa): Likewise.
31871 (execute_update_addresses_taken): Likewise.
31872 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31874 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31876 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31878 (post_order_compute): Likewise.
31879 (inverted_post_order_compute): Likewise.
31880 (pre_and_rev_post_order_compute_fn): Likewise.
31882 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31884 * genrecog.c (int_set::int_set): Explicitly construct our
31885 auto_vec base class.
31886 * vec.h (auto_vec::auto_vec): New constructor.
31888 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31890 * bitmap.h (class auto_bitmap): New constructor taking
31891 bitmap_obstack * argument.
31893 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31895 * bitmap.h (class auto_bitmap): Change type of m_bits to
31896 bitmap_head, and adjust ctor / dtor and member operators.
31898 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31900 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31901 when returned register mode doesn't match original mode.
31903 2017-05-12 Jeff Law <law@redhat.com>
31904 Jakub Jelinek <jakub@redhat.com>
31906 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31907 we look for cc setter after the compare-elim changes.
31908 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31909 within the vector to match what compare-elim now expects.
31910 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31911 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31913 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31914 after the compare-elim changes.
31915 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31916 the vector to match what compare-elim now expects.
31917 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31918 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31919 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31920 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31921 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31923 * config/visium/visium.c (single_set_and_flags): Fix where
31924 we look for cc setter after the compare-elim changes.
31925 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31926 with the vector to match what compare-elim now expects.
31927 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31928 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31929 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31930 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31931 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31932 (neg<mode>2_insn_set_overflow): Likewise.
31934 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31936 PR middle-end/79794
31937 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31938 maybe_expand_insn call, set ops[0].target. If still set after call,
31939 set alt_rtl. Add extra arg to recursive calls.
31940 (extract_bit_field): Add alt_rtl argument. Pass to
31942 * expmed.h (extract_bit_field): Fix prototype.
31943 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31944 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31945 to extract_bit_field_calls.
31946 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31947 Pass alt_rtl to extract_bit_field calls.
31948 * calls.c (store_unaligned_arguments_into_psuedos)
31949 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31950 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31952 * optabs.h (struct expand_operand): Add target bitfield.
31954 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31956 * compare-elim.c (try_eliminate_compare): Canonicalize
31957 operation with embedded compare to
31958 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31959 (set (reg) (operation)].
31961 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31963 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31966 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31967 cost of adding a carry flag for ADC instruction.
31968 [case MINUS]: Ignore the cost of subtracting a carry flag
31969 for SBB instruction.
31971 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31973 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31975 * config/rs6000/bmiintrin.h: New file.
31976 * config/rs6000/bmi2intrin.h: New file.
31977 * config/rs6000/x86intrin.h: New file.
31979 2017-05-12 Jeff Law <law@redhat.com>
31981 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31984 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31986 PR middle-end/80707
31987 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31989 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31991 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31992 early expansion of vector divide builtins.
31993 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31994 builtins identified as having unsigned arguments.
31996 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31998 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31999 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
32000 expansion of vector logical operations (and, andc, or, xor,
32003 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
32005 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
32006 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
32008 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
32010 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
32011 early GIMPLE expansion of vector multiplies.
32013 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32015 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
32016 TARGET_HAVE_MOVT conditional.
32017 (movt splitter): Likewise.
32019 2017-05-12 Richard Biener <rguenther@suse.de>
32021 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
32022 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
32023 Fold all stmts not inplace.
32025 2017-05-12 Richard Biener <rguenther@suse.de>
32027 PR tree-optimization/80713
32028 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
32029 inserted_exprs bit for not removed stmts.
32031 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
32033 PR middle-end/69921
32034 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
32035 parallelized" attribute for parallelized OpenACC kernels.
32036 * omp-offload.c (execute_oacc_device_lower): Use it.
32038 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
32039 Set "oacc kernels" attribute.
32040 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
32041 parameter. Adjust all users.
32042 (oacc_fn_attrib_kernels_p): Remove function.
32043 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
32044 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
32045 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
32046 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
32047 assert "oacc kernels" attribute is set.
32049 2017-05-11 Carl Love <cel@us.ibm.com>
32051 * config/rs6000/rs6000-c: Add support for built-in functions
32052 vector unsigned char vec_popcnt (vector signed char)
32053 vector unsigned char vec_popcnt (vector unsigned char)
32054 vector unsigned short vec_popcnt (vector signed short)
32055 vector unsigned short vec_popcnt (vector unsigned short)
32056 vector unsigned int vec_popcnt (vector signed int)
32057 vector unsigned int vec_popcnt (vector unsigned int)
32058 vector unsigned long long vec_popcnt (vector signed long long)
32059 vector unsigned long long vec_popcnt (vector unsigned long long)
32060 vector signed long long vec_slo (vector signed long long,
32061 vector signed char)
32062 vector signed long long vec_slo (vector signed long long,
32063 vector unsigned char)
32064 vector unsigned long long vec_slo (vector unsigned long long,
32065 vector signed char)
32066 vector unsigned long long vec_slo (vector unsigned long long,
32067 vector unsigned char)
32068 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
32069 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
32070 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
32071 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
32072 * doc/extend.texi: Update the built-in documentation file for the
32073 new built-in functions.
32075 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
32077 * attribs.h (sorted_attr_string): Move machine independent
32078 functions for target clone support from the i386 port to common
32079 code. Rename ix86_function_versions to common_function_versions.
32080 Rename make_name to make_unique_name.
32081 (common_function_versions): Likewise.
32082 (make_unique_name): Likewise.
32083 (make_dispatcher_decl): Likewise.
32084 (is_function_default_version): Likewise.
32085 * attribs.c (attr_strcmp): Likewise.
32086 (sorted_attr_string): Likewise.
32087 (common_function_versions): Likewise.
32088 (make_unique_name): Likewise.
32089 (make_dispatcher_decl): Likewise.
32090 (is_function_default_version): Likewise.
32091 * config/i386/i386.c (attr_strcmp): Likewise.
32092 (sorted_attr_string): Likewise.
32093 (ix86_function_versions): Likewise.
32094 (make_name): Likewise.
32095 (make_dispatcher_decl): Likewise.
32096 (is_function_default_version): Likewise.
32097 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
32099 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32102 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
32103 Account for direct move costs for vec_construct of integer
32106 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
32109 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
32110 (UNSPEC_STX_ATOMIC): Ditto.
32111 (loaddi_via_sse): New insn.
32112 (storedi_via_sse): Ditto.
32113 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
32114 Update corresponding peephole2 patterns.
32115 (atomic_storedi_fpu): Ditto.
32117 2017-05-11 Julia Koval <julia.koval@intel.com>
32119 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
32120 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
32122 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
32123 (__builtin_ia32_rsqrt14ss_mask): New builtins.
32124 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
32126 2017-05-11 Nathan Sidwell <nathan@acm.org>
32128 * graphite-poly.c: Include dumpfile.h.
32130 * dumpfle.h (dump_function): Declare here ...
32131 * tree-dump.h (dump_function): ... not here.
32132 * dumpfile.c: #include tree-cfg.h.
32133 (dump_function): Move here from ...
32134 * tree-dump.c (dump_function): ... here.
32135 * gimplify.c: #include splay-tree.h, not tree-dump.h.
32136 * graphite-poly.c: Don't include tree-dump.h.
32137 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
32138 * print-tree.c: Likewise.
32139 * stor-layout.c: Likewise.
32140 * tree-nested.c: Likewise.
32142 * dumpfile.c (dump_start): Use TDF_FLAGS.
32143 (dump_enable_all): Fix TDF_KIND check thinko.
32145 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
32147 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32148 array entries to represent two legal parameterizations of the
32149 overloaded __builtin_cmpb function, as represented by the
32150 P6_OV_BUILTIN_CMPB constant.
32151 (altivec_resolve_overloaded_builtin): Add special case handling
32152 for the __builtin_cmpb function, as represented by the
32153 P6_OV_BUILTIN_CMPB constant.
32154 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
32155 (BU_P6_64BIT_2): New macro.
32156 (BU_P6_OVERLOAD_2): New macro
32157 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
32158 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
32159 (CMPB): Add overload support to represent both 32-bit and 64-bit
32160 compare-bytes function.
32161 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
32162 support for TARGET_CMPB.
32163 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
32164 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32165 documentation of the __builtin_cmpb overloaded built-in function.
32167 2017-05-11 Richard Biener <rguenther@suse.de>
32169 PR tree-optimization/80705
32170 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
32171 bases are not vectorizable.
32173 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32175 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
32176 when counting register pressure.
32178 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32180 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
32181 (iv_ca_more_deps): Renamed to ...
32182 (iv_ca_compare_deps): ... this.
32183 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
32185 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32187 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
32189 (determine_group_iv_costs): ... here.
32190 (find_inv_vars_cb): Record inv var if it's not recorded before.
32192 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32194 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
32195 (get_shiftadd_cost): Ditto.
32197 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32199 * tree-ssa-address.c: Include header file.
32200 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
32202 (add_to_parts): Refactor.
32203 (addr_to_parts): New parameter. Update use of move_hint_to_base.
32204 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
32207 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32209 PR tree-optimization/53090
32210 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
32212 (extract_cond_operands): Detect condition with IV on both sides
32213 and return COMP_IV_EXPR_2.
32214 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
32215 (rewrite_use_compare): Simplify by removing call to function
32216 extract_cond_operands.
32218 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32220 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
32221 (extract_cond_operands): Detect condition comparing against non-
32222 invariant bound and return appropriate enum value.
32223 (find_interesting_uses_cond): Update use of extract_cond_operands.
32224 Handle its return value accordingly.
32225 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
32227 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32229 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
32230 nonlinear iv_use computation in loop invariant sensitive way.
32232 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32234 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
32235 (find_iv_candidates): Call relate_compare_use_with_all_cands.
32237 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32239 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
32240 (dump_cand): Support iv_cand.inv_exprs.
32241 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
32243 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
32246 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32248 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
32250 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
32251 as local function. Include necessary header files.
32252 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
32254 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32256 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
32258 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32260 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
32261 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
32262 RSHIFT_EXPR and BIT_NOT_EXPR.
32264 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32266 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
32267 (adjust_setup_cost): New parameter supporting round up adjustment.
32268 (struct address_cost_data): Delete.
32269 (force_expr_to_var_cost): Don't bound cost with spill_cost.
32270 (split_address_cost, ptr_difference_cost): Delete.
32271 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
32272 (struct ainc_cost_data): New struct.
32273 (get_address_cost_ainc): New function.
32274 (get_address_cost, get_computation_cost): Reimplement.
32275 (determine_group_iv_cost_address): Record inv_expr for all uses of
32277 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
32278 (iv_ca_has_deps): Reimplemented to ...
32279 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
32281 (iv_ca_extend): Call iv_ca_more_deps.
32283 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32285 * tree-ssa-address.c (struct mem_address): Move to header file.
32286 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
32287 * tree-ssa-address.h (struct mem_address): Move from C file.
32288 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
32290 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32292 * tree-affine.h (aff_combination_type): New interface.
32293 (aff_combination_zero_p): Remove static.
32294 (aff_combination_const_p): New interface.
32295 (aff_combination_singleton_var_p): New interfaces.
32297 2017-05-11 Richard Biener <rguenther@suse.de>
32299 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
32300 Skip unreachable blocks and destinations.
32301 (eliminate): Move stmt removal and fixup ...
32302 (fini_eliminate): ... here. Skip inserted exprs.
32303 (pass_pre::execute): Move fini_pre after fini_eliminate.
32304 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
32305 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
32306 PRE to get rid of dead code that has invalid SSA form and
32307 split critical edges again.
32309 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32311 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32313 2017-05-11 Richard Biener <rguenther@suse.de>
32315 * passes.c (execute_function_todo): Verify loops if they are
32316 said to be up-to-date.
32317 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
32318 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
32320 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
32323 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
32324 handle calling assemble_external ourself.
32327 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
32328 modes with zero size. Enhance comment.
32330 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32332 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
32333 built-ins for vec_xl and vec_xst with short and char pointer
32336 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32338 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
32339 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
32340 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
32341 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
32342 (_mm_maskz_min_round_ss): New intrinsics.
32343 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
32344 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
32345 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
32346 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
32347 (__builtin_ia32_minss_mask_round): New builtins.
32348 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32349 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32350 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
32352 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
32353 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
32355 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
32358 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32360 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
32361 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
32362 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
32363 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
32364 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
32365 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
32366 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
32367 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32368 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32369 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
32370 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
32371 (__builtin_ia32_mulss_mask_round): New builtins.
32372 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32373 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32374 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
32376 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
32377 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32379 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32382 2017-05-10 Julia Koval <julia.koval@intel.com>
32384 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
32385 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
32386 (_mm256_setr_m128i): New intrinsics.
32388 2017-05-10 Julia Koval <julia.koval@intel.com>
32390 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
32391 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
32392 (_mm_maskz_rcp14_ss): New intrinsics.
32393 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
32394 (__builtin_ia32_rcp14ss_mask): New builtins.
32395 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
32397 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
32399 PR tree-optimization/51513
32400 * tree-cfg.c (gimple_seq_unreachable_p): New function.
32401 (assert_unreachable_fallthru_edge_p): Use it.
32402 (group_case_labels_stmt): Likewise.
32403 * tree-cfg.h: Prototype it.
32404 * stmt.c: Include cfghooks.h and tree-cfg.h.
32405 (emit_case_dispatch_table) <gap_label>: New local variable.
32406 Use it to fill dispatch table gaps.
32407 Test for default_label before updating probabilities.
32408 (expand_case) <default_label>: Remove unneeded initialization.
32409 Test for unreachable default case statement and remove its edge.
32410 Set default_label accordingly.
32411 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
32413 2017-05-10 Carl Love <cel@us.ibm.com>
32415 * config/rs6000/rs6000-c: Add support for built-in functions
32416 vector signed char vec_neg (vector signed char)
32417 vector signed short int vec_neg (vector short int)
32418 vector signed int vec_neg (vector signed int)
32419 vector signed long long vec_neg (vector signed long long)
32420 vector float vec_neg (vector float)
32421 vector double vec_neg (vector double)
32422 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
32424 * config/rs6000/altivec.h: Add define for vec_neg
32425 * doc/extend.texi: Update the built-in documentation for the
32426 new built-in functions.
32428 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32430 PR tree-optimization/77644
32431 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
32433 2017-05-10 Nathan Sidwell <nathan@acm.org>
32435 * dumpfile.h (TDI_lang_all): New.
32436 (TDF_KIND): New. Renumber others
32437 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
32439 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
32441 (get_dump_file_name): Adjust suffix generation.
32442 (dump_enable_all): Use TDF_KIND.
32443 * doc/invoke.texi (-fdump-lang-all): Document.
32445 * dumpfile.h: Tabify.
32447 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
32450 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
32451 Move member access before delete.
32453 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
32455 * tree-inline.c (expand_call_inline): Split block at stmt
32458 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
32461 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
32462 are now unused after splitting mov{sf,sd}_hardfloat.
32463 (f32_lr2): Likewise.
32464 (f32_lm): Likewise.
32465 (f32_lm2): Likewise.
32466 (f32_li): Likewise.
32467 (f32_li2): Likewise.
32468 (f32_lv): Likewise.
32469 (f32_sr): Likewise.
32470 (f32_sr2): Likewise.
32471 (f32_sm): Likewise.
32472 (f32_sm2): Likewise.
32473 (f32_si): Likewise.
32474 (f32_si2): Likewise.
32475 (f32_sv): Likewise.
32476 (f32_dm): Likewise.
32477 (f32_vsx): Likewise.
32478 (f32_av): Likewise.
32479 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
32480 For movsf, order stores so the VSX stores occur before the GPR
32481 store which encourages the register allocator to use a traditional
32482 FPR instead of a GPR. For movsd, order the stores so that the GPR
32483 store comes before the VSX stores to allow the power6 to work.
32484 This is due to the power6 not having a 32-bit integer store
32485 instruction from a FPR.
32486 (movsf_hardfloat): Likewise.
32487 (movsd_hardfloat): Likewise.
32489 2017-05-09 Martin Sebor <msebor@redhat.com>
32491 PR translation/80280
32492 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
32495 PR translation/80280
32496 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
32497 data member added in r247778.
32498 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
32500 2017-05-09 Nathan Sidwell <nathan@acm.org>
32502 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
32504 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
32507 2017-05-09 Marek Polacek <polacek@redhat.com>
32509 * doc/invoke.texi: Fix typo.
32511 2017-05-09 Richard Biener <rguenther@suse.de>
32513 * tree-vrp.c (vrp_val_is_max): Adjust comment.
32514 (vrp_val_is_min): Likewise.
32515 (set_value_range_to_value): Likewise.
32516 (set_value_range_to_nonnegative): Likewise.
32517 (gimple_assign_nonzero_p): Likewise.
32518 (gimple_stmt_nonzero_p): Likewise.
32519 (vrp_int_const_binop): Likewise. Remove unreachable case.
32520 (adjust_range_with_scev): Adjust comments.
32521 (compare_range_with_value): Likewise.
32522 (extract_range_from_phi_node): Likewise.
32523 (test_for_singularity): Likewise.
32525 2017-05-09 Richard Biener <rguenther@suse.de>
32527 * tree-vrp.c (get_single_symbol): Add assert that we don't
32528 get overflowed constants as invariant part.
32529 (compare_values_warnv): Add comment before the TREE_NO_WARNING
32530 checks. Use wi::cmp instead of recursing for integer constants.
32531 (compare_values): Just ignore whether we assumed undefined
32532 overflow instead of failing the compare.
32533 (extract_range_for_var_from_comparison_expr): Add comment before the
32534 TREE_NO_WARNING sets.
32535 (test_for_singularity): Likewise.
32536 (extract_range_from_comparison): Do not disable optimization
32537 when we assumed undefined overflow.
32538 (extract_range_basic): Remove init of unused var.
32540 2017-05-09 Richard Biener <rguenther@suse.de>
32542 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
32543 (extract_range_from_multiplicative_op_1): Adjust.
32544 (extract_range_from_binary_expr_1): Use int_const_binop.
32546 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
32549 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
32550 rs6000_store_data_bypass_p in seven define_bypass directives and
32551 in several comments.
32552 * config/rs6000/rs6000-protos.h: Add prototype for
32553 rs6000_store_data_bypass_p function.
32554 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
32555 function implements slightly different (rs6000-specific) semantics
32556 than store_data_bypass_p, returning false rather than aborting
32557 with assertion error when arguments do not satisfy the
32558 requirements of store data bypass.
32559 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
32560 rs6000_store_data_bypass_p.
32562 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
32564 * config/xtensa/xtensa-protos.h
32565 (xtensa_initial_elimination_offset): New declaration.
32566 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
32567 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
32568 macro definition, add case for FRAME_POINTER_REGNUM when
32569 FRAME_GROWS_DOWNWARD.
32570 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
32571 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
32572 xtensa_initial_elimination_offset.
32574 2017-05-08 Nathan Sidwell <nathan@acm.org>
32576 * doc/invoke.texi: Alphabetize -fdump options.
32578 2017-05-08 Martin Sebor <msebor@redhat.com>
32580 PR translation/80280
32581 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
32583 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
32585 * target.def (compute_frame_layout): New optional target hook.
32586 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
32587 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
32588 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
32590 * reload1.c (verify_initial_elim_offsets): Likewise.
32591 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
32592 (use_simple_return_p): Call arm_compute_frame_layout if needed.
32593 (arm_get_frame_offsets): Split up into this ...
32594 (arm_compute_frame_layout): ... and this function.
32596 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
32598 * config/aarch64/constraints.md (Usa): New constraint.
32599 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
32601 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
32603 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
32604 with_multilib_list after it has been checked.
32606 2017-05-08 Richard Biener <rguenther@suse.de>
32608 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
32609 (bitmap_set_subtract_values): Likewise.
32611 2017-05-08 Richard Biener <rguenther@suse.de>
32613 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
32614 (gimple_assign_nonzero): ... this and remove strict_overflow_p
32616 (gimple_stmt_nonzero_warnv_p): Rename to ...
32617 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
32619 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
32620 (extract_range_basic): Adjust, do not disable propagation on
32621 strict overflow sensitive simplification.
32622 (vrp_visit_cond_stmt): Likewise.
32624 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
32626 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
32627 body size unconditionally.
32629 2017-05-07 Jeff Law <law@redhat.com>
32632 2017-05-06 Jeff Law <law@redhat.com>
32633 PR tree-optimization/78496
32634 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32637 PR tree-optimization/78496
32638 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32639 (simplify_stmt_using_ranges): Call it.
32640 (vrp_dom_walker::before_dom_children): Extract equivalences
32641 from an ASSERT_EXPR with an equality comparison against a
32644 2017-05-06 Jeff Law <law@redhat.com>
32646 PR tree-optimization/78496
32647 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32650 PR tree-optimization/78496
32651 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32652 (simplify_stmt_using_ranges): Call it.
32653 (vrp_dom_walker::before_dom_children): Extract equivalences
32654 from an ASSERT_EXPR with an equality comparison against a
32657 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32659 * lra-constraints.c (lra_copy_reg_equiv): New function.
32660 (split_reg): Use it to copy equivalence information from the
32661 original register to the spill register.
32663 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32665 PR rtl-optimization/75964
32666 * simplify-rtx.c (simplify_const_relational_operation): Remove
32667 invalid handling of comparisons of integer ABS.
32669 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
32671 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
32672 initialize to zero.
32673 (init_regs): Remove declaration.
32674 (function_arg_advance_32): Initialize error_p as boolean variable.
32676 2017-05-05 Nathan Sidwell <nathan@acm.org>
32678 * store-motion.c (remove_reachable_equiv_notes): Reformat long
32679 lines. Use for (;;).
32681 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32683 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
32684 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
32685 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
32686 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
32687 VF=2 that require versioning.
32689 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32691 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
32694 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32696 * diagnostic.h (diagnostic_override_option_index): Convert from
32697 macro to inline function.
32699 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32701 * diagnostic.c (last_module_changed_p): New function.
32702 (set_last_module): New function.
32703 (diagnostic_report_current_module): Convert macro usage to
32704 the above functions.
32705 * diagnostic.h (diagnostic_context::last_module): Strengthen
32706 from const line_map * to const line_map_ordinary *.
32707 (diagnostic_last_module_changed): Delete macro.
32708 (diagnostic_set_last_module): Delete macro.
32710 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32712 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
32713 with diagnostic_report_diagnostic.
32714 (diagnostic_n_impl_richloc): Likewise.
32715 * diagnostic.h (report_diagnostic): Delete macro.
32716 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
32717 with diagnostic_report_diagnostic.
32718 * substring-locations.c (format_warning_va): Likewise.
32720 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32722 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
32723 save/restor of format_spec. Move option-printing code to...
32724 (print_option_information): ...this new function, and
32725 reimplement by simply printing to the pretty_printer,
32726 rather than appending to the format string.
32728 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32730 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
32731 handling logic into...
32732 (update_effective_level_from_pragmas): ...this new function.
32734 2017-05-04 Andrew Waterman <andrew@sifive.com>
32736 * config/riscv/riscv.opt (mstrict-align): New option.
32737 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
32738 (SLOW_UNALIGNED_ACCESS): Define.
32739 (riscv_slow_unaligned_access): Declare.
32740 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
32742 (riscv_slow_unaligned_access): New variable.
32743 (rocket_tune_info): Set slow_unaligned_access to true.
32744 (optimize_size_tune_info): Set slow_unaligned_access to false.
32745 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
32746 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
32747 (riscv_option_override): Set riscv_slow_unaligned_access.
32748 * doc/invoke.texi: Add -mstrict-align to RISC-V.
32750 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
32752 * config/riscv/riscv.md: Unify indentation.
32754 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
32759 * config/rs6000/rs6000.md (u code attribute): Add FIX and
32761 (extendsi<mode>2): Add support for doing sign extension via
32762 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
32763 don't have ISA 3.0 instructions.
32764 (extendsi<mode>2 splitter): Likewise.
32765 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
32766 generate the normal insns since SImode can now go in vector
32767 registers. Disallow the special UNSPECs needed for previous
32768 machines to hide SImode being used. Add new insns
32769 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
32770 (fix_trunc<mode>si2_stfiwx): Likewise.
32771 (fix_trunc<mode>si2_internal): Likewise.
32772 (fixuns_trunc<mode>si2): Likewise.
32773 (fixuns_trunc<mode>si2_stfiwx): Likewise.
32774 (fctiw<u>z_<mode>_smallint): Likewise.
32775 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32776 of floating point to 32-bit integer from doing a direct move to
32777 the GPR registers to do a store.
32778 (fctiwz_<mode>): Break long line.
32780 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32782 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32783 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32784 (addr_list, addr_offset_valid_p): New.
32785 (split_address_groups): Check offset validity with above function.
32786 (gt-tree-ssa-loop-ivopts.h): Include header file.
32788 2017-05-05 Nathan Sidwell <nathan@acm.org>
32790 * config.gcc (arm*-*-*): Add missing 'fi'.
32792 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32794 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32795 included in -fopt-info does not matter.
32796 * doc/optinfo.texi (-fopt-info): Fix description of default
32797 behavour. Explicitly say order of options included in -fopt-info
32800 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32802 * config.gcc: Allow combinations of aprofile and rmprofile values for
32803 --with-multilib-list.
32804 * config/arm/t-multilib: New file.
32805 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32806 variables. Remove setting of ISA and floating-point ABI in
32807 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32808 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32809 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32810 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32812 * config/arm/t-rmprofile: Likewise except for the matches changes.
32813 * doc/install.texi (--with-multilib-list): Document the combination of
32814 aprofile and rmprofile values and warn about pitfalls in doing that.
32816 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32818 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32819 (movdi_aarch64): Likewise.
32821 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32823 PR tree-optimization/80632
32824 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32826 (build_arrays): Initialize it for virtual phis.
32827 (fix_phi_nodes): Use it for virtual phis.
32829 PR tree-optimization/80558
32830 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32831 [x, y] op z into [x op, y op z] for op & or | if conditions
32834 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32835 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32838 * config/arm/arm.md (use_literal_pool): Remove.
32839 (64-bit immediate split): No longer takes cost into consideration
32840 if arm_disable_literal_pool is enabled.
32841 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32842 used when arm_disable_literal_pool is enabled.
32843 (arm_max_const_double_inline_cost): Remove use of
32844 arm_disable_literal_pool.
32845 (push_minipool_fix): Add assert.
32846 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32847 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32848 (no_literal_pool_sf_immediate): New.
32850 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32852 PR tree-optimization/80613
32853 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32854 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32856 2017-05-05 Richard Biener <rguenther@suse.de>
32858 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32860 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32862 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32863 of this flag from insn conditions due to removal from r247495.
32865 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32867 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32869 (arm_early_store_addr_dep_ptr): Likewise.
32870 * config/arm/aarch-common-protos.h
32871 (arm_early_load_addr_dep_ptr): Add prototype.
32872 (arm_early_store_addr_dep_ptr): Likewise.
32873 * config/arm/cortex-a53.md: Add new bypasses.
32875 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32877 * tree.c (next_type_uid): Change type to unsigned.
32878 (type_hash_canon): Decrement back next_type_uid if
32879 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32880 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32883 2017-05-04 Martin Sebor <msebor@redhat.com>
32885 * builtins.c: Fix a trivial typo in a comment.
32887 PR middle-end/79234
32888 * builtins.c (check_sizes): Adjust to handle reading past the end.
32889 Avoid printing excessive upper bound of ranges. Use %E to print
32890 tree nodes instead of converting them to %wu.
32891 (expand_builtin_memchr): New function.
32892 (compute_dest_size): Rename...
32893 (compute_objsize): ...to this.
32894 (expand_builtin_memcpy): Adjust.
32895 (expand_builtin_mempcpy): Adjust.
32896 (expand_builtin_strcat): Adjust.
32897 (expand_builtin_strcpy): Adjust.
32898 (check_strncat_sizes): Adjust.
32899 (expand_builtin_strncat): Adjust.
32900 (expand_builtin_strncpy): Adjust and simplify.
32901 (expand_builtin_memset): Adjust.
32902 (expand_builtin_bzero): Adjust.
32903 (expand_builtin_memcmp): Adjust.
32904 (expand_builtin): Handle memcmp.
32905 (maybe_emit_chk_warning): Check strncat just once.
32907 2017-05-04 Martin Sebor <msebor@redhat.com>
32909 PR preprocessor/79214
32910 PR middle-end/79222
32911 PR middle-end/79223
32912 * builtins.c (check_sizes): Add inlining context and issue
32913 warnings even when -Wno-system-headers is set.
32914 (check_strncat_sizes): Same.
32915 (expand_builtin_strncat): Same.
32916 (expand_builtin_memmove): New function.
32917 (expand_builtin_stpncpy): Same.
32918 (expand_builtin): Handle memmove and stpncpy.
32920 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32922 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32923 which is not used any more.
32925 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32927 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32929 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32931 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32932 (cortexa53_tunings): Likewise.
32933 (cortexa57_tunings): Likewise.
32934 (cortexa72_tunings): Likewise.
32935 (cortexa73_tunings): Likewise.
32937 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32939 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32940 Set loop alignment to 8.
32942 2017-05-04 Martin Sebor <msebor@redhat.com>
32944 PR translation/80280
32945 * builtins.c (expand_builtin_object_size): Add missing quoting to
32946 %D and like directives.
32947 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32948 (hsa_type_for_tree_type): Same.
32949 (verify_function_arguments): Same.
32950 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32951 * varasm.c (get_section): Same.
32954 2017-05-04 Martin Sebor <msebor@redhat.com>
32956 PR translation/80280
32957 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32959 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32961 * config/aarch64/aarch64.c (generic_addrcost_table):
32962 Change HI/TI mode setting.
32964 2017-05-04 Martin Jambor <mjambor@suse.cz>
32966 PR tree-optimization/80622
32967 * tree-sra.c (comes_initialized_p): New function.
32968 (build_accesses_from_assign): Only set write lazily when
32969 comes_initialized_p is false.
32970 (analyze_access_subtree): Use comes_initialized_p.
32971 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32972 instead of testing for PARM_DECL.
32974 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32976 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32977 constraint on operand 0 to allow more general addressing modes.
32978 Adjust output template.
32979 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32981 * config/aarch64/aarch64-protos.h
32982 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32983 * config/aarch64/constraints.md (Dp): New address constraint.
32984 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32987 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32989 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32990 update use of estimate_ipcp_clone_size_and_time.
32991 (estimate_local_effects): Update use of
32992 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32993 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32994 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32995 Return nonspecialized time.
32997 2017-05-04 Richard Biener <rguenther@suse.de>
32999 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
33000 for the last VUSE which def dominates the PHI. Directly call
33002 (get_continuation_for_phi_1): Remove.
33004 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
33006 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
33007 to explain the use of truncating division. Cap the number of
33008 iterations to the maximum given by nb_iterations_upper_bound,
33011 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33013 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
33014 * configure: Regenerate.
33015 * config.in: Regenerate.
33016 * config/i386/driver-mingw32.c: new file.
33017 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
33018 * config.host: Link driver-mingw32.o on MinGW host.
33019 * doc/install.texi: Document new --enable-mingw-wildcard configure
33022 2017-05-04 Marek Polacek <polacek@redhat.com>
33024 PR tree-optimization/80612
33025 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
33027 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
33028 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
33030 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
33031 (movt splitter): Likewise.
33032 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
33033 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
33034 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
33035 block for Thumb-1 with MOVT.
33036 (thumb2_legitimate_address_p): Move code block ...
33037 (can_avoid_literal_pool_for_label_p): ... into this new function.
33038 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
33040 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
33041 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
33042 "M-profile targets with the MOVT instruction".
33044 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
33046 * config/arm/arm-builtins.c (arm_init_builtins): Rename
33047 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
33048 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
33050 2017-05-04 Martin Liska <mliska@suse.cz>
33052 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
33053 variable cond_code.
33055 2017-05-04 Richard Biener <rguenther@suse.de>
33057 * tree.c (array_at_struct_end_p): Handle arrays at struct
33058 end with flexarrays more conservatively. Refactor and treat
33059 arrays of arrays or aggregates more strict. Fix
33060 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
33061 * tree.c (array_at_struct_end_p): Adjust prototype.
33062 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
33063 * gimple-fold.c (get_range_strlen): Likewise.
33064 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
33066 2017-05-04 Richard Biener <rguenther@suse.de>
33068 PR tree-optimization/31130
33069 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
33071 (supports_overflow_infinity): Likewise.
33072 (is_negative_overflow_infinity): Likewise.
33073 (is_positive_overflow_infinity): Likewise.
33074 (is_overflow_infinity): Likewise.
33075 (stmt_overflow_infinity): Likewise.
33076 (overflow_infinity_range_p): Likewise.
33077 (usable_range_p): Remove as always returning true.
33078 (make_overflow_infinity): Remove.
33079 (negative_overflow_infinity): Likewise.
33080 (positive_overflow_infinity): Likewise.
33081 (avoid_overflow_infinity): Likewise.
33082 (set_value_range): Adjust accordingly.
33083 (set_value_range_to_nonnegative): Likewise, remove now unused
33084 overflow_infinity arg.
33085 (vrp_operand_equal_p): Adjust.
33086 (update_value_range): Likewise.
33087 (range_int_cst_singleton_p): Likewise.
33088 (operand_less_p): Likewise.
33089 (compare_values_warnv): Likewise.
33090 (extract_range_for_var_from_comparison_expr): Likewise.
33091 (vrp_int_const_binop): Likewise.
33092 (zero_nonzero_bits_from_vr): Likewise.
33093 (extract_range_from_multiplicative_op_1): Likewise.
33094 (extract_range_from_binary_expr_1): Likewise.
33095 (extract_range_from_unary_expr): Likewise.
33096 (extract_range_from_comparison): Likewise.
33097 (extract_range_basic): Likewise.
33098 (adjust_range_with_scev): Likewise.
33099 (compare_ranges): Likewise.
33100 (compare_range_with_value): Likewise.
33101 (dump_value_range): Likewise.
33102 (test_for_singularity): Likewise, remove strict_overflow_p parameter
33104 (simplify_cond_using_ranges): Adjust.
33106 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
33108 * brig-builtins.def: Added a builtin for class_f64.
33109 * builtin-types.def: Added a builtin type needed by class_f64.
33111 2017-05-03 Jason Merrill <jason@redhat.com>
33113 * timevar.def: Add TV_CONSTEXPR.
33115 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33117 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
33119 2017-05-03 Martin Jambor <mjambor@suse.cz>
33121 * ipa-prop.c (ipa_update_after_lto_read): Removed.
33122 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
33123 * ipa-cp.c (ipcp_propagate_stage): Do not call
33124 ipa_update_after_lto_read.
33125 * ipa-inline.c (ipa_inline): Likewise.
33127 2017-05-03 Martin Jambor <mjambor@suse.cz>
33129 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
33130 tag. Added a default constructor and a destructor.
33131 (ipa_edge_args_sum_t): New class;
33132 (ipa_edge_args_sum): Declare.
33133 (ipa_edge_args_vector): Remove declaration.
33134 (IPA_EDGE_REF): Use ipa_edge_args_sum.
33135 (ipa_free_edge_args_substructures): Remove declaration.
33136 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
33137 (ipa_edge_args_info_available_for_edge_p): Likewise.
33138 * ipa-prop.c (ipa_edge_args_vector): Removed.
33139 (edge_removal_hook_holder): Likewise.
33140 (edge_duplication_hook_holder): Likewise.
33141 (ipa_edge_args_sum): New variable.
33142 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
33143 ipa_edge_args_vector.
33144 (ipa_free_edge_args_substructures): Likewise.
33145 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
33146 ipa_edge_args_vector.
33147 (ipa_edge_removal_hook): Turned into method
33148 ipa_edge_args_sum_t::remove.
33149 (ipa_edge_duplication_hook): Turned into method
33150 ipa_edge_args_sum_t::duplicate.
33151 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
33152 registering edge hooks.
33153 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
33154 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
33155 ipa_edge_args_sum instead of ipa_edge_args_vector.
33156 * ipa-profile.c (ipa_profile): Likewise.
33158 2017-05-03 Martin Jambor <mjambor@suse.cz>
33160 * symbol-summary.h (function_summary): New method exists.
33161 (function_summary::symtab_removal): Deallocate through release.
33162 (call_summary): New class.
33163 (gt_ggc_mx): New overload.
33164 (gt_pch_nx): Likewise.
33165 (gt_pch_nx): Likewise.
33167 2017-05-03 Jeff Law <law@redhat.com>
33169 PR tree-optimization/78496
33170 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
33171 from simplify_cond_using_ranges. Split off code to walk
33172 backwards through casts into ...
33173 (simplify_cond_using_ranges_2): New function.
33174 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
33175 (execute_vrp): After identifying jump threads, call
33176 simplify_cond_using_ranges_2.
33178 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
33181 * ipa-inline.h (inline_summary): Add ctor.
33182 (create_ggc): Do not use ggc_cleared_alloc.
33184 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
33185 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33187 * gcc.c (handle_braces): Support escaping in switch matching
33189 * doc/invoke.texi (Spec Files): Document it.
33190 Remove superfluous @code markup in items.
33192 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33194 * diagnostic-show-locus.c (struct column_range): New struct.
33195 (get_affected_columns): New function.
33196 (get_printed_columns): New function.
33197 (struct correction): New struct.
33198 (correction::ensure_capacity): New function.
33199 (correction::ensure_terminated): New function.
33200 (struct line_corrections): New struct.
33201 (line_corrections::~line_corrections): New dtor.
33202 (line_corrections::add_hint): New function.
33203 (layout::print_trailing_fixits): Reimplement in terms of the new
33205 (selftest::test_overlapped_fixit_printing): New function.
33206 (selftest::diagnostic_show_locus_c_tests): Call it.
33208 2017-05-03 Nathan Sidwell <nathan@acm.org>
33210 Canonicalize canonical type hashing
33211 * tree.h (type_hash_canon_hash): Declare.
33212 * tree.c (type_hash_list, attribute_hash_list): Move into
33213 type_hash_canon_hash.
33214 (build_type_attribute_qual_variant): Break out hash code calc into
33215 type_hash_canon_hash.
33216 (type_hash_canon_hash): New. Generic type hash computation.
33217 (build_range_type_1, build_array_type_1, build_function_type,
33218 build_method_type_directly, build_offset_type, build_complex_type,
33219 make_vector_type): Call it.
33221 2017-05-03 Richard Biener <rguenther@suse.de>
33223 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
33224 When all DRs have unknown misaligned do not always peel
33225 when there is a store but apply the same costing model as if
33226 there were only loads.
33228 2017-05-03 Richard Biener <rguenther@suse.de>
33231 PR tree-optimization/80492
33232 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33233 compare_base_decls returning dont-know properly.
33235 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
33237 * config/arm/iterators.md (CCSI): New mode iterator.
33238 (arch): New mode attribute.
33239 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
33240 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
33241 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
33242 code iterator for success result mode.
33243 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
33244 the corresponding new insn generators.
33246 2017-05-03 Bin Cheng <bin.cheng@arm.com>
33249 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33250 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33252 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
33254 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
33255 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
33256 (DDR_A): Wrap DDR argument in brackets.
33257 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
33258 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
33259 (DDR_REVERSED_P): Likewise.
33261 2017-05-03 Jakub Jelinek <jakub@redhat.com>
33263 PR tree-optimization/79472
33264 * tree-switch-conversion.c (struct switch_conv_info): Add
33265 contiguous_range and default_case_nonstandard fields.
33266 (collect_switch_conv_info): Compute contiguous_range and
33267 default_case_nonstandard fields, don't clear final_bb if
33268 contiguous_range and only the default case doesn't have the required
33270 (check_all_empty_except_final): Set default_case_nonstandard instead
33271 of failing if contiguous_range and the default case doesn't have empty
33273 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
33274 and only the default case doesn't have the required constants. Skip
33276 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
33277 if default_case_nonstandard.
33278 (build_constructors): Build constant 1 just once. Assert that default
33279 values aren't inserted in between cases if contiguous_range. Skip
33281 (build_arrays): Skip virtual phis.
33282 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
33283 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
33284 Handle virtual phis.
33285 (gen_inbound_check): Handle default_case_nonstandard case.
33286 (process_switch): Adjust check_final_bb caller. Call
33287 gather_default_values with the first non-default case instead of
33288 default case if default_case_nonstandard.
33290 2017-05-02 Nathan Sidwell <nathan@acm.org>
33292 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
33293 check. Fix formatting.
33295 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
33297 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
33298 errors when comparing specialized and unspecialized times.
33300 2017-05-02 David Malcolm <dmalcolm@redhat.com>
33302 * diagnostic-show-locus.c
33303 (layout::should_print_annotation_line_p): Make private.
33304 (layout::print_annotation_line): Make private.
33305 (layout::annotation_line_showed_range_p): Make private.
33306 (layout::show_ruler): Make private.
33307 (layout::print_source_line): Make private. Pass in line and
33308 line_width, rather than calling location_get_source_line. Drop
33310 (layout::print_leading_fixits): New method.
33311 (layout::print_any_fixits): Rename to...
33312 (layout::print_trailing_fixits): ...this, and make private.
33313 Don't print newline fixits.
33314 (diagnostic_show_locus): Move logic for printing one row into...
33315 (layout::print_line): ...this new function. Move the
33316 location_get_source_line call and error-handling from
33317 print_source_line to here. Call print_leading_fixits, and rename
33318 print_any_fixits to print_trailing_fixits.
33319 (selftest::test_fixit_insert_containing_newline): Update now that
33320 newlines are partially supported.
33321 (selftest::test_fixit_insert_containing_newline_2): New test.
33322 (selftest::test_fixit_replace_containing_newline): Update comments.
33323 (selftest::diagnostic_show_locus_c_tests): Call the new test.
33324 * edit-context.c (class added_line): New class.
33325 (class edited_line): Describe newline handling in comment.
33326 (edited_line::actually_edited_p): New method.
33327 (edited_line::print_content): Delete redundant decl.
33328 (edited_line::m_predecessors): New field.
33329 (edited_file::print_content): Call edited_line::print_content.
33330 (edited_file::print_diff): Update to support newlines.
33331 (edited_file::print_diff_hunk): Likewise.
33332 (edited_file::print_run_of_changed_lines): New function.
33333 (edited_file::print_diff_line): Convert to...
33334 (print_diff_line): ...this.
33335 (edited_file::get_effective_line_count): New function.
33336 (edited_line::edited_line): Initialize new field m_predecessors.
33337 (edited_line::~edited_line): Clean up m_predecessors.
33338 (edited_line::apply_fixit): Handle newlines.
33339 (edited_line::get_effective_line_count): New function.
33340 (edited_line::print_content): New function.
33341 (edited_line::print_diff_lines): New function.
33342 (selftest::test_applying_fixits_insert_containing_newline): New
33344 (selftest::test_applying_fixits_replace_containing_newline): New
33346 (selftest::insert_line): New function.
33347 (selftest::test_applying_fixits_multiple_lines): Add example of
33349 (selftest::edit_context_c_tests): Call the new tests.
33351 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33353 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
33354 parameter cand. Update dump information.
33355 (get_computation_cost): Update uses.
33357 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33359 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
33360 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
33361 (get_computation_at, rewrite_use_address): Update use of
33362 get_computation_aff.
33364 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33366 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
33367 (get_computation): Delete.
33368 (get_computation_cost): Implement like get_computation_cost_at.
33369 Use get_computation_at.
33370 (get_computation_cost_at): Delete.
33371 (rewrite_use_nonlinear_expr): Use get_computation_at.
33372 (rewrite_use_compare, remove_unused_ivs): Ditto.
33374 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33376 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
33378 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33380 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
33381 (ivopts_global_cost_for_size): Rename parameter and update uses.
33382 (iv_ca_recount_cost): Update uses.
33383 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
33384 candidates seperately in n_invs and n_cands.
33385 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
33387 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33389 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
33390 (find_inv_vars_cb): New.
33391 (find_depends): Renamed to ...
33392 (find_inv_vars): ... this.
33393 (add_candidate_1, force_var_cost): Call find_inv_vars.
33394 (split_address_cost, determine_group_iv_cost_cond): Ditto.
33396 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33398 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
33399 inv_vars. Add inv_exprs.
33400 (struct iv_cand): Rename depends_on to inv_vars.
33401 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
33402 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
33403 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
33404 (dump_cand): Dump inv_vars.
33405 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
33406 (record_invariant, find_depends, add_candidate_1): Ditto.
33407 (set_group_iv_cost, force_var_cost): Ditto.
33408 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
33409 (get_computation_cost_at, get_computation_cost): Ditto.
33410 (determine_group_iv_cost_generic): Ditto.
33411 (determine_group_iv_cost_address): Ditto.
33412 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
33413 (determine_group_iv_costs): Ditto.
33414 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
33415 (iv_ca_set_remove_invariants): Renamed to ...
33416 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
33417 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
33418 (iv_ca_set_add_invariants): Renamed to ...
33419 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
33420 (iv_ca_set_cp): Use iv_ca_set_add_invs.
33421 (iv_ca_has_deps): Support inv_vars and inv_exprs.
33422 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
33423 (create_new_ivs): Remove useless dump.
33425 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33427 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
33429 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
33430 (iv_ca_set_no_cp, create_new_iv): Ditto.
33432 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33434 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33436 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33438 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
33439 function tree_check2.
33441 2017-05-02 Martin Liska <mliska@suse.cz>
33443 * doc/gcov.texi: Add missing preposition.
33444 * gcov.c (function_info::function_info): Properly fill up
33445 all member variables.
33447 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33449 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
33451 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33453 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
33455 2017-05-02 Martin Liska <mliska@suse.cz>
33458 * lto-streamer-in.c (lto_read_tree_1): Remove
33459 LTO_STREAMER_DEBUG.
33460 * lto-streamer.c (struct tree_hash_entry): Likewise.
33461 (struct tree_entry_hasher): Likewise.
33462 (tree_entry_hasher::hash): Likewise.
33463 (tree_entry_hasher::equal): Likewise.
33464 (lto_streamer_init): Likewise.
33465 (lto_orig_address_map): Likewise.
33466 (lto_orig_address_get): Likewise.
33467 (lto_orig_address_remove): Likewise.
33468 * lto-streamer.h: Likewise.
33469 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
33470 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
33472 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
33474 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
33475 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
33476 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
33477 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
33478 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
33479 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
33480 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
33481 (mm_maskz_sub_ss): New intrinsics.
33482 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33483 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
33484 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
33485 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
33486 (__builtin_ia32_subss_mask_round): New builtins.
33487 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33488 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
33489 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
33491 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
33492 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
33494 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
33497 2017-05-02 Martin Jambor <mjambor@suse.cz>
33499 PR tree-optimization/78687
33500 * tree-sra.c (access): New field parent.
33501 (process_subtree_disqualification): New function.
33502 (disqualify_candidate): Call it.
33503 (build_accesses_from_assign): Reset write flag if creating an
33505 (build_access_subtree): Fill in parent field and also prpagate
33506 down grp_write flag.
33507 (create_artificial_child_access): New parameter set_grp_write, set
33508 grp_write to its value.
33509 (propagate_subaccesses_across_link): Also propagate grp_write flag
33511 (propagate_all_subaccesses): Push the closest parent back to work
33512 queue if add_access_to_work_queue returned true.
33514 2017-05-02 Richard Biener <rguenther@suse.de>
33516 * common.opt (fstrict-overflow): Alias negative to fwrapv.
33517 * doc/invoke.texi (fstrict-overflow): Remove all traces of
33518 -fstrict-overflow documentation.
33519 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
33520 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
33521 flag_strict_overflow.
33522 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
33523 * lto-opts.c (lto_write_options): Do not stream it.
33524 * lto-wrapper.c (merge_and_complain): Do not handle it.
33525 * opts.c (default_options_table): Do not set -fstrict-overflow.
33526 (finish_options): Likewise do not clear it when sanitizing.
33527 * simplify-rtx.c (simplify_const_relational_operation): Do not
33528 test flag_strict_overflow.
33530 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33532 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
33533 using enabled attribute.
33534 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
33535 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
33536 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
33537 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
33538 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
33539 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
33540 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
33541 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
33542 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
33543 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
33545 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33547 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
33549 2017-05-02 Richard Biener <rguenther@suse.de>
33551 PR tree-optimization/80591
33553 2017-04-10 Richard Biener <rguenther@suse.de>
33555 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33558 2017-05-02 Richard Biener <rguenther@suse.de>
33560 PR tree-optimization/80549
33561 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
33562 (cleanup_tree_cfg_noloop): Create forwarders to known loop
33563 headers if they do not have a preheader.
33565 2017-05-02 Martin Liska <mliska@suse.cz>
33568 * common.opt: Fix typo.
33569 * doc/invoke.texi: Likewise.
33571 2017-05-01 Jan Beulich <jbeulich@suse.com>
33573 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
33574 swapping, add (x,x,m,x,n) alternative.
33576 2017-05-01 Nathan Sidwell <nathan@acm.org>
33578 * calls.c (combine_pending_stack_adjustment_and_call): Remove
33579 unnecessary unadjusted_alignment check.
33581 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33584 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
33586 * gimplify.c (gimplify_cilk_detach): New function.
33587 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
33588 * tree-core.h: Document EXPR_CILK_SPAWN.
33589 * tree.h (EXPR_CILK_SPAWN): Define.
33591 2017-05-01 David Malcolm <dmalcolm@redhat.com>
33593 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
33594 to use new fixit_hint representation, using the "replace" logic.
33595 (get_line_span_for_fixit_hint): Likewise.
33596 (layout::print_any_fixits): Likewise.
33597 (selftest::test_one_liner_many_fixits): Rename to...
33598 (selftest::test_one_liner_many_fixits_1): ...this, and update
33599 comment and expected output to reflect that the multiple fix-it
33600 hints are now consolidated into one insertion.
33601 (selftest::test_one_liner_many_fixits_2): New test.
33602 (selftest::test_diagnostic_show_locus_one_liner): Update for
33604 (selftest::test_fixit_consolidation): Update for fix-it API
33606 * diagnostic.c (print_parseable_fixits): Likewise.
33607 * edit-context.c (edited_line::m_line_events): Convert from
33608 auto_vec <line_event *> to auto_vec <line_event>.
33609 (class line_event): Convert from abstract base class to a concrete
33610 class, taking over the role of replace_event.
33611 (class insert_event): Delete.
33612 (class replace_event): Rename to class line_event. Convert to
33614 (edit_context::add_fixits): Reimplement.
33615 (edit_context::apply_insert): Delete.
33616 (edit_context::apply_replace): Rename to...
33617 (edit_context::apply_fixit): ...this. Convert to half-open range.
33618 (edited_file::apply_insert): Delete.
33619 (edited_file::apply_replace): Rename to...
33620 (edited_file::apply_fixit): ...this.
33621 (edited_line::~edited_line): Drop deletion of events.
33622 (edited_line::apply_insert): Delete.
33623 (edited_line::apply_replace): Rename to...
33624 (edited_line::apply_fixit): ...this. Convert to half-open range.
33625 Update for change to type of m_line_events.
33626 * edit-context.h (edit_context::apply_insert): Delete.
33627 (edit_context::apply_replace): Rename to...
33628 (edit_context::apply_fixit): ...this.
33630 2017-05-01 Martin Sebor <msebor@redhat.com>
33632 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
33635 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
33638 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
33639 __get_cpuid_max returns 0.
33640 (__get_cpuid_count): Ditto.
33642 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
33644 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
33645 replacement expression is another instance of one of its arguments.
33647 2017-05-01 Jakub Jelinek <jakub@redhat.com>
33650 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
33651 check for stack push/pop autoinc.
33652 * config/i386/i386.c (ix86_agi_dependent): Return false
33653 if the only reason why modified_in_p returned true is that
33654 addr is SP based and set_insn is a push or pop.
33656 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33658 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
33661 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33664 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
33665 (account_size_time): Use two predicates - exec_pred and
33667 (evaluate_conditions_for_known_args): Compute both clause and
33669 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
33670 (inline_summary_t::duplicate): Update.
33671 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
33673 (compute_inline_parameters): Likewise.
33674 (estimate_edge_size_and_time): Update caluclation of time.
33675 (estimate_node_size_and_time): Compute both time and nonspecialized
33677 (estimate_ipcp_clone_size_and_time): Update.
33678 (inline_merge_summary): Update.
33679 (do_estimate_edge_time): Update.
33680 (do_estimate_edge_size): Update.
33681 (do_estimate_edge_hints): Update.
33682 (inline_read_section, inline_write_summary): Stream both new predicates.
33683 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
33685 (compute_inlined_call_time): Cleanup.
33686 (big_speedup_p): Update.
33687 (edge_badness): Update.
33688 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
33689 (size_time_entry): Replace predicate by exec_predicate and
33690 nonconst_predicate.
33691 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
33692 (estimate_edge_time): Return also nonspec_time.
33693 (reset_edge_growth_cache): Update.
33695 2017-04-29 Jakub Jelinek <jakub@redhat.com>
33697 PR rtl-optimization/80491
33698 * ifcvt.c (noce_process_if_block): When looking for x setter
33699 with missing else_bb, don't check only the insn right before
33700 cond_earliest, but look for the last insn that x is modified in
33701 within the same bb.
33703 PR rtl-optimization/80491
33704 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
33706 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
33708 PR tree-optimization/80487
33709 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
33711 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33713 PR tree-optimization/79697
33714 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
33715 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
33716 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
33718 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
33719 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
33721 2017-04-28 Martin Sebor <msebor@redhat.com>
33723 PR tree-optimization/80523
33724 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
33725 (init_target_to_host_charmap, target_to_host, target_strtol10): New
33727 (maybe_warn, format_directive, parse_directive): Use new functions.
33728 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
33730 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
33732 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
33734 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
33736 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
33737 target_header_dir): Set correctly.
33738 * configure: Regenerated.
33739 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33740 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33741 instead of SYSTEM_HEADER_DIR.
33743 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
33745 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
33746 (estimate_local_effects): Likewise.
33747 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
33748 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
33749 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
33750 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
33751 do_estimate_edge_time, estimate_edge_time): Likewise.
33752 * ipa-inline-analysis.c (estimate_node_size_and_time,
33753 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
33754 (estimate_time_after_inlining): Remove.
33756 2017-04-28 Martin Liska <mliska@suse.cz>
33758 * doc/gcov.texi: Enhance documentation of gcov.
33760 2017-04-28 Martin Liska <mliska@suse.cz>
33762 * doc/gcov.texi: Sort options in alphabetic order.
33763 * doc/gcov-dump.texi: Likewise.
33764 * doc/gcov-tool.texi: Likewise.
33765 * gcov.c (print_usage): Likewise.
33766 * gcov-dump.c (print_usage): Likewise.
33767 * gcov-tool.c (print_merge_usage_message): Likewise.
33768 (print_rewrite_usage_message): Likewise.
33769 (print_overlap_usage_message): Likewise.
33771 2017-04-28 Martin Liska <mliska@suse.cz>
33773 PR gcov-profile/53915
33774 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33776 2017-04-28 Martin Liska <mliska@suse.cz>
33778 PR gcov-profile/79891
33779 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33780 is marked by compiler as living on a line.
33781 (get_cycles_count): Remove usage of the union.
33782 (output_intermediate_file): Likewise.
33783 (find_source): Fix GNU coding style.
33784 (accumulate_line_counts): Remove old non-all block mode.
33785 (output_lines): Remove usage of the union.
33786 * profile.c (output_location): Include all BBs, even if
33787 belonging to a same line (and file) as a previous BB.
33789 2017-04-28 Martin Liska <mliska@suse.cz>
33791 * gcov.c (process_args): Handle new argument 'w'.
33792 (read_graph_file): Assign ID to BBs.
33793 (output_branch_count): Display BB # if verbose flag is set.
33794 (output_lines): Likewise for arcs.
33795 (print_usage): Add '--verbose' option help.
33796 * doc/gcov.texi: Document --verbose (-w) option.
33798 2017-04-28 Martin Liska <mliska@suse.cz>
33800 * gcov.c (struct block_location_info): New struct.
33801 (process_file): Fill up the new structure.
33802 (read_graph_file): Replace usage of encoding by the newly added
33804 (add_line_counts): Likewise.
33805 (accumulate_line_counts): Remove usage of the union.
33806 (function_info::function_info): New function.
33807 (function_info::~function_info): Likewise.
33808 (process_file): Call delete instead of release_function.
33809 (release_function): Release the function.
33810 (release_structures): Call delete instead of release_function.
33811 (solve_flow_graph): Replace usage of num_blocks.
33812 (find_exception_blocks): Likewise.
33813 (output_lines): Fix GNU coding style.
33815 2017-04-28 Martin Liska <mliska@suse.cz>
33818 * coverage.c (coverage_remove_note_file): New function.
33819 * coverage.h: Declare the function.
33820 * toplev.c (finalize): Clean if an error has been seen.
33822 2017-04-28 Martin Liska <mliska@suse.cz>
33824 PR gcov-profile/80031
33825 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33826 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33827 * gcov.c (read_graph_file): Read just number of blocks.
33828 * profile.c (branch_prob): Do not stream 0 flags per a basic
33831 2017-04-28 Martin Liska <mliska@suse.cz>
33833 * gcov-dump.c (tag_*): Add new argument to declarations.
33834 (dump_gcov_file): Likewise.
33835 (tag_blocks): Add and use new argument depth.
33836 (tag_arcs): Likewise.
33837 (tag_lines): Likewise.
33838 (tag_counters): Likewise.
33839 (tag_summary): Likewise.
33840 (dump_working_sets): Use depth to do a proper indentation.
33842 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33845 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33846 * symtab.c (symtab_node::debug_symtab): Move definition here.
33848 2017-04-28 Richard Biener <rguenther@suse.de>
33850 * lto-streamer.h (LTO_major_version): Bump to 7.
33852 2017-04-28 Richard Biener <rguenther@suse.de>
33854 * tree-vrp.c (assert_info): New struct.
33855 (add_assert_info): New helper.
33856 (register_edge_assert_for_2): Refactor to add asserts to a vector
33858 (register_edge_assert_for_1): Likewise.
33859 (register_edge_assert_for): Likewise.
33860 (finish_register_edge_assert_for): New helper actually registering
33861 asserts where live on edge.
33862 (find_conditional_asserts): Adjust.
33863 (find_switch_asserts): Likewise.
33864 (evrp_dom_walker::try_find_new_range): Generalize.
33865 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33867 2017-04-27 Marek Polacek <polacek@redhat.com>
33870 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33871 arg10 and arg11 to itype.
33873 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33875 * doc/extend.texi (Object Size Checking): Improve grammar.
33877 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33880 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33881 that the logic for permitting reciprocal estimates matches that
33884 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33887 * tree.c (type_cache_hasher::equal): Only compare
33888 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33889 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33890 non-aggregate element types.
33891 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33892 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33894 2017-04-27 Richard Biener <rguenther@suse.de>
33896 PR middle-end/80533
33897 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33898 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33899 keeping a reference to a trailing array.
33901 2017-04-27 Richard Biener <rguenther@suse.de>
33903 PR middle-end/80539
33904 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33905 being in loop-closed SSA form conservatively.
33906 (chrec_fold_multiply_poly_poly): Likewise.
33908 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33910 PR middle-end/79665
33911 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33912 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33914 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33917 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33918 (aarch64_function_arg_alignment): Return unsigned int again, but still
33919 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33920 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33921 Don't emit -Wpsabi note.
33922 (aarch64_function_arg_boundary): Likewise.
33923 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33926 2017-04-26 Nathan Sidwell <nathan@acm.org>
33928 * tree.h (crc32_unsigned_n): Declare.
33929 (crc32_unsigned, crc32_unsigned): Make inline.
33930 * tree.c (crc32_unsigned_bits): Replace with ...
33931 (crc32_unsigned_n): ... this.
33932 (crc32_unsigned, crc32_byte): Remove.
33933 (crc32_string): Remove unnecessary braces.
33935 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33937 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33938 * ipa-inline-analysis.c (MAX_TIME): Remove.
33939 (account_size_time): Use sreal for time.
33940 (dump_inline_summary): Update.
33941 (estimate_function_body_sizes): Update.
33942 (estimate_edge_size_and_time): Update.
33943 (estimate_calls_size_and_time): Update.
33944 (estimate_node_size_and_time): Update.
33945 (inline_merge_summary): Update.
33946 (inline_update_overall_summary): Update.
33947 (estimate_time_after_inlining): Update.
33948 (inline_read_section): Update.
33949 (inline_write_summary): Update.
33950 * ipa-inline.c (compute_uninlined_call_time): Update.
33951 (compute_inlined_call_time): Update.
33952 (recursive_inlining): Update.
33953 (inline_small_functions): Update.
33954 (dump_overall_stats): Update.
33955 * ipa-inline.h: Include sreal.h.
33956 (size_time_entry): Turn time to sreal.
33957 (inline_summary): Turn self_time nad time to sreal.
33959 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33961 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33963 (sreal::stream_out, sreal::stream_in): New.
33964 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33966 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33968 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33971 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33974 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33976 (dimode_scalar_chain::compute_convert_gain): Ditto.
33977 (dimode_scalar_chain::make_vector_copies): Ditto.
33978 (dimode_scalar_chain::convert_reg): Ditto.
33979 (dimode_scalar_chain::convert_insn): Ditto.
33980 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33981 (VI248_AVX512BW_1): New mode iterator.
33982 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33983 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33986 2017-04-25 Martin Sebor <msebor@redhat.com>
33988 PR tree-optimization/80497
33989 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33990 constants are representable in HOST_WIDE_INT.
33991 (parse_directive): Ditto.
33993 2017-04-25 Martin Sebor <msebor@redhat.com>
33996 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33997 (new_zero_array): Adjust signature.
33998 (dom_info::dom_init): Used unsigned rather that size_t.
33999 (dom_info::dom_info): Same.
34001 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
34002 Jakub Jelinek <jakub@redhat.com>
34005 * config/arm/arm.c: Include gimple.h.
34006 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
34007 returns negative, increment ncrn only if it returned positive.
34008 (arm_needs_doubleword_align): Return int instead of bool,
34009 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
34010 members, but if there is any such non-FIELD_DECL
34011 > PARM_BOUNDARY aligned decl, return -1 instead of false.
34012 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
34013 returns negative, increment nregs only if it returned positive.
34014 (arm_setup_incoming_varargs): Likewise.
34015 (arm_function_arg_boundary): Emit -Wpsabi note if
34016 arm_needs_doubleword_align returns negative, return
34017 DOUBLEWORD_ALIGNMENT only if it returned positive.
34019 2017-04-25 Marek Polacek <polacek@redhat.com>
34022 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34023 first argument to type.
34025 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
34028 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
34029 type checks to test for compatibility instead of equality.
34031 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
34032 Jakub Jelinek <jakub@redhat.com>
34035 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
34037 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
34038 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
34039 the alignment computation, but return their maximum in warn_alignment.
34040 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
34041 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
34043 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
34044 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
34047 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34049 * config/arc/simdext.md (dmpyh): Fix typo.
34051 2017-04-25 Richard Biener <rguenther@suse.de>
34053 PR tree-optimization/80492
34054 * alias.c (compare_base_decls): Handle registers with asm
34055 specification conservatively.
34056 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
34057 compare_base_decls returning dont-know properly.
34059 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34061 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
34062 (legitimate_offset_address_p): New function.
34063 (arc_legitimate_address_p): Use above function.
34065 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34067 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
34069 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34071 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
34072 ACCH registers whenever they are available.
34074 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34076 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
34077 double regs fix when not used.
34079 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34081 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
34083 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
34084 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
34086 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34088 * config/arc/arc.c (arc_output_addsi): Check for h-register class
34089 when emitting short ADD instructions.
34091 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34093 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
34095 (cmpsi_cc_c_insn): Likewise.
34096 (cbranchsi4_scratch): Compute proper instruction length using
34097 compact_hreg_operand.
34098 * config/arc/predicates.md (compact_hreg_operand): New predicate.
34100 2017-04-25 Richard Biener <rguenther@suse.de>
34102 PR middle-end/80509
34103 * passes.c (pass_manager::pass_manager): Initialize
34104 m_name_to_pass_map.
34106 2017-04-25 Richard Biener <rguenther@suse.de>
34108 PR tree-optimization/79201
34109 * tree-ssa-sink.c (statement_sink_location): Handle calls.
34111 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34114 * config/s390/vector.md: Split MEM->GPR vector moves for
34115 non-s_operand addresses.
34117 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34120 * config/s390/predicates.md (reload_const_wide_int_operand): New
34122 * config/s390/s390.md ("movti"): Remove d/P alternative.
34123 ("movti_bigconst"): New pattern definition.
34125 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34128 * s390-protos.h (s390_expand_cs_hqi): Removed.
34129 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
34130 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
34131 modes as well as CCZ1mode and CCZmode.
34132 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
34133 signature of s390_emit_compare_and_swap.
34134 (s390_expand_cs_hqi): Likewise, make static.
34135 (s390_expand_cs_tdsi): Generate an explicit compare before trying
34136 compare-and-swap, in some cases.
34137 (s390_expand_cs): Wrapper function.
34138 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
34140 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
34141 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
34142 patterns for small and large integers. Forbid symref memory operands.
34143 Move expander to s390.c. Require cc register.
34144 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
34145 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
34146 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
34147 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
34148 symref memory operands. Remove CC mode and call s390_match_ccmode
34150 ("atomic_exchange<mode>"): Allow and implement all integer modes.
34152 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34154 * config/s390/s390.md (define_peephole2): New peephole to help
34155 combining the load-and-test pattern with volatile memory.
34157 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34159 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
34160 with CCZmode for TARGET_Z196.
34162 2017-04-25 Jakub Jelinek <jakub@redhat.com>
34164 PR rtl-optimization/80501
34165 * combine.c (make_compound_operation_int): Set subreg_code to SET
34166 even for AND with mask of the sign bit of mode.
34168 PR rtl-optimization/80500
34169 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
34170 sum's initial value.
34172 2017-04-25 Julian Brown <julian@codesourcery.com>
34173 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34175 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
34177 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
34179 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
34181 2017-04-25 Julian Brown <julian@codesourcery.com>
34182 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34184 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
34185 (thunderx2t99_sha): New Reservation.
34187 2017-04-25 Julian Brown <julian@codesourcery.com>
34188 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34190 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
34191 type for 1-element load.
34193 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
34195 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
34197 2017-04-24 Martin Jambor <mjambor@suse.cz>
34199 PR tree-optimization/80293
34200 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
34201 char arrays not totally scalarizable if it is false.
34202 (analyze_all_variable_accesses): Pass correct value in the new
34203 parameter. Add a statistics counter.
34205 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
34207 PR middle-end/79931
34208 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
34210 2017-04-24 Richard Biener <rguenther@suse.de>
34212 PR tree-optimization/80494
34213 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
34214 out for complex types.
34216 2017-04-24 Richard Biener <rguenther@suse.de>
34218 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
34219 * tree-ssa-sccvn.c (print_scc): Print SCC size.
34220 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
34221 (DFS): Adjust and never fail.
34222 (sccvn_dom_walker::fail): Remove.
34223 (sccvn_dom_walker::before_dom_children): Adjust.
34224 (run_scc_vn): Likewise and never fail.
34225 * tree-ssa-pre.c (pass_pre::execute): Adjust.
34226 (pass_fre::execute): Likewise.
34228 2017-04-24 Richard Biener <rguenther@suse.de>
34230 PR tree-optimization/79725
34231 * tree-ssa-sink.c (statement_sink_location): Return whether
34232 failure reason was zero uses. Move that check later.
34233 (sink_code_in_bb): Deal with zero uses by removing the stmt
34236 2017-04-24 Richard Biener <rguenther@suse.de>
34239 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
34240 pointer-based references.
34242 2017-04-24 Richard Biener <rguenther@suse.de>
34245 * pass_manager.h (pass_manager::operator new): Remove.
34246 (pass_manager::operator delete): Likewise.
34247 * passes.c (pass_manager::operator new): Remove.
34248 (pass_manager::operator delete): Likewise.
34249 (pass_manager::pass_manager): Zero individual pass members.
34251 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
34254 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
34255 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
34256 Check "XEXP (src, 1)" operand here.
34257 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
34258 Check "XEXP (src, 1)" operand here.
34259 (dimode_scalar_chain::make_vector_copies): Detect count register
34260 of a shift instruction. Zero extend count register from QImode
34261 to DImode to satisfy vector shift pattern count operand predicate.
34262 Substitute vector shift count operand with a DImode copy.
34263 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
34266 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
34268 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
34269 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34270 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
34271 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34272 (UNSPEC_NOREX_MEM): Remove definition.
34274 2017-04-21 Richard Biener <rguenther@suse.de>
34276 PR tree-optimization/79547
34277 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
34278 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
34279 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
34280 without any constraints.
34282 2017-04-21 Richard Biener <rguenther@suse.de>
34284 PR tree-optimization/78847
34285 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
34287 2017-04-21 Richard Biener <rguenther@suse.de>
34289 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
34290 (build_distinct_type_copy): Likewise.
34291 (build_variant_type_copy): Likewise.
34292 * tree.c (build_qualified_type): Pass down mem-stat info.
34293 (build_distinct_type_copy): Likewise.
34294 (build_variant_type_copy): Likewise.
34296 2017-04-21 Richard Biener <rguenther@suse.de>
34298 PR tree-optimization/80237
34299 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
34301 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
34302 for a simplified result.
34304 2016-04-21 Richard Biener <rguenther@suse.de>
34306 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
34307 sth as strict as a simple_iv but a chrec without symbols and an
34308 operand defined in the loop we are peeling (and not some subloop).
34309 (propagate_constants_for_unrolling): Propagate all constants.
34311 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34314 * config/i386/i386.c (print_reg): Remove assert for disalowed
34315 regno values, call output_operand_lossage instead.
34317 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34320 * config/i386/constraints.md (Yc): New register constraint.
34321 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
34322 Use Yc constraint for alternative 2 of operand 0. Remove
34323 preferred_for_speed attribute.
34325 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34327 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
34328 lastprivate clauses in SIMT case.
34330 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
34332 * doc/invoke.texi (-Wextra-semi): Document new warning option.
34334 2017-04-20 Richard Biener <rguenther@suse.de>
34336 PR tree-optimization/57796
34337 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
34338 as N scalar stores.
34339 (vect_model_load_cost): Cost gathers as N scalar loads.
34341 2017-04-20 Richard Biener <rguenther@suse.de>
34343 * ggc-page.c (ggc_allocated_p): Rename to ...
34344 (safe_lookup_page_table_entry): ... this and return the lookup
34346 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
34348 2017-04-20 Richard Biener <rguenther@suse.de>
34350 PR tree-optimization/80453
34351 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
34352 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
34353 from the conditions.
34354 (vn_phi_eq): Pass them down.
34355 (vn_phi_lookup): Record them.
34356 (vn_phi_insert): Likewise.
34358 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
34360 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
34361 uninitialized variable warning to avoid buffer overrun.
34363 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34366 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
34367 is suppressed for '{ 0 }' in C.
34369 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34371 * BASE-VER: Set to 8.0.0.
34373 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
34375 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
34376 priority .init_array and .fini_array section with SECTION_NOTYPE
34379 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34381 PR middle-end/80423
34382 * tree.h (build_array_type): Add typeless_storage default argument.
34383 * tree.c (type_cache_hasher::equal): Also compare
34384 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
34385 (build_array_type): Add typeless_storage argument, set
34386 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
34388 (build_nonshared_array_type): Adjust build_array_type_1 caller.
34389 (build_array_type): Likewise. Add typeless_storage argument.
34391 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34392 Jakub Jelinek <jakub@redhat.com>
34394 PR tree-optimization/80426
34395 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
34396 operation on symbolic operands, also compute the overflow for the
34397 invariant part when the operation degenerates into a negation.
34399 2017-04-19 Jakub Jelinek <jakub@redhat.com>
34402 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
34403 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
34406 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
34408 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
34411 * config/avr/avr.c (tree.h): Include it.
34412 (cgraph.h): Include it.
34413 (avr_encode_section_info): Don't warn for uninitialized progmem
34414 variable if it's just an alias.
34416 2017-04-19 Richard Biener <rguenther@suse.de>
34419 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
34420 when needed by AutoPGO.
34422 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
34425 * doc/lto.texi: Remove an extra 'that'.
34427 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
34429 PR rtl-optimization/80429
34430 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
34431 are only used in debug insns.
34433 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34434 Vladimir Makarov <vmakarov@redhat.com>
34436 * config/sparc/predicates.md (input_operand): Add comment. Return
34437 true for any memory operand when LRA is in progress.
34438 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
34440 2017-04-18 Jeff Law <law@redhat.com>
34443 * mips.md ({return,simple_return}_internal): Do not overwrite
34446 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34448 PR tree-optimization/80443
34449 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
34450 instead of adding 1, subtract -1 and similarly instead of subtracting
34453 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
34455 PR rtl-optimization/80357
34456 * haifa-sched.c (tmp_bitmap): New variable.
34457 (model_recompute): Handle duplicate use records.
34458 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
34459 (free_global_sched_pressure_data): Free it.
34461 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
34464 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34465 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34466 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34467 instead of SYSTEM_HEADER_DIR.
34469 2017-04-18 Jeff Law <law@redhat.com>
34471 PR middle-end/80422
34472 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
34473 predecessors after walking up the insn chain.
34475 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34478 * dwarf2out.c (modified_type_die): Try harder not to emit internal
34479 sizetype type into debug info.
34481 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34484 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
34485 unneeded test for TARGET_UPPER_REGS_SF.
34486 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
34488 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34491 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
34492 instead of gsi_after_labels.
34494 2017-04-18 Jeff Law <law@redhat.com>
34496 * regcprop.c (maybe_mode_change): Avoid creating copies of the
34500 2017-04-13 Jeff Law <law@redhat.com>
34501 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34502 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34504 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
34507 * config/avr/avr.c (intl.h): Include it.
34508 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
34510 2017-04-18 Martin Liska <mliska@suse.cz>
34512 PR gcov-profile/78783
34513 * gcov-tool.c (gcov_output_files): Validate that destination
34514 file is either removed by the tool or by a user.
34516 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34517 Guy Benyei <guybe@mellanox.com>
34519 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
34520 block, and do not negate it, the stored id is already negative.
34522 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34524 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
34526 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34529 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
34530 masks of options that should be turned off if the VSX vector
34531 options are turned off.
34532 (OTHER_P8_VECTOR_MASKS): Likewise.
34533 (OTHER_VSX_VECTOR_MASKS): Likewise.
34534 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
34535 rs6000_disable_incompatible_switches to validate no type switches
34537 (rs6000_incompatible_switch): New function to disallow turning on
34538 other vector options if -mno-vsx, -mno-power8-vector, or
34539 -mno-power9-vector are specified.
34541 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34543 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
34545 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34547 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
34548 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
34549 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
34550 (ARG_POINTER_CFA_OFFSET): Likewise.
34552 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34554 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
34555 conditions to take advantage of various optimizations.
34557 2017-04-13 Jeff Law <law@redhat.com>
34559 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34560 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34561 (zero_extendsidi2_dext): Likewise.
34563 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34566 * fold-const.c (fold_ternary_loc): Revert
34567 use op0 instead of fold_convert_loc (loc, type, arg0) part of
34570 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
34572 PR rtl-optimization/80343
34573 * lra-remat.c (update_scratch_ops): Assign original hard reg to
34574 new scratch pseudo.
34576 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
34579 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
34580 to ubsan_encode_value.
34582 2017-04-13 Jeff Law <law@redhat.com>
34584 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
34585 appearing in DEBUG_INSNs.
34587 2017-04-13 Martin Liska <mliska@suse.cz>
34589 PR gcov-profile/80413
34590 * gcov-io.c (gcov_write_string): Copy to buffer just when
34591 allocated size is greater than zero.
34593 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34596 * dwarf2out.c (decls_for_scope): Ignore declarations of
34597 current_function_decl in BLOCK_NONLOCALIZED_VARS.
34599 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
34602 * ipa-visibility.c (non_local_p): Fix typos.
34603 (localize_node): When localizing symbol in same comdat group,
34604 dissolve the group only when we know external symbols are going
34606 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
34608 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34610 PR tree-optimization/79390
34611 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
34612 order does not result in usable sequence, retry with reversed operand
34618 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
34619 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
34620 op0 instead of fold_convert_loc (loc, type, arg0).
34622 2017-04-12 Jeff Law <law@redhat.com>
34624 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
34625 has a delay slot in the generated code.
34627 * config/cris/cris.md (cris_preferred_reload_class): Return
34628 GENNONACR_REGS rather than GENERAL_REGS.
34630 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34633 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
34634 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
34635 signedness of the result type.
34637 2017-04-12 Richard Biener <rguenther@suse.de>
34638 Jeff Law <law@redhat.com>
34640 PR tree-optimization/80359
34641 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
34642 trim stores to TARGET_MEM_REFs.
34644 2017-04-12 Richard Biener <rguenther@suse.de>
34646 PR tree-optimization/79390
34647 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
34648 threading case even more.
34650 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
34653 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
34654 for quad_address_p for TImode, instead of just not indexed_address.
34656 2017-04-12 Richard Biener <rguenther@suse.de>
34657 Bernd Edlinger <bernd.edlinger@hotmail.de>
34659 PR middle-end/79671
34660 * alias.c (component_uses_parent_alias_set_from): Handle
34661 TYPE_TYPELESS_STORAGE.
34662 (get_alias_set): Likewise.
34663 * tree-core.h (tree_type_common): Add typeless_storage flag.
34664 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
34665 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
34666 for types containing members with TYPE_TYPELESS_STORAGE.
34667 (place_field): Likewise.
34668 (layout_type): Likewise for ARRAY_TYPE.
34669 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
34670 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
34671 TYPE_TYPELESS_STORAGE.
34672 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
34674 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34677 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34678 first argument to type.
34680 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34684 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
34685 CONST0_RTX (mode) rather than const0_rtx where appropriate.
34686 (rs6000_expand_binop_builtin): Likewise.
34687 (rs6000_expand_ternop_builtin): Likewise; also add missing
34688 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
34689 vshasigma built-ins.
34690 * doc/extend.texi: Document that vec_xxpermdi's third argument
34691 must be a constant.
34693 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
34695 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
34696 Use shift_const cost parameter when calculating gain of STV shifts.
34698 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
34700 PR rtl-optimization/70478
34701 * lra-constraints.c (process_alt_operands): Check memory for
34702 disfavoring memory insn operand.
34704 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34706 PR middle-end/80100
34707 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
34708 left shift in unsigned HOST_WIDE_INT type.
34710 PR rtl-optimization/80385
34711 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
34712 (not (neg X)) into (plus X -1) for complex or non-integral modes.
34715 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
34716 if they have any depend clauses.
34718 2017-04-11 Martin Liska <mliska@suse.cz>
34721 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
34722 * ipa-split.c (split_function): Create a local comdat symbol
34723 if caller is in a comdat group.
34725 2017-04-11 Martin Liska <mliska@suse.cz>
34728 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
34731 2017-04-11 Martin Sebor <msebor@redhat.com>
34733 PR middle-end/80364
34734 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
34735 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
34737 (directive::set_width, directive::set_precision, format_character):
34739 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
34742 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
34745 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
34746 conflict, set target->arch_name instead of target->cpu_name.
34748 2017-04-11 Richard Biener <rguenther@suse.de>
34750 PR tree-optimization/80374
34751 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
34752 build_zero_cst, remove fold_convertible_p check again.
34754 2017-04-11 Martin Liska <mliska@suse.cz>
34757 * ubsan.c (instrument_object_size): Do not instrument register
34760 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34763 * config/i386/i386-builtin-types.def
34764 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
34765 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
34766 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
34767 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
34768 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
34769 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
34770 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
34771 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
34772 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
34773 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
34774 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34775 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34776 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34777 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34778 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34779 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34780 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34781 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34782 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34783 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34784 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34785 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34786 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34787 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34788 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34789 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34790 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34791 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34792 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34793 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34794 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34795 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34796 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34797 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34798 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34799 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34800 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34801 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34802 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34803 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34804 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34805 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34806 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34807 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34808 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34809 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34811 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34812 flag to second_arg_count, handle 4 argument function type _COUNT
34813 aliases, handle second_arg_count on second argument rather than last.
34815 2017-04-10 Jeff Law <law@redhat.com>
34817 PR tree-optimization/80374
34818 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34819 record anything if we can not convert integer_zero_node to the
34822 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34825 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34826 Enhance special handling given to the TARGET_P9_MINMAX option in
34827 relation to certain other options.
34829 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34831 PR tree-optimization/80153
34832 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34833 remove POINTER_PLUS_EXPR's base part directly, rather than through
34836 2017-04-10 Richard Biener <rguenther@suse.de>
34837 Bin Cheng <bin.cheng@arm.com>
34839 PR tree-optimization/80153
34840 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34841 the first element of pointer type aff_tree. Build result expr in
34843 (add_elt_to_tree): Convert to type unconditionally. Remove other
34844 fold_convert calls.
34845 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34846 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34848 2017-04-10 Richard Biener <rguenther@suse.de>
34850 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34853 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34855 PR rtl-optimization/70478
34856 * lra-constraints.c (curr_small_class_check): New.
34857 (update_and_check_small_class_inputs): New.
34858 (process_alt_operands): Update curr_small_class_check. Disfavor
34859 alternative insn memory operands. Check available regs for small
34862 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34865 * config/mips/mips.opt (-mvirt): Update description.
34866 * doc/invoke.texi (-mvirt): Likewise.
34868 2017-04-10 Richard Biener <rguenther@suse.de>
34870 PR middle-end/80362
34871 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34872 looking for NEGATE_EXPR in -A / -B to A / B folding.
34874 2017-04-10 Martin Liska <mliska@suse.cz>
34876 PR gcov-profile/80224
34877 * gcov.c (print_usage): Fix usage string.
34878 (get_gcov_intermediate_filename): Remove.
34879 (output_gcov_file): Use both for normal and intermediate format.
34880 (generate_results): Do not initialize special file for
34881 intermediate format.
34883 2017-04-10 Richard Biener <rguenther@suse.de>
34885 PR tree-optimization/80304
34886 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34889 2017-04-10 Nathan Sidwell <nathan@acm.org>
34892 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34893 (rs6000_init_builtins): Use it.
34895 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34897 * config/arm/arm.md (<mrc>): Add mode to SET source.
34898 (<mrrc>): Likewise.
34900 2017-04-10 Richard Biener <rguenther@suse.de>
34902 PR middle-end/80344
34903 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34905 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34908 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34909 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34910 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34911 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34912 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34913 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34914 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34915 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34916 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34917 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34918 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34919 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34920 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34921 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34922 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34923 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34924 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34925 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34926 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34927 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34928 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34929 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34930 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34932 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34934 PR rtl-optimization/70478
34935 * lra-constraints.c: Reverse the last patch.
34937 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34939 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34940 Add comment for WCHAR_T.
34942 2017-04-08 Martin Liska <mliska@suse.cz>
34945 2017-04-07 Martin Liska <mliska@suse.cz>
34948 * ipa-split.c (split_function): Add function part to a same comdat
34951 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34954 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34956 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34958 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34959 (vec_pairing): ...to this.
34960 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34961 (rs6000_sched_init): Adjust for name change.
34962 (struct rs6000_sched_context): Likewise.
34963 (rs6000_init_sched_context): Likewise.
34964 (rs6000_set_sched_context): Likewise.
34966 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34972 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34974 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34975 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34976 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34978 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34980 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34982 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34984 PR rtl-optimization/70703
34985 * ira-color.c (update_conflict_hard_regno_costs): Use
34986 int64_t instead of HOST_WIDE_INT.
34988 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34990 PR rtl-optimization/70478
34991 * lra-constraints.c (process_alt_operands): Disfavor alternative
34992 insn memory operands.
34994 2017-04-07 Jeff Law <law@redhat.com>
34996 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34997 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34999 2017-04-07 Martin Liska <mliska@suse.cz>
35002 * config/aarch64/aarch64.c (aarch64_process_target_attr):
35003 Show error message instead of an ICE.
35005 2017-04-07 Martin Liska <mliska@suse.cz>
35008 * ipa-split.c (split_function): Add function part to a same comdat
35011 2017-04-07 Richard Biener <rguenther@suse.de>
35013 PR middle-end/80341
35014 * tree.c (get_unwidened): Also handle ! for_type case for
35016 * convert.c (do_narrow): Split out from ...
35017 (convert_to_integer_1): ... here. Do not pass final truncation
35018 type to get_unwidened for TRUNC_DIV_EXPR.
35020 2017-04-07 Richard Biener <rguenther@suse.de>
35022 * tree-affine.c (wide_int_ext_for_comb): Take type rather
35024 (aff_combination_const): Adjust.
35025 (aff_combination_scale): Likewise.
35026 (aff_combination_add_elt): Likewise.
35027 (aff_combination_add_cst): Likewise.
35028 (aff_combination_convert): Likewise.
35029 (add_elt_to_tree): Likewise. Remove unused argument.
35030 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
35032 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
35034 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
35036 * config/arm/arm.c (arm_default_short_enums): Use
35037 ARM_DEFAULT_SHORT_ENUMS.
35038 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
35040 2017-04-06 Jakub Jelinek <jakub@redhat.com>
35043 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
35044 members with redundant out-of-class redeclaration.
35046 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
35049 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
35050 * config/i386/i386.md (*zero_extendsidi2):
35051 Add (?*x,*x) and (?*v,*v) alternatives.
35053 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
35056 * config/i386/i386.c (ix86_expand_builtin)
35057 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
35058 mode from insn data. Convert operands to insn operand mode.
35059 Copy operands that don't satisfy insn predicate to a register.
35061 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
35063 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
35066 2017-04-06 Richard Biener <rguenther@suse.de>
35068 PR tree-optimization/80334
35069 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
35070 preserve alignment of accesses.
35072 2017-04-06 Richard Biener <rguenther@suse.de>
35074 PR tree-optimization/80262
35075 * tree-sra.c (build_ref_for_offset): Preserve address-space
35077 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
35078 Drop useless address-space information on MEM_REF offsets.
35080 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
35082 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
35084 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
35086 PR rtl-optimization/70703
35087 * ira-color.c (update_conflict_hard_regno_costs): Use
35088 HOST_WIDE_INT instead of long.
35090 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
35093 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
35094 not defined for x86_64 target. Add -mmmx target option when __SSE2__
35096 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
35097 for x86_64 target. Handle -m3dnowa option.
35099 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
35101 PR rtl-optimization/70703
35102 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
35103 (update_conflict_hard_regno_costs): Use long instead of unsigned
35104 arithmetic for cost calculation.
35106 2017-04-05 Jakub Jelinek <jakub@redhat.com>
35107 Bernd Edlinger <bernd.edlinger@hotmail.de>
35110 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
35113 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
35116 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
35117 ptr_mode with Pmode throughout.
35118 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
35119 into probe_stack_range and use DImode.
35121 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
35124 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
35125 call_eh_return is true.
35127 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35129 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
35130 Initialize last_match_fntype_index.
35132 2017-04-05 Jakub Jelinek <jakub@redhat.com>
35135 * tree-nvr.c: Include internal-fn.h.
35136 (pass_return_slot::execute): Ignore internal calls without
35139 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35140 Richard Biener <rguenther@suse.de>
35143 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
35144 captures used multiple times, except for the last use.
35145 * generic-match-head.c: Include gimplify.h.
35147 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35149 PR tree-optimization/79390
35150 * target.h (struct noce_if_info): Declare.
35151 * targhooks.h (default_noce_conversion_profitable_p): Declare.
35152 * target.def (noce_conversion_profitable_p): New target hook.
35153 * ifcvt.h (struct noce_if_info): New type, moved from ...
35154 * ifcvt.c (struct noce_if_info): ... here.
35155 (noce_conversion_profitable_p): Renamed to ...
35156 (default_noce_conversion_profitable_p): ... this. No longer
35158 (noce_try_store_flag_constants, noce_try_addcc,
35159 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
35160 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
35161 instead of noce_conversion_profitable_p.
35162 * config/i386/i386.c: Include ifcvt.h.
35163 (ix86_option_override_internal): Don't override
35164 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
35165 (ix86_noce_conversion_profitable_p): New function.
35166 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
35167 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
35168 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
35169 * doc/tm.texi: Regenerated.
35171 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35173 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
35176 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
35179 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
35180 instructions for small multiply cores.
35182 2017-04-04 Jeff Law <law@redhat.com>
35184 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
35186 (mips_expand_vec_perm_const): Initialize elements in orig_perm
35187 that are not set by the loop over the elements.
35189 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35192 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
35193 int mode, convert_modes it to mode as unsigned, otherwise use
35194 lowpart_subreg to mode rather than SImode.
35195 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
35196 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
35197 Use DImode instead of SImode for the shift count operand.
35198 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
35201 2017-04-04 Richard Biener <rguenther@suse.de>
35203 PR middle-end/80281
35204 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
35205 arithmetic done for the negate or the plus. Simplify.
35206 (A - (-B) -> A + B): Likewise.
35207 * fold-const.c (split_tree): Make sure to not negate pointers.
35209 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
35211 PR rtl-optimization/60818
35212 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
35213 a compare of comparisons with the thing compared if this results
35214 in a different machine mode.
35216 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
35218 * alias.c (base_alias_check): Fix typo in comment.
35219 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
35220 * cgraphunit.c (symbol_table::compile): Likewise.
35221 * collect2.c (maybe_run_lto_and_relink): Likewise.
35222 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
35223 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
35224 * config/avr/avr.c (avr_map_op_t): Likewise.
35225 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
35226 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
35227 * config/epiphany/epiphany.md (movcc): Likewise.
35228 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
35229 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
35231 * config/mips/mips.c (mips_save_restore_reg): Likewise.
35232 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
35233 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
35234 * config/sh/sh.c (sh_rtx_costs): Likewise.
35235 * fold-const.c (fold_truth_andor): Likewise.
35236 * genautomata.c (collapse_flag): Likewise.
35237 * gengtype.h (struct type::u::s): Likewise.
35238 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
35239 * input.c (FORMAT_AMOUNT): Likewise.
35240 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
35241 (known_aggs_to_agg_replacement_list): Likewise.
35242 * ipa-inline-analysis.c: Likewise.
35243 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
35244 * ipa-polymorphic-call.c
35245 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
35246 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
35247 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
35249 * modulo-sched.c (apply_reg_moves): Likewise.
35250 * omp-expand.c (build_omp_regions_1): Likewise.
35251 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
35252 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
35253 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
35254 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
35255 * value-prof.c: Likewise.
35256 * var-tracking.c (val_reset): Likewise.
35258 2017-04-03 Richard Biener <rguenther@suse.de>
35260 PR tree-optimization/80275
35261 * fold-const.c (split_address_to_core_and_offset): Handle
35264 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
35266 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
35267 descriptors is at least equal to that of functions.
35269 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35271 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
35273 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35276 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
35277 (mov<IMOD4:mode>): New expander.
35278 (*mov<IMOD4:mode>_internal): New insn and split pattern.
35280 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
35282 PR rtl-optimization/79405
35283 * fwprop.c (propagations_left): New variable.
35284 (forward_propagate_into): Decrement it.
35285 (fwprop_init): Initialize it.
35286 (fw_prop): If the variable has reached zero, stop propagating.
35287 (fwprop_addr): Ditto.
35289 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35292 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
35293 a FUNCTION_DECL, pass it as decl instead of origin to
35296 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
35298 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
35301 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
35304 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
35305 TARGET_VSX_SMALL_INTEGER.
35307 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35309 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
35310 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
35312 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
35314 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
35315 extraction from odd-numbered MSA register.
35317 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35319 PR middle-end/80173
35320 * expmed.c (store_bit_field_1): Don't attempt to create
35321 a word subreg out of hard registers wider than word if they
35322 have HARD_REGNO_NREGS of 1 for their mode.
35324 PR middle-end/80163
35325 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
35326 conversions to integer types wider than word and pointer.
35329 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
35330 (rtx_equal_for_cselib_p): Pass 0 to it.
35331 * cselib.c (cselib_hasher::equal): Likewise.
35332 (rtx_equal_for_cselib_1): Add depth argument. If depth
35333 is 128, don't look up VALUE locs and punt. Increment
35334 depth in recursive calls when walking VALUE locs.
35336 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
35338 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
35339 (make_gcov_file_name): Use the canonical path name for generating
35341 (read_line): Fix handling of files with ascii null bytes.
35343 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
35345 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
35346 to initialise a vector register instead
35347 of using a const_int.
35349 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35351 PR translation/80189
35352 * gimplify.c (omp_default_clause): Use %qs instead of %s in
35353 diagnostic messages.
35355 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
35358 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
35359 (dfp_diex_<mode>): Update mode of operand 1.
35360 * doc/extend.texi (dxex, dxexq): Document change to return type.
35361 (diex, diexq): Document change to argument type.
35363 2017-03-30 Martin Jambor <mjambor@suse.cz>
35366 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
35367 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
35368 it reflects the signature changes performed at the callee side.
35369 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
35370 to cgraph_build_function_type_skip_args.
35371 (build_function_decl_skip_args): Adjust call to the above function.
35373 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35376 * config/i386/sse.md
35377 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
35378 register as dest whenever it is a MEM not rtx_equal_p to the
35379 corresponding dup operand, and when forcing into reg move the
35380 reg into the memory afterwards.
35381 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
35382 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
35383 for the force_reg mode.
35384 (avx512vl_vextractf128<mode>): Use register as dest either
35385 always when a MEM, or when it is a MEM not rtx_equal_p to the
35386 corresponding dup operand, or even not when it is a CONST_VECTOR
35387 depending on the mode and lo vs. hi.
35388 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
35390 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
35391 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
35392 Likewise. Require that operands[2] is even.
35393 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
35394 Remove extraneous parens. Require that operands[2] is a multiple
35396 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
35397 operands[0] is a MEM if <mask_applied>, the predicates/constraints
35398 disallow memory then.
35400 2017-03-30 Richard Biener <rguenther@suse.de>
35402 PR tree-optimization/77498
35403 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
35404 to non-constants over backedges.
35406 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
35408 PR rtl-optimization/80233
35409 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
35410 as last_combined_insn. Do not test for BARRIER_P separately.
35412 2017-03-29 Andreas Schwab <schwab@suse.de>
35415 * calls.c (prepare_call_address): Convert funexp to Pmode before
35416 copying to temp reg.
35418 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35420 PR tree-optimization/80158
35421 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
35422 Handle possible future case of more than one alternate
35424 (replace_rhs_if_not_dup): Likewise.
35425 (replace_one_candidate): Likewise.
35427 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
35429 PR rtl-optimization/80193
35430 * ira.c (ira): Do not check allocation for LRA.
35432 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
35434 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
35435 (nvptx_output_simt_exit): Declare.
35436 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
35437 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
35438 (init_softstack_frame): Move initialization of crtl->is_leaf to...
35439 (nvptx_declare_function_name): ...here. Emit declaration of local
35440 memory space buffer for omp_simt_enter insn.
35441 (nvptx_output_unisimt_switch): New.
35442 (nvptx_output_softstack_switch): New.
35443 (nvptx_output_simt_enter): New.
35444 (nvptx_output_simt_exit): New.
35445 * config/nvptx/nvptx.h (struct machine_function): New fields
35446 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
35447 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
35448 (UNSPECV_SIMT_EXIT): Ditto.
35449 (omp_simt_enter_insn): New insn.
35450 (omp_simt_enter): New expansion.
35451 (omp_simt_exit): New insn.
35452 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
35454 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
35455 (expand_GOMP_SIMT_ENTER_ALLOC): New.
35456 (expand_GOMP_SIMT_EXIT): New.
35457 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
35458 (GOMP_SIMT_ENTER_ALLOC): Ditto.
35459 (GOMP_SIMT_EXIT): Ditto.
35460 * target-insns.def (omp_simt_enter): New insn.
35461 (omp_simt_exit): Ditto.
35462 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
35464 (lower_rec_simd_input_clauses): Implement SIMT privatization.
35465 (lower_rec_input_clauses): Likewise.
35466 (lower_lastprivate_clauses): Handle SIMT privatization.
35468 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
35469 (ompdevlow_adjust_simt_enter): New.
35470 (find_simtpriv_var_op): New.
35471 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
35472 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
35474 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
35475 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
35476 (copy_decl_for_dup_finish): Ditto.
35478 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
35480 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
35483 * config/i386/i386.c (ix86_option_override_internal): Always
35484 allow -mpreferred-stack-boundary=3 for 64-bit targets.
35486 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35488 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
35490 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35492 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
35493 mark new edge's irreducible flag accordign to it.
35494 (vect_do_peeling): Check loop preheader edge's irreducible flag
35495 and pass it to function slpeel_add_loop_guard.
35497 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
35499 PR tree-optimization/80218
35500 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35501 Update block frequencies and counts.
35503 2017-03-28 Richard Biener <rguenther@suse.de>
35505 PR tree-optimization/78644
35506 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
35507 of a simplification result we may not use it at all.
35509 2017-03-28 Richard Biener <rguenther@suse.de>
35512 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
35513 without arguments, generate default definition of a SSA name.
35515 2017-03-28 Richard Biener <rguenther@suse.de>
35517 PR middle-end/80222
35518 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
35519 TYPE_REF_CAN_ALIAS_ALL references.
35520 * fold-const.c (fold_indirect_ref_1): Likewise.
35522 2017-03-28 Martin Liska <mliska@suse.cz>
35525 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
35526 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
35528 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35529 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
35531 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
35532 (EXTRA_SPECS): Define.
35533 (SUBTARGET_EXTRA_SPECS): Likewise.
35534 (SUBTARGET_CPP_SPEC): Likewise.
35535 * config/arc/elf.h (EXTRA_SPECS): Renamed to
35536 SUBTARGET_EXTRA_SPECS.
35537 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
35539 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35541 * config/arc/simdext.md (vst64_insn): Update pattern.
35542 (vld32wh_insn): Likewise.
35543 (vld32wl_insn): Likewise.
35544 (vld64_insn): Likewise.
35545 (vld32_insn): Likewise.
35547 2017-03-28 Marek Polacek <polacek@redhat.com>
35550 * fold-const.c (fold_comparison): Use protected_set_expr_location
35551 instead of SET_EXPR_LOCATION.
35553 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
35555 * tree.c (add_expr): Avoid name lookup warning.
35557 2017-03-27 Jeff Law <law@redhat.com>
35559 PR tree-optimization/80216
35560 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
35561 function name. Limit recursion depth.
35562 (record_temporary_equivalences): Corresponding changes.
35564 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
35566 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
35569 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35572 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
35574 * cfgcleanup.c (reg_note_cfa_p): New array.
35575 (insns_have_identical_cfa_notes): New function.
35576 (old_insns_match_p): Don't cross-jump in between /f
35577 and non-/f instructions. If both i1 and i2 are frame related,
35578 verify all CFA notes, their order and content.
35580 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
35583 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
35584 HImode and SImode with zero extend to DImode to one insn.
35585 (bswap<mode>2_extenddi): Likewise.
35586 (bswapsi2_extenddi): Likewise.
35587 (bswaphi2_extendsi): Likewise.
35588 (bswaphi2): Combine bswap HImode and SImode into one insn.
35589 Separate memory insns from swapping register.
35590 (bswapsi2): Likewise.
35591 (bswap<mode>2): Likewise.
35592 (bswaphi2_internal): Delete, no longer used.
35593 (bswapsi2_internal): Likewise.
35594 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
35595 store, and gpr<-gpr swap insns.
35596 (bswap<mode>2_store): Likewise.
35597 (bswaphi2_reg): Register only splitter, combine with the splitter.
35598 (bswaphi2 splitter): Likewise.
35599 (bswapsi2_reg): Likewise.
35600 (bswapsi2 splitter): Likewise.
35601 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
35602 the insns into load, store, and register/register insns.
35603 (bswapdi2_ldbrx): Likewise.
35604 (bswapdi2_load): Likewise.
35605 (bswapdi2_store): Likewise.
35606 (bswapdi2_reg): Likewise.
35608 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
35610 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
35611 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
35613 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
35616 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
35618 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35619 special handling for target option conflicts between dform
35620 options (-mpower9-dform, -mpower9-dform-vector,
35621 -mpower9-dform-scalar) and -mno-direct-move.
35623 2017-03-27 Richard Biener <rguenther@suse.de>
35625 PR tree-optimization/80181
35626 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
35628 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35630 * config/arc/predicates.md (move_double_src_operand): Replace the
35631 call to move_double_src_operand with a call to address_operand.
35633 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35635 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
35636 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
35637 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
35639 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35641 * config/arc/predicates.md (long_immediate_loadstore_operand):
35642 Consider scaled addresses cases.
35644 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35646 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
35647 restored when in interrupt.
35648 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
35649 doesn't have delay slot.
35651 2017-03-27 Richard Biener <rguenther@suse.de>
35654 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
35655 inlined thunk clones.
35657 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35660 * asan.c (instrument_derefs): Copy over last operand from
35661 original COMPONENT_REF to the new COMPONENT_REF with
35662 DECL_BIT_FIELD_REPRESENTATIVE.
35663 * ubsan.c (instrument_object_size): Likewise.
35665 2017-03-27 Richard Biener <rguenther@suse.de>
35667 PR tree-optimization/80170
35668 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
35669 sure DR/SCEV didnt fold in constants we do not see when looking
35670 at the reference base alignment.
35672 2017-03-27 Richard Biener <rguenther@suse.de>
35674 PR middle-end/80171
35675 * gimple-fold.c (fold_ctor_reference): Properly guard against
35676 NULL return value from canonicalize_constructor_val.
35678 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
35681 * config/i386/i386.c (ix86_expand_builtin)
35682 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
35683 flags reg setting and flags reg using instructions.
35684 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
35685 clobbering instructions to zero extend op2.
35687 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
35689 * doc/install.texi (Configuration) <--with-aix-soname>:
35690 Update link to AIX ld.
35692 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
35694 PR rtl-optimization/80160
35695 PR rtl-optimization/80159
35696 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
35697 reg_alternate_class into account.
35699 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
35702 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
35703 to consider in curr_insn_transform.
35705 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35707 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
35708 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
35709 and emit_mode_inner.
35711 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35713 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
35714 argument to the overloaded builtin variants. Use the new flag to
35715 deprecate certain builtin variants.
35716 * config/s390/s390-builtin-types.def: Add new builtin types.
35717 * config/s390/s390-builtins.h: Support new flags field for
35718 overloaded builtins.
35719 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
35720 (s390_macro_to_expand): Enable vector float data type.
35721 (s390_cpu_cpp_builtins_internal): Indicate support of the new
35722 builtins by incrementing the __VEC__ version number.
35723 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
35725 (s390_resolve_overloaded_builtin): Emit error messages depending
35726 on the builtin flags.
35727 * config/s390/s390.c (s390_expand_builtin): Support additional
35728 flags argument. Change error message to match the messages
35729 emitted in s390-c.c.
35730 * config/s390/s390.md: New UNSPEC_* constants.
35731 (op_type): Add new instruction types.
35732 * config/s390/vecintrin.h: Add new builtins and test data class
35734 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
35735 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
35736 (VEC_INEXACT, VEC_NOINEXACT): New constants.
35737 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
35738 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
35739 ("vec_mergel<mode>"): V_HW -> VEC_HW.
35741 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
35742 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
35743 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
35744 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
35746 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
35747 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
35748 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
35749 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
35751 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
35752 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
35753 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
35754 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
35755 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
35756 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
35757 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
35759 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
35760 ("vec_scatter_element<V_HW_4:mode>_DI")
35761 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
35762 ("vec_fpint<mode>", "vflls")
35763 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
35764 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
35765 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
35766 ("*vec_cmphe<mode>_cc"): ... these.
35768 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
35769 mode constant instead of magic value.
35771 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35773 * config/s390/s390.c (s390_expand_vec_compare): Support other
35774 vector floating point modes than just V2DF.
35775 (s390_expand_vcond): Likewise.
35776 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35777 (s390_cannot_change_mode_class): Prevent mode changes between TF
35778 and V1TF in vector registers.
35779 * config/s390/s390.md (DF, SF): New mode attributes.
35780 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35781 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35782 SFmode support for VRs.
35783 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35785 (VFT, VF_HW): New mode iterators.
35786 (vw, sdx): New mode attributes.
35787 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35788 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35789 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35790 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35791 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35792 also the new vector floating point modes. Renaming to ...
35794 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35795 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35796 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35797 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35798 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35799 ("vec_unordered<mode>"): ... these.
35801 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35802 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35803 ("*vec_extendv2df"): New insn definitions.
35805 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35807 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35808 ("mulditi3_2", "*muldi3_sign"): New patterns.
35809 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35810 rename the pattern definition.
35812 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35814 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35816 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35818 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35820 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35821 instruction if possible.
35822 * config/s390/vector.md (vec_halfnumelts): New mode
35824 ("*vec_vllezlf<mode>"): New pattern.
35826 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35828 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35829 ("popcountv4si2", "popcountv2di2"): Rename to ...
35830 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35831 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35833 ("popcount<mode>2_vxe"): New pattern.
35835 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35837 * common/config/s390/s390-common.c (processor_flags_table): Add
35839 * config.gcc: Add arch12.
35840 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35841 Default to arch12 for unknown CPU model numbers.
35842 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35843 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35844 PROCESSOR_max sanity check.
35845 * config/s390/s390-opts.h (enum processor_type): Add
35847 * config/s390/s390.c (processor_table): Add arch12.
35848 (s390_expand_builtin): Add check for B_VXE flag.
35849 (s390_issue_rate): Add PROCESSOR_ARCH12.
35850 (s390_get_sched_attrmask): Likewise.
35851 (s390_get_unit_mask): Likewise.
35852 (s390_sched_score): Enable z13 scheduling for arch12.
35853 (s390_sched_reorder): Likewise.
35854 (s390_sched_variable_issue): Likewise.
35855 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35857 (s390_tune_attr): Use z13 scheduling also for arch12.
35858 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35859 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35860 (TARGET_VXE_P): New macros.
35861 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35862 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35863 * config/s390/s390.opt: Add arch12 as processor_type.
35865 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35867 * config/s390/s390.md
35868 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35869 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35870 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35872 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35873 Rename expanders to ...
35875 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35876 ("fixuns_truncdddi2_emu"): ... these.
35878 ("fixuns_trunc<mode>si2_emu"): New expander.
35880 ("*fixuns_truncdfdi2_z13"): Rename to ...
35881 ("*fixuns_truncdfdi2_vx"): ... this.
35883 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35885 * config/s390/2964.md: Remove the single element vector compare
35886 instructions which are no longer used.
35887 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35889 (s390_canonicalize_comparison): Remove handling of DFmode
35891 (s390_expand_vec_compare_scalar): Remove function.
35892 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35893 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35895 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35897 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35899 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35901 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35902 will anyway by matched by mov<mode>_64dfp.
35904 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35906 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35907 vlef/vstef. Add missing operand to vleif.
35909 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35911 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35912 pair for all vector types with 64 bit elements.
35913 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35914 * config/s390/vector.md (V_HW_64): ... here.
35915 (V_128_NOSINGLE): New mode iterator.
35916 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35917 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35918 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35919 ("*vec_load_pairv2di"): Change to ...
35920 ("*vec_load_pair<mode>"): ... this one.
35922 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35924 * config/s390/constraints.md: Add comments.
35925 (jKK): Reject element sizes > 8 bytes.
35926 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35928 * config/s390/s390.md: Add the s_operand checks formerly in
35929 s390_split_ok_p to various splitters where they are still
35931 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35932 for 128 bit vectors. Plus two splitters.
35934 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35936 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35939 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35942 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35943 error if the boundary argument is not constant.
35945 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35947 PR rtl-optimization/80112
35948 * loop-doloop.c (doloop_condition_get): Don't check condition
35949 if cmp isn't SET with IF_THEN_ELSE src.
35951 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35953 PR tree-optimization/80158
35954 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35955 replacing a candidate statement, also replace it for the
35956 candidate's alternate interpretation.
35957 (replace_rhs_if_not_dup): Likewise.
35958 (replace_one_candidate): Likewise.
35960 2017-03-24 Richard Biener <rguenther@suse.de>
35962 PR tree-optimization/80167
35963 * graphite-isl-ast-to-gimple.c
35964 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35966 (translate_isl_ast_to_gimple::get_rename): Likewise.
35968 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35970 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35971 handling of certain combinations of target options, including the
35972 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35973 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35975 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35978 * config/arm/arm.md (*load_multiple): Add reload_completed to
35979 matching condition.
35981 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35982 Richard Biener <rguenth@suse.de>
35984 PR tree-optimization/79908
35985 PR tree-optimization/80136
35986 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35987 been cast away, gimplify_and_add suffices.
35989 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35991 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35993 2017-03-23 Richard Biener <rguenther@suse.de>
35995 PR tree-optimization/80032
35996 * gimplify.c (gimple_push_cleanup): Forced unconditional
35997 cleanups still have to go to the conditional_cleanups
36000 2017-03-22 Jakub Jelinek <jakub@redhat.com>
36002 PR tree-optimization/80072
36003 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
36005 (next_operand_entry_id): Change type to unsigned int.
36006 (sort_by_operand_rank): Make sure to return the right return value
36007 even if unsigned fields are bigger than INT_MAX.
36008 (struct oecount): Change cnt and id type to unsigned int.
36009 (oecount_hasher::equal): Formatting fix.
36010 (oecount_cmp): Make sure to return the right return value
36011 even if unsigned fields are bigger than INT_MAX.
36012 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
36015 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
36016 TREE_READONLY on result if writing it more than once.
36019 * doc/invoke.texi (-fsanitize=thread): Document that with
36020 -fnon-call-exceptions atomics are not able to throw
36024 * tsan.c: Include tree-eh.h.
36025 (instrument_builtin_call): Call maybe_clean_eh_stmt or
36026 maybe_clean_or_replace_eh_stmt where needed.
36027 (instrument_memory_accesses): Add cfg_changed argument.
36028 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
36029 if it returned true.
36030 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
36032 PR rtl-optimization/63191
36033 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
36034 wrapper function, moved the whole old content into ...
36035 (ix86_delegitimize_address_1): ... this. New inline function.
36036 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
36037 true as last argument instead of ix86_delegitimize_address.
36039 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
36041 * config/aarch64/aarch64.c (generic_branch_cost): Copy
36042 cortexa57_branch_cost.
36044 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
36046 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
36048 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36051 * doc/md.texi (Constraints): Document wA constraint.
36052 * config/rs6000/constraints.md (wA): New.
36053 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
36054 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
36055 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
36056 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
36058 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
36061 * gimplify.c (is_oacc_declared): New function.
36062 (oacc_default_clause): Use it to set default flags for acc declared
36063 variables inside parallel regions.
36064 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
36065 declared variables.
36066 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
36067 declare attribute to any decl as necessary.
36069 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
36072 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
36073 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
36074 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
36075 (arm_arch_lpae): This.
36076 * config/arm/arm.c (arm_arch7ve): Rename into ...
36077 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
36078 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
36081 2017-03-22 Martin Liska <mliska@suse.cz>
36084 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
36085 error message instead of an ICE.
36087 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36089 * doc/extend.texi (6.11 Additional Floating Types): Revise.
36091 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
36093 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36095 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
36098 2017-03-21 Martin Sebor <msebor@redhat.com>
36100 * doc/extend.texi: Use "cannot" instead of "can't."
36101 * doc/hostconfig.texi: Same.
36102 * doc/install.texi: Same.
36103 * doc/invoke.texi: Same.
36104 * doc/loop.texi: Same.
36105 * doc/md.texi: Same.
36106 * doc/objc.texi: Same.
36107 * doc/rtl.texi: Same.
36108 * doc/tm.texi: Same.
36109 * doc/tm.texi.in: Same.
36110 * doc/trouble.texi: Same.
36112 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
36115 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
36116 (collect_checksum_attributes): Set it.
36117 (die_checksum_ordered): Use it.
36119 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36121 PR tree-optimization/79908
36122 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
36123 change: For a VA_ARG whose LHS has been cast away, use
36124 force_gimple_operand to construct the side effects.
36126 2017-03-21 David Malcolm <dmalcolm@redhat.com>
36128 PR translation/80001
36129 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
36130 more amenable to translation.
36131 (oacc_loop_auto_partitions): Likewise.
36133 2017-03-21 Marek Polacek <polacek@redhat.com>
36134 Martin Sebor <msebor@redhat.com>
36136 PR tree-optimization/80109
36137 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
36138 on INTEGRAL_TYPE_P.
36140 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36141 Segher Boessenkool <segher@kernel.crashing.org>
36144 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
36145 check reg_used_between_p between insn and one of succ or succ2
36146 depending on if succ is artificial insn not inserted into insn
36149 2017-03-21 Martin Liska <mliska@suse.cz>
36151 PR gcov-profile/80081
36152 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
36153 * doc/gcc.texi: Include gcov-dump stuff.
36154 * doc/gcov-dump.texi: New file.
36156 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
36158 PR rtl-optimization/79150
36159 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
36160 conditional jump, if the jump is the last insn of the loop.
36162 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36163 Richard Biener <rguenth@suse.de>
36165 PR tree-optimization/79908
36166 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
36167 been cast away, use force_gimple_operand to construct the side
36170 2017-03-21 Martin Liska <mliska@suse.cz>
36172 PR libfortran/79956
36173 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
36176 2017-03-21 Brad Spengler <spender@grsecurity.net>
36179 * plugin.c (htab_hash_plugin): New function.
36180 (add_new_plugin): Use it and adjust.
36181 (parse_plugin_arg_opt): Adjust.
36182 (init_one_plugin): Likewise.
36184 2017-03-21 Richard Biener <rguenther@suse.de>
36186 PR tree-optimization/80032
36187 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
36188 if set force the cleanup to happen unconditionally.
36189 (gimplify_target_expr): Push inserted clobbers with force_uncond
36190 to avoid them being removed by control-dependent DCE.
36192 2017-03-21 Richard Biener <rguenther@suse.de>
36194 PR tree-optimization/80122
36195 * tree-inline.c (copy_bb): Do not expans va-arg packs or
36196 va_arg_pack_len when the inlined call stmt requires pack
36198 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
36200 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36203 * tsan.c (instrument_builtin_call): If the memory model argument
36204 is not a constant, assume it is valid.
36207 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
36210 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36212 PR rtl-optimization/79910
36213 * combine.c (can_combine_p): Do not allow combining an I0 or I1
36214 if its dest is used by an insn before I2 (other than the combined
36215 insns themselves, which are properly handled already).
36217 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36220 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36222 * combine.c (record_used_regs): New static function.
36223 (try_combine): Handle situations where there is an additional
36224 instruction between I2 and I3 which needs to have a LOG_LINK
36228 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36230 * combine.c (try_combine): Delete redundant i1 test. Call
36231 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36233 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36236 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
36237 alternatives 13/14.
36239 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36241 PR tree-optimization/80054
36242 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
36243 the optimization if a PHI or any of its arguments is not dominated
36244 by the candidate's basis. Use gphi* rather than gimple* as
36246 (replace_profitable_candidates): Clean up a gimple* variable that
36247 should be a gphi* variable.
36249 2017-03-20 Martin Sebor <msebor@redhat.com>
36252 * doc/extend.texi (attribute constructor): Document present limitation.
36254 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
36257 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
36258 __POWER9_VECTOR__ #ifdef control, change template definition to
36259 use Power9-specific built-in function.
36260 (vec_any_eq): Likewise.
36261 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
36262 to control outcomes from this test.
36263 (vector_ae_<mode>p): For VEC_F modes, likewise.
36265 2017-03-20 Ian Lance Taylor <iant@google.com>
36267 * config/i386/i386.c (ix86_function_regparm): Save an extra
36268 register for -fsplit-stack with DECL_STATIC_CHAIN.
36270 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36273 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
36274 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
36276 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36278 * config/riscv/riscv.c (riscv_print_operand): Use "fence
36280 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
36283 2017-03-20 Marek Polacek <polacek@redhat.com>
36286 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
36288 2017-03-20 Richard Biener <rguenther@suse.de>
36290 PR tree-optimization/80113
36291 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
36292 allocate extra SSA name for PHI def.
36293 (add_close_phis_to_outer_loops): Likewise.
36294 (add_close_phis_to_merge_points): Likewise.
36295 (copy_loop_close_phi_args): Likewise.
36296 (copy_cond_phi_nodes): Likewise.
36298 2017-03-20 Martin Liska <mliska@suse.cz>
36300 PR middle-end/79753
36301 * tree-chkp.c (chkp_build_returned_bound): Do not build
36302 returned bounds for a LHS that's not a BOUNDED_P type.
36304 2017-03-20 Martin Liska <mliska@suse.cz>
36308 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
36309 COMPLEX_CST and VECTOR_CST.
36311 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36314 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
36315 target operand. A new splitter adds the clobber statement in case
36316 the target operand is dead anyway.
36318 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
36320 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
36321 to age-old versions of binutils and glibc.
36323 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
36325 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
36327 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36329 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
36331 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36333 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
36334 requirement for binutils 2.13.
36336 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36338 * combine.c (try_combine): Delete redundant i1 test. Call
36339 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36341 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
36343 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
36344 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
36346 <riscv64-*-elf>: Re-arrange section
36347 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
36348 <riscv32-*-linux>: Likewise.
36349 <riscv64-*-elf>: Likewise
36350 <riscv64-*-linux>: Likewise.
36352 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
36355 * aarch64.opt(verbose-cost-dump): Fix typo.
36357 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
36360 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
36361 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
36363 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36365 * reload.c (find_reloads): When reloading a nonoffsettable address,
36366 use RELOAD_OTHER for it and its address reloads.
36368 PR rtl-optimization/79910
36369 * combine.c (record_used_regs): New static function.
36370 (try_combine): Handle situations where there is an additional
36371 instruction between I2 and I3 which needs to have a LOG_LINK
36374 2017-03-17 Jeff Law <law@redhat.com>
36376 PR tree-optimization/71437
36377 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
36378 conditional in the hash table first.
36379 (vrp_dom_walker::before_dom_children): Extract condition from
36380 ASSERT_EXPR. Record condition, its inverion and any implied
36381 conditions as well.
36383 2017-03-17 Marek Polacek <polacek@redhat.com>
36384 Markus Trippelsdorf <markus@trippelsdorf.de>
36386 PR tree-optimization/80079
36387 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
36390 2017-03-17 Richard Biener <rguenther@suse.de>
36392 PR middle-end/80075
36393 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
36394 Properly verify the LHS before the RHS possibly claims to be
36396 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
36399 2017-03-17 Martin Jambor <mjambor@suse.cz>
36401 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
36402 (List of -O2 options): Likewise.
36403 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
36406 2017-03-17 Tom de Vries <tom@codesourcery.com>
36408 * gcov-dump.c (print_usage): Print bug_report_url.
36410 2017-03-17 Richard Biener <rguenther@suse.de>
36412 PR middle-end/80050
36413 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
36414 (parser::peek): Likewise.
36416 2017-03-17 Richard Biener <rguenther@suse.de>
36418 PR tree-optimization/80048
36419 * sese.c (free_sese_info): Properly release rename_map and
36420 copied_bb_map elements.
36422 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
36424 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
36425 Add linked-list forward and backlinks. Insert on
36426 construction, remove on destruction.
36427 (class pass_store_merging): Add m_stores_head field.
36428 (pass_store_merging::terminate_and_process_all_chains):
36429 Iterate over m_stores_head list.
36430 (pass_store_merging::terminate_all_aliasing_chains):
36432 (pass_store_merging::execute): Check for debug stmts first.
36433 Push new chains onto the m_stores_head stack.
36435 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
36438 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
36439 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
36440 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
36442 2017-03-16 Jeff Law <law@redhat.com>
36444 PR tree-optimization/71437
36445 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
36446 member function. Implementation moved into after_dom_children
36447 member function and into the threader's thread_outgoing_edges
36449 (dom_opt_dom_walker::after_dom_children): Simplify by moving
36450 some code into new thread_outgoing_edges.
36451 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
36452 definition. Simplify marker handling (do it here). Assume we always
36453 have the available expression and the const/copies tables.
36454 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
36456 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
36457 * tree-vrp.c (equiv_stack): No longer file scoped.
36458 (vrp_dom_walker): New class.
36459 (vrp_dom_walker::before_dom_children): New member function.
36460 (vrp_dom_walker::after_dom_children): Likewise.
36461 (identify_jump_threads): Setup domwalker. Use it rather than
36462 walking edges in a random order by hand. Simplify setup/finalization.
36463 (finalize_jump_threads): Remove.
36464 (vrp_finalize): Do not call identify_jump_threads here.
36465 (execute_vrp): Do it here instead and call thread_through_all_blocks
36468 PR tree-optimization/71437
36469 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
36471 (simplify_stmt_for_jump_threading): Add basic_block argument. All
36473 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
36474 (dom_opt_dom_walker::thread_across_edge): Remove
36475 handle_dominating_asserts argument. All callers changed.
36476 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
36477 changes. Remove calls to lhs_of_dominating_assert. Other
36478 uses of handle_dominating_asserts turn into unconditional code
36479 (simplify_control_stmt_condition_1): Likewise.
36480 (simplify_control_stmt_condition): Likewise.
36481 (thread_through_normal_block, thread_across_edge): Likewise.
36482 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
36483 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
36484 object if it is not an SSA_NAME.
36485 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
36486 before calling into the VRP specific simplifiers.
36487 (identify_jump_threads): Remove handle_dominating_asserts
36490 2017-03-16 Jakub Jelinek <jakub@redhat.com>
36493 * tree-diagnostic.c (default_tree_printer): No longer static.
36494 * tree-diagnostic.h (default_tree_printer): New prototype.
36496 2017-03-16 Tamar Christina <tamar.christina@arm.com>
36498 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
36499 Change ins into fmov.
36501 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36503 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
36504 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
36505 Use h_con constraint for operand 1.
36506 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
36507 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
36509 2017-03-15 Jeff Law <law@redhat.com>
36511 PR tree-optimization/71437
36512 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
36513 (record_temporary_equivalences): Use it.
36515 PR tree-optimization/71437
36516 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
36517 tree-ssa-scopedtables.
36518 (lookup_avail_expr, build_and_record_new_cond): Likewise.
36519 (record_conditions, record_cond, vuse_eq): Likewise.
36520 (record_edge_info): Adjust to API tweak of record_conditions.
36521 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
36522 (record_temporary_equivalences, optimize_stmt): Likewise.
36523 (eliminate_redundant_computations): Likewise.
36524 (record_equivalences_from_stmt): Likewise.
36525 * tree-ssa-scopedtables.c: Include options.h and params.h.
36526 (vuse_eq): New function, moved from tree-ssa-dom.c
36527 (build_and_record_new_cond): Likewise.
36528 (record_conditions): Likewise. Accept vector of conditions rather
36529 than edge_equivalence structure for first argument.
36530 for the first argument.
36531 (avail_exprs_stack::lookup_avail_expr): New member function, moved
36532 from tree-ssa-dom.c.
36533 (avail_exprs_stack::record_cond): Likewise.
36534 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
36535 from tree-ssa-dom.c.
36536 (avail_exprs_stack): Add new member functions lookup_avail_expr
36538 (record_conditions): Declare.
36540 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
36543 * lra-constraints.c (process_alt_operands): Increase reject for
36544 reloading an input/output operand.
36546 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
36549 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
36550 insns to convert from signed/unsigned char/short to IEEE 128-bit
36552 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
36554 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
36557 * config/i386/i386.c (ix86_vector_duplicate_value): Create
36558 subreg of inner mode for values already in registers.
36560 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
36562 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
36563 iteration reg is used after the loop.
36565 2017-03-14 Martin Sebor <msebor@redhat.com>
36567 PR tree-optimization/79800
36568 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
36569 precision in negative-positive range.
36570 (format_floating): Call non-const overload with adjusted precision.
36572 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
36575 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
36578 2017-03-14 Martin Sebor <msebor@redhat.com>
36580 PR middle-end/80020
36581 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
36582 * builtins.def (aligned_alloc): Use it.
36585 * Makefile.in (GTFILES): Add calls.c.
36586 * calls.c: Include "gt-calls.h".
36588 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
36590 PR rtl-optimization/79728
36591 * regs.h (struct target_regs): New field
36592 x_contains_allocatable_regs_of_mode.
36593 (contains_allocatable_regs_of_mode): New macro.
36594 * reginfo.c (init_reg_sets_1): Initialize it, and change
36595 contains_reg_of_mode so it includes global regs as well.
36596 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
36597 rather than contains_regs_of_mode.
36599 2017-03-14 Martin Liska <mliska@suse.cz>
36601 * doc/invoke.texi: Document options that can't be combined with
36602 -fcheck-pointer-bounds.
36604 2017-03-14 Martin Liska <mliska@suse.cz>
36606 PR middle-end/79831
36607 * doc/invoke.texi (-Wchkp): Document the option.
36609 2017-03-14 Martin Liska <mliska@suse.cz>
36611 * Makefile.in: Install gcov-dump.
36613 2017-03-14 Martin Liska <mliska@suse.cz>
36615 * multiple_target.c (expand_target_clones): Bail out for
36616 an invalid attribute.
36618 2017-03-14 Richard Biener <rguenther@suse.de>
36620 * alias.c (struct alias_set_entry): Pack properly.
36621 * cfgloop.h (struct loop): Likewise.
36622 * cse.c (struct set): Likewise.
36623 * ipa-utils.c (struct searchc_env): Likewise.
36624 * loop-invariant.c (struct invariant): Likewise.
36625 * lra-remat.c (struct cand): Likewise.
36626 * recog.c (struct change_t): Likewise.
36627 * rtl.h (struct address_info): Likewise.
36628 * symbol-summary.h (function_summary): Likewise.
36629 * tree-loop-distribution.c (struct partition): Likewise.
36630 * tree-object-size.c (struct object_size_info): Likewise.
36631 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
36632 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
36633 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
36634 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
36635 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
36636 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
36637 (struct _stmt_vec_info): Likewise.
36639 2017-03-14 Martin Liska <mliska@suse.cz>
36642 * multiple_target.c (create_dispatcher_calls): Check that
36643 a target can create a function dispatcher.
36645 2017-03-14 Martin Liska <mliska@suse.cz>
36648 * multiple_target.c (expand_target_clones): Drop local.local
36649 flag for default implementation.
36651 2017-03-14 Richard Biener <rguenther@suse.de>
36653 PR tree-optimization/80030
36654 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
36656 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
36658 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
36659 gcc_fallthrough() instead of __attribute__((fallthrough));
36661 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36663 * doc/gcc.texi: Remove "up" link to (DIR).
36664 * doc/gccint.texi: Ditto.
36666 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36668 * doc/install.texi (Specific) <avr>: Remove reference to
36671 2017-03-13 Jeff Law <law@redhat.com>
36673 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
36674 attribute rather than comments.
36676 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
36677 match_scratch operand is highest.
36679 2017-03-13 Martin Liska <mliska@suse.cz>
36681 PR middle-end/78339
36682 * ipa-pure-const.c (warn_function_noreturn): If the declarations
36683 is a CHKP clone, use original declaration.
36685 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36687 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
36688 (arc_conditional_register_usage): Use a different allocation order
36689 when optimizing for size.
36690 * common/config/arc/arc-common.c (arc_option_optimization_table):
36691 Section anchors default on when optimizing for size.
36693 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36695 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
36697 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36699 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
36700 * config/arc/arc.md (cpu_facility): Add cd variant.
36701 (*movqi_insn): Add code density variant.
36702 (*movhi_insn): Likewise.
36703 (*movqi_insn): Likewise.
36704 (*addsi3_mixed): Likewise.
36705 (subsi3_insn): Likewise.
36707 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36709 * config/arc/arc.md (movsi_cond_exec): Update constraint.
36711 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36713 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
36714 expressions with MINUS and UNARY ops.
36716 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36719 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
36721 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
36722 between vec_select and vector argument.
36723 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
36724 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
36725 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
36726 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
36727 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
36728 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
36730 2017-03-13 Richard Biener <rguenther@suse.de>
36733 * params.def (vect-max-peeling-for-alignment): Fix typo.
36735 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36737 * doc/install.texi (Specific) <mips-*-*>: Remove description of
36738 issue that only occurred with binutils below 2.18.
36740 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36742 * doc/install.texi (Specific) <cris-axis-elf>: No longer
36743 refer to binutils 2.11/2.12 minimum.
36745 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36747 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
36748 ftp.kernel.org and simplify binutils requirement.
36750 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
36752 * doc/invoke.texi (Warning Options): Fix spelling of link-time
36754 (Optimize Options): Ditto. Also remove redundancy.
36756 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36758 PR translation/79848
36759 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
36761 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
36762 to G_ to avoid double translation.
36764 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36766 PR translation/79923
36767 * auto-profile.c (get_combined_location): Convert leading
36768 character of diagnostics to lower case and remove trailing period.
36769 (read_profile): Likewise for various diagnostics.
36770 * config/arm/arm.c (arm_option_override): Remove trailing period
36771 from various diagnostics.
36772 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
36773 (msp430_expand_delay_cycles): Likewise.
36775 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36778 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36779 full command-line argument, rather than just "str".
36780 (aarch64_validate_march): Likewise.
36781 (aarch64_validate_mtune): Likewise.
36783 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36785 PR rtl-optimization/78911
36786 * lra-assigns.c (must_not_spill_p): New function.
36787 (spill_for): Use it.
36789 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36791 PR tree-optimization/79981
36792 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36793 ATOMIC_COMPARE_EXCHANGE ifn result.
36794 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36795 IFN_ATOMIC_COMPARE_EXCHANGE.
36797 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36800 * opts.c (parse_sanitizer_options): Add missing question mark to
36801 "did you mean" message.
36803 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36805 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36807 (VMULEUH_UNS): Likewise.
36808 (VMULOUB_UNS): Likewise.
36809 (VMULOUH_UNS): Likewise.
36810 * config/rs6000/rs6000.c (builtin_function_type): Remove
36811 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36813 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36816 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36817 x with result of extra_parsing_for_operand_code_0.
36818 (function_reader::extra_parsing_for_operand_code_0): Convert
36819 return type from void to rtx, returning x. When reading
36820 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36821 larger size containing struct block_symbol.
36823 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36825 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36826 -mfloat128-hardware without -m64.
36828 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36831 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36832 entries to the case statement that marks unsigned arguments to
36833 overloaded functions.
36835 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36837 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36838 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36840 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36843 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36844 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36846 2017-03-10 Martin Liska <mliska@suse.cz>
36850 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36851 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36854 2017-03-10 Olivier Hainque <hainque@adacore.com>
36856 * tree-switch-conversion (array_value_type): Start by resetting
36857 candidate type to it's main variant.
36859 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36861 PR rtl-optimization/79909
36862 * combine.c (try_combine): Use simplify_replace_rtx on individual
36863 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36864 of the whole CALL_INSN_FUNCTION_USAGE.
36866 PR tree-optimization/79972
36867 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36868 get_range_info on SSA_NAMEs. Formatting fixes.
36870 2017-03-10 Richard Biener <rguenther@suse.de>
36871 Jakub Jelinek <jakub@redhat.com>
36873 PR tree-optimization/77975
36874 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36875 edge to be constant.
36876 (get_val_for): For constant x return it. Formatting fix.
36877 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36878 would use the same bases as the current one.
36880 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36882 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36883 instead of vec_select for V1TImode.
36884 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36886 (VSX_LE_128): Add V1TI to this mode iterator.
36887 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36888 (*vsx_le_perm_store_<mode>): Likewise.
36889 (pre-reload splitter for VSX stores): Likewise.
36890 (post-reload splitter for VSX stores): Likewise.
36891 (*vsx_xxpermdi2_le_<mode>): Likewise.
36892 (*vsx_lxvd2x2_le_<mode>): Likewise.
36893 (*vsx_stxvd2x2_le_<mode>): Likewise.
36895 2017-03-09 Michael Eager <eager@eagercon.com>
36897 Correct failures with --enable-checking=yes,rtl.
36899 * config/microblaze/microblaze.c (microblaze_expand_shift):
36900 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36901 test for const0_rtx.
36902 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36903 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36905 2017-03-09 Richard Biener <rguenther@suse.de>
36907 PR tree-optimization/79977
36908 * graphite-scop-detection.c (scop_detection::merge_sese):
36909 Handle the case of extra exits to blocks dominating the entry.
36911 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36913 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36916 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36918 PR rtl-optimization/79949
36919 * lra-constraints.c (process_alt_operands): Check memory when
36920 trying to predict a cycle. Print about the overall increase.
36922 2017-03-09 Richard Biener <rguenther@suse.de>
36924 PR middle-end/79971
36925 * gimple-expr.c (useless_type_conversion_p): Preserve
36926 TYPE_SATURATING for fixed-point types.
36928 2017-03-09 Richard Biener <rguenther@suse.de>
36931 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36932 alignment of BLKmode params.
36934 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36937 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36938 (VALL_NO_V2Q): Likewise.
36940 * config/aarch64/aarch64-simd.md
36941 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36943 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36944 VALL_NO_V2Q mode iterator.
36945 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36947 2017-03-09 Martin Liska <mliska@suse.cz>
36949 PR tree-optimization/79631
36950 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36951 tree_int_cst_sign_bit just for INTEGER constants.
36953 2017-03-09 Martin Liska <mliska@suse.cz>
36957 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36960 2017-03-09 Marek Polacek <polacek@redhat.com>
36963 * tree.c (inchash::add_expr): Handle TREE_VEC.
36965 2017-03-09 Martin Liska <mliska@suse.cz>
36968 (chkp_narrow_size_and_offset): New function.
36969 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36970 (void chkp_parse_bit_field_ref): New function.
36971 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36972 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36974 2017-03-09 Martin Liska <mliska@suse.cz>
36977 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36978 (chkp_find_bounds_1): Remove gcc_unreachable.
36980 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36983 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36984 BUILT_IN_SYNC*, determine the access type from the size suffix and
36985 always build a MEM_REF with that type. Handle forgotten
36986 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36989 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36990 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36991 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36992 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36993 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36994 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36995 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36996 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36997 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36998 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36999 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
37000 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
37001 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
37002 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
37003 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
37004 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
37005 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
37006 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
37007 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
37008 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
37009 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
37010 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
37011 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
37012 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
37013 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
37014 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
37015 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
37016 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
37017 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
37018 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
37019 definitions outside of __OPTIMIZE__ guarded section.
37022 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
37023 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
37024 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
37025 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
37028 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37030 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
37031 ("vfenez<mode>"): Add missing constraints.
37033 2017-03-08 Martin Sebor <msebor@redhat.com>
37036 * config/nds32/nds32.c (nds32_option_override):
37037 Fix misspelled diagnostic.
37039 2017-03-08 Jakub Jelinek <jakub@redhat.com>
37042 * gimplify.c (gimplify_omp_for): Replace index var in outer
37043 taskloop statement with an artificial variable and add
37044 OMP_CLAUSE_PRIVATE clause for it.
37046 2017-03-08 Richard Biener <rguenther@suse.de>
37048 PR tree-optimization/79955
37049 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
37050 for accesses that are completely outside of the variable.
37052 2017-03-08 Andrew Haley <aph@redhat.com>
37054 PR tree-optimization/79943
37055 * tree-ssa-loop-split.c (compute_new_first_bound): When
37056 calculating the new upper bound, (END-BEG) should be added, not
37059 2017-03-08 Jakub Jelinek <jakub@redhat.com>
37061 * config/avr/avr.md (setmemhi): Make sure match_dup
37062 operand number comes before match_scratch.
37064 2017-03-08 Richard Biener <rguenther@suse.de>
37066 PR tree-optimization/79920
37067 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
37068 with ncopies == 1 to ...
37069 (vect_transform_slp_perm_load): ... here. Properly compute
37070 all element loads by iterating VF times over the group. Do
37071 not handle ncopies (computed in a broken way) in
37072 vect_create_mask_and_perm.
37074 2017-03-08 Jakub Jelinek <jakub@redhat.com>
37077 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
37078 is a uniform vector, use uniform_vector_p return value instead of
37079 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
37081 2017-03-07 Marek Polacek <polacek@redhat.com>
37083 PR middle-end/79809
37084 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
37085 (alloca_call_type): Likewise.
37087 2017-03-07 Martin Liska <mliska@suse.cz>
37089 * gcov.c (process_args): Put comment to correct location.
37091 2017-03-07 Martin Liska <mliska@suse.cz>
37093 PR middle-end/68270
37094 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
37095 Use array_at_struct_end_p instead of DECL_CHAIN (field).
37096 (chkp_narrow_bounds_for_field): Likewise.
37097 (chkp_parse_array_and_component_ref): Pass one more argument to
37100 2017-03-07 Richard Biener <rguenther@suse.de>
37102 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
37105 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
37107 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
37108 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
37110 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37113 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
37114 to end of description.
37115 (PARAM_MAX_STORES_TO_MERGE): Likewise.
37117 2017-03-07 Jakub Jelinek <jakub@redhat.com>
37119 PR rtl-optimization/79901
37120 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
37122 (*avx512f_<code><mode>3<mask_name>): ... this.
37123 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
37124 iterator instead of VI8_AVX2_AVX512BW.
37126 PR rtl-optimization/79901
37127 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
37128 min/max expander, expand it using expand_vec_cond_expr.
37131 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
37134 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37137 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
37138 to void * for PCH reasons.
37139 * dwarf2out.c (output_loc_operands, output_die): Cast
37140 v.val_vec.array to unsigned char *.
37142 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
37145 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
37148 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
37150 PR rtl-optimization/79571
37151 * lra-constraints.c (process_alt_operands): Calculate static
37152 reject and subtract it from overall when only addresses will be
37155 2017-03-06 Julia Koval <julia.koval@intel.com>
37158 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
37159 incoming stack boundary to 128 for 64-bit targets.
37161 2017-03-06 Richard Biener <rguenther@suse.de>
37163 PR tree-optimization/79894
37164 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
37165 to NULL after folding it.
37167 2017-03-06 Richard Biener <rguenther@suse.de>
37169 PR tree-optimization/79824
37170 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
37171 check disabling peeling for gaps.
37173 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
37175 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
37176 attributes): Document gettimeofday.
37178 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37180 * config/s390/s390.c (s390_option_override_internal): Set
37181 PARAM_MIN_VECT_LOOP_BOUND
37183 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37185 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
37186 * config/s390/s390.md: Likewise.
37188 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37191 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
37192 (<avx2_avx512>_perm<mode>): Rename to ...
37193 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
37195 (<avx512>_perm<mode>_mask): Rename to ...
37196 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
37198 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
37199 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
37200 instead of VI8F_256_512.
37201 (avx512f_perm<mode>): New define_expand.
37202 (avx512f_perm<mode>_mask): Likewise.
37203 (avx512f_perm<mode>_1<mask_name>): New define_insn.
37204 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
37206 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37208 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
37209 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
37211 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
37213 2017-03-06 Martin Liska <mliska@suse.cz>
37216 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
37217 when having a SSA NAME w/o VAR_DECL assigned to it.
37219 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37221 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
37222 msa_dpsub_<su>_d): Fix MODE for vec_select.
37224 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37226 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
37228 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
37230 2017-03-06 Richard Biener <rguenther@suse.de>
37232 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
37233 * plugin.c (register_plugin_info): Likewise.
37234 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
37236 2017-03-05 Jakub Jelinek <jakub@redhat.com>
37238 * config/i386/sse.md (sse_storehps, sse_storelps,
37239 avx_<castmode><avxsizesuffix>_<castmode>,
37240 avx512f_<castmode><avxsizesuffix>_<castmode>,
37241 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
37242 in condition that at least one operand is not a MEM.
37244 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37246 PR middle-end/79805
37247 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
37248 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
37250 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
37251 gimple_call_nothrow_p flag based on whether original builtin can throw.
37252 If it can, emit following stmts on the fallthrough edge.
37253 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
37254 don't create new bb if inserting just debug stmts on the edge, try to
37255 insert them on the fallthru bb or just reset debug stmts.
37257 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
37260 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
37261 restore recog_data (including the operand rtxes inside it) around
37262 the call to get_insn_template.
37264 2017-03-03 Martin Sebor <msebor@redhat.com>
37266 PR tree-optimization/79699
37267 * context.c (context::~context): Free MPFR caches to avoid
37268 a memory leak on program exit.
37270 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37272 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
37273 Use wide_int::ulow () instead of .elt (0).
37275 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37277 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
37278 (*pushxf): Limit oF constraint to 32bit targets and add oC
37279 constraint for 64bit targets.
37280 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
37281 (*pushdf): Change rmF constraint to rmC.
37283 2017-03-03 Martin Liska <mliska@suse.cz>
37285 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
37286 Remove unused variable.
37288 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37291 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
37292 is a memory operand, increase num_memory.
37293 (ix86_expand_args_builtin): Likewise.
37295 2017-03-03 Jan Hubicka <jh@suse.cz>
37298 * ipa-devirt.c (maybe_record_node): Properly handle
37299 __cxa_pure_virtual visibility.
37301 2017-03-03 Martin Liska <mliska@suse.cz>
37303 PR tree-optimization/79803
37304 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
37306 (pass_loop_prefetch::execute): Disabled optimization if an
37307 assumption about L1 cache size is not met.
37309 2017-03-03 Martin Liska <mliska@suse.cz>
37311 PR rtl-optimization/79574
37312 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
37313 (hash_scan_set): Likewise.
37314 (dump_hash_table): Likewise.
37315 (hoist_code): Likewise.
37317 2017-03-03 Richard Biener <rguenther@suse.de>
37319 * fixed-value.c (fixed_from_string): Restore use of elt (1)
37320 in place of uhigh ().
37321 (fixed_convert_from_real): Likewise.
37323 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37326 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
37328 2017-03-03 Richard Biener <rguenther@suse.de>
37330 PR middle-end/79818
37331 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
37332 TYPE_OVERFLOW_UNDEFINED check.
37334 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37336 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
37338 (vector_ae_<mode>_p): Likewise.
37339 (vector_nez_<mode>_p): Likewise.
37340 (vector_ne_v2di_p): Likewise.
37341 (vector_ae_v2di_p): Likewise.
37342 (vector_ne_<mode>_p): Likewise.
37343 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
37345 (vsx_tsqrt<mode>2_fe): Likewise.
37347 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
37350 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
37352 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37354 PR rtl-optimization/79780
37355 * cprop.c (one_cprop_pass): When second and further conditional trap
37356 in a single basic block is turned into an unconditional trap, turn it
37357 into a deleted note to avoid RTL verification failures.
37359 2017-03-02 Richard Biener <rguenther@suse.de>
37361 * fold-const.c (const_binop): Use ulow () instead of elt (0).
37363 2017-03-02 Richard Biener <rguenther@suse.de>
37365 PR tree-optimization/79345
37367 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
37368 param and abort the walk, returning -1 if it is hit.
37369 (walk_aliased_vdefs): Take a limit param and pass it on.
37370 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
37371 defaulting to 0 and return a signed int.
37372 * tree-ssa-uninit.c (struct check_defs_data): New struct.
37373 (check_defs): New helper.
37374 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
37375 about uninitialized memory.
37376 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
37377 bogus uninitialized warning.
37378 (fixed_convert_from_real): Likewise.
37380 2017-03-02 Bin Cheng <bin.cheng@arm.com>
37382 PR tree-optimization/66768
37383 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
37384 iv_use if base object can't be determined.
37386 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37388 PR tree-optimization/79345
37389 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
37390 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
37391 (get_pattern_stats): Initialize it.
37392 * genemit.c (gen_expand): Verify match_scratch numbers come after
37393 match_operand/match_dup numbers.
37394 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
37395 match_scratch numbers.
37396 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
37398 * config/s390/s390.md (trunctdsd2): Likewise.
37400 2017-03-02 Richard Biener <rguenther@suse.de>
37402 * wide-int.h (wide_int_storage::operator=): Implement in terms
37405 2017-03-02 Richard Biener <rguenther@suse.de>
37407 PR tree-optimization/79777
37408 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
37409 the to insert expression to sth existing.
37411 2017-03-01 Martin Sebor <msebor@redhat.com>
37413 PR middle-end/79692
37414 * gimple-ssa-sprintf.c
37415 (directive::known_width_and_precision): New function.
37416 (format_integer): Use it.
37417 (get_mpfr_format_length): Consider the full range of precision
37418 when computing %g output with the # flag. Set the likely byte
37419 count to 3 rather than 1 when precision is indeterminate.
37420 (format_floating): Correct the lower bound of precision.
37422 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37424 * doc/invoke.texi: Document default code model for 64-bit Linux.
37426 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37429 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
37430 udiv rather than div since input pattern is unsigned.
37432 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
37434 * config/i386/i386.c (print_reg): Warn for values of
37435 unsupported size in integer register.
37437 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
37440 * config/rs6000/predicates.md (current_file_function_operand): Do
37441 not allow self calls to be local if the function is replaceable.
37443 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
37446 * config/rs6000/altivec.h (vec_ctz and others): Change the
37447 preprocessor macro that controls conditional compilation from
37448 _ARCH_PWR9 to __POWER9_VECTOR__.
37449 (vec_all_ne): Change parameterization of __altivec_scalar_pred
37450 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
37451 control (instead of _ARCH_PWR9 control) so that template
37452 definition uses power9-specific function.
37453 (vec_any_eq): Likewise.
37454 (vec_all_ne): Change macro definition to use a power9-specific
37455 expansion under #ifdef __POWER9_VECTOR__ control (instead of
37456 _ARCH_PWR9 control).
37457 (vec_any_eq) Likewise.
37458 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
37459 expansion for CMPNEF to remove support for xvcmpnesp instruction.
37460 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
37461 support for xvcmpnedp instruction.
37462 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
37463 macro expansion so that Power9 implementation of vec_all_ne does
37464 not use the AltiVec predicate framework.
37465 (VCMPNEH_P): Likewise.
37466 (VCMPNEW_P): Likewise.
37467 (VCMPNED_P): Likewise.
37468 (VCMPNEFP_P): Likewise.
37469 (VCMPNEDP_P): Likewise.
37470 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
37471 implementation of vec_any_eq to not use AltiVec predicate
37473 (VCMPAEH_P): Likewise.
37474 (VCMPAEW_P): Likewise.
37475 (VCMPAED_P): Likewise.
37476 (VCMPAEFP_P): Likewise.
37477 (VCMPAEDP_P): Likewise.
37478 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
37479 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
37480 not use the AltiVec predicate framework.
37481 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
37482 of vec_any_eq to not use AltiVec predicate framework.
37483 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
37484 support for predefined __POWER9_VECTOR__ macro to indicate that
37485 Power9 instruction selection is enabled.
37486 (altivec_overloaded_builtins): Remove extraneous
37487 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
37488 function argument types RS6000_BTI_bool_V16QI and
37489 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
37490 entry for overloaded function argument types RS6000_BTI_bool_V4SI
37491 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
37492 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
37493 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
37494 Power9 for implementations of vec_cmpne. Change the signature for
37495 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
37496 (representing vec_all_ne) to remove the previously described first
37497 argument of type RS6000_BTI_INTSI, as this was an artifact of
37498 reliance on the AltiVec predicate framework, which is no longer
37499 used in the implementation of these functions. Add
37500 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
37501 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
37502 since, unlike the AltiVec predicate framework implementation, we
37503 do not share function descriptors between vec_alle and vec_anyeq.
37504 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
37505 set of modes that receive special treatment even when
37506 TARGET_P9_VECTOR is true. The special treatment emits code that
37507 does not depend on Power9 instructions.
37508 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
37509 define_expand to not rely on AltiVec predicate framework.
37510 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37512 (vector_ne_v2di_p): Change this define_expand to not rely on
37513 AltiVec predicate framework.
37514 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
37516 (vector_ne_<mode>_p): Change this define_expand to not rely on
37517 AltiVec predicate framework.
37518 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37520 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
37521 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
37522 define_insn pattern.
37523 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
37524 define_insn pattern because the xvcmpne<VSs>. instruction is not
37526 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
37527 instruction is not supported.
37529 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37531 * config/nvptx/nvptx.c: Include intl.h.
37533 2017-03-01 Martin Jambor <mjambor@suse.cz>
37536 * ipa-prop.h (ipa_bits): Removed field known.
37537 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
37538 to pointers. Adjusted their comments to warn about their sharing.
37539 (ipcp_transformation_summary): Change bits to a vector of pointers.
37540 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
37541 (ipa_get_ipa_bits_for_value): Declare.
37542 * tree-vrp.h (value_range): Mark as GTY((for_user)).
37543 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
37544 (ipa_bits_hash_table): Likewise.
37545 (ipa_vr_ggc_hash_traits): Likewise.
37546 (ipa_vr_hash_table): Likewise.
37547 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
37548 being pointers and vr_known being removed.
37549 (ipa_set_jf_unknown): Likewise.
37550 (ipa_get_ipa_bits_for_value): New function.
37551 (ipa_set_jfunc_bits): Likewise.
37552 (ipa_get_value_range): New overloaded functions.
37553 (ipa_set_jfunc_vr): Likewise.
37554 (ipa_compute_jump_functions_for_edge): Use the above functions to
37555 construct bits and vr parts of jump functions.
37556 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
37557 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37559 (ipcp_grow_transformations_if_necessary): Also allocate
37560 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37562 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
37563 them. Fix too long lines.
37564 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
37565 vr_known being removed.
37566 (ipa_read_jump_function): Use new setter functions to construct bits
37567 and vr parts of jump functions or set them to NULL.
37568 (write_ipcp_transformation_info): Adjust for bits being pointers.
37569 (read_ipcp_transformation_info): Likewise.
37570 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
37572 Include gt-ipa-prop.h.
37573 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
37575 (ipcp_store_bits_results): Likewise.
37576 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
37577 Do not write to existing jump functions but use a temporary instead.
37579 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37582 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
37583 attempt to use its first operand as BIT_FIELD_REF base.
37585 2017-03-01 Richard Biener <rguenther@suse.de>
37587 PR middle-end/79721
37588 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
37589 interpolating formula in wrapping arithmetic.
37590 (chrec_apply): Convert chrec_evaluate return value to wanted type.
37592 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37594 PR tree-optimization/79734
37595 * tree-vect-generic.c (expand_vector_condition): Optimize
37596 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
37597 Handle VEC_COND_EXPR where comparison has different inner width from
37598 type's inner width.
37600 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
37602 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
37603 markup, and similar issues. Remove @opindex entries for things
37604 that aren't options. Add missing -mmpy-option entries.
37606 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37608 PR tree-optimization/79737
37609 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
37610 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
37611 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
37612 instead of byte_size. Formatting fix.
37613 (shift_bytes_in_array_right): Formatting fix.
37615 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
37618 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
37619 condition on optimize for the leaf function test.
37621 2017-02-28 Martin Liska <mliska@suse.cz>
37624 * read-rtl-function.c (function_reader::handle_unknown_directive):
37625 Bail out when one uses -flto.
37627 2017-02-28 Martin Liska <mliska@suse.cz>
37629 * common.opt: Replace space with tabular for options of <number>
37631 * config/i386/i386.opt: Show <number> value for
37632 -mlarge-data-threshold.
37633 * opts.c (print_filtered_help): Do not display number in hexadecimal
37636 2017-02-28 Martin Liska <mliska@suse.cz>
37638 * common.opt: Fix --help=option -Q for options which are of
37641 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
37643 * config/i386/i386.c (print_reg): Error out for values
37644 of 8-bit size in invalid integer register.
37646 2017-02-28 Martin Sebor <msebor@redhat.com>
37648 PR tree-optimization/79691
37649 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
37651 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37654 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
37655 gcc_unreachable with output_operand_lossage.
37657 2017-02-28 Richard Biener <rguenther@suse.de>
37659 PR tree-optimization/79740
37660 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
37662 (visit_nary_op): Insert the nary into the hashtable if we
37663 pattern-matched sth.
37664 * tree-ssa-pre.c (eliminate_insert): Robustify.
37666 2017-02-28 Richard Biener <rguenther@suse.de>
37668 PR middle-end/79731
37669 * fold-const.c (decode_field_reference): Reject out-of-bound
37672 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37674 * config/i386/i386.c: Include intl.h.
37675 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
37676 instead of just cond ? "..." : "...".
37677 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
37678 * coverage.c (read_counts_file): Likewise.
37679 * omp-offload.c: Include intl.h.
37680 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
37681 of just cond ? "..." : "...".
37682 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
37683 of just cond ? "..." : "...".
37685 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
37688 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
37690 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
37691 'tune for' CPU name.
37692 * config/arm/arm-cpu-data.h: Regenerated.
37694 2017-02-28 Richard Biener <rguenther@suse.de>
37696 PR tree-optimization/79732
37697 * tree-inline.c (expand_call_inline): Do not shadow var.
37699 2017-02-28 Richard Biener <rguenther@suse.de>
37701 PR tree-optimization/79723
37702 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
37703 address-space properly.
37705 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
37707 * doc/optinfo.texi (Optimization groups): Fix option used for
37709 * doc/invoke.texi (-fopt-info): Document "omp".
37710 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
37711 (OPTGROUP_ALL): Add OPTGROUP_OMP.
37712 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
37713 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
37714 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
37716 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
37718 * dumpfile.c (optgroup_options): Instead of "openmp", associate
37719 OPTGROUP_OMP with "omp".
37721 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
37724 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
37725 for arithmetic shift of unsigned V2DI.
37727 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
37729 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
37730 arc/linux.h headers.
37731 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
37732 (LINK_SPEC): Likewise.
37733 (ARC_TLS_EXTRA_START_SPEC): Likewise.
37734 (EXTRA_SPECS): Likewise.
37735 (STARTFILE_SPEC): Likewise.
37736 (ENDFILE_SPEC): Likewise.
37737 (LIB_SPEC): Likewise.
37738 (TARGET_SDATA_DEFAULT): Likewise.
37739 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
37740 (MULTILIB_DEFAULTS): Likewise.
37741 (DWARF2_UNWIND_INFO): Likewise.
37742 * config/arc/big.h: New file.
37743 * config/arc/elf.h: Likewise.
37744 * config/arc/linux.h: Likewise.
37745 * config/arc/t-uClibc: Remove.
37747 2017-02-27 Bin Cheng <bin.cheng@arm.com>
37749 PR tree-optimization/77536
37750 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
37751 (tree_transform_and_unroll_loop): Use above function to compute the
37752 estimated niter of unrolled loop and use it when scaling profile.
37753 Also use count info rather than frequency if it's non-zero.
37754 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
37755 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
37756 (vect_transform_loop): Call above function.
37758 2017-02-27 Richard Biener <rguenther@suse.de>
37760 PR tree-optimization/45397
37761 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
37762 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
37763 (visit_nary_op): Add pattern matching for CSEing sign-changed
37764 or truncated operations with wider ones.
37766 2017-02-27 Richard Biener <rguenther@suse.de>
37768 PR tree-optimization/79690
37769 * tree-vect-stmts.c (vectorizable_store): Use vector type
37770 built from the DR with address-space.
37772 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
37774 * doc/invoke.texi (Optimize Options): Refine the description
37775 of asan-use-after-return.
37777 2017-02-25 Alan Modra <amodra@gmail.com>
37779 PR rtl-optimization/79584
37780 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37781 base, not ad->base_term, the reg within base. Remove assertion
37782 that ad->base == ad->base_term. Replace gen_int_mode using
37783 bogus mode with const0_rtx.
37785 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37787 PR middle-end/79396
37788 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37789 FMA_EXPR like tcc_binary or tcc_unary.
37791 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37794 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37796 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37797 (output_loc_operands): Handle DW_OP_call_ref and
37798 DW_OP_GNU_variable_value.
37799 (struct variable_value_struct): New type.
37800 (struct variable_value_hasher): Likewise.
37801 (variable_value_hash): New variable.
37802 (string_types): Remove.
37803 (copy_loc_descr): New function.
37804 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37805 (prepend_loc_descr_to_each): New function.
37806 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37807 instead of add_loc_descr_to_each if the first argument is single
37808 location list and the second has multiple.
37809 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37810 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37811 when looking for variable value which doesn't have other location info.
37812 (loc_list_from_tree): Formatting fix.
37813 (gen_array_type_die): Simplify DW_AT_string_length handling.
37814 (adjust_string_types): Remove.
37815 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37816 string_types. Call resolve_variable_values.
37817 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37818 (resolve_addr_in_expr): Likewise. Add A argument.
37819 (copy_deref_exprloc): Remove deref argument. Adjust for the
37820 original expression being DW_OP_GNU_variable_value with optionally
37821 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37822 optionally after it.
37823 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37824 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37825 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37826 (variable_value_hasher::hash, variable_value_hasher::equal): New
37828 (resolve_variable_value_in_expr, resolve_variable_value,
37829 resolve_variable_values, note_variable_value_in_expr,
37830 note_variable_value): New functions.
37831 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37834 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37837 * opts.h (handle_generated_option): Add GENERATED_P argument.
37838 * opts-common.c (handle_option): Adjust function comment.
37839 (handle_generated_option): Add GENERATED_P argument, pass it to
37841 (control_warning_option): Pass false to handle_generated_option
37843 * opts.c (maybe_default_option): Pass true to handle_generated_option
37845 * optc-gen.awk: Likewise.
37847 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37849 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37850 a REG, look at the REG it is a SUBREG of.
37851 (splitter for cmpeqsi_t): Ditto.
37853 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37855 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37856 the special USEs with the pattern of the insn, not the insn itself.
37858 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37861 * doc/invoke.texi: Document -mload-store-pairs.
37863 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37864 Sandra Loosemore <sandra@codesourcery.com>
37866 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37867 argument isn't a CONST_INT.
37868 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37869 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37870 (nios2_validate_compare): Bypass alternate compare logic if *op2
37871 is not a CONST_INT.
37872 (ldstwm_operation_p): Return false if first_base is not a REG or
37873 if first_offset is not a CONST_INT.
37875 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37877 * config/cris/cris.md: Use correct operand in a define_peephole2.
37879 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37881 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37883 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37885 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37886 this_insn if it is an INSN or JUMP_INSN.
37887 (force_offsettable): Look at base, not at addr.
37888 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37889 on things that aren't necessarily CONST_INTs.
37891 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37893 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37894 -mfpmath=sse is the default also for x86-32 targets with SSE2
37895 instruction set when @option{-ffast-math} is enabled
37897 2017-02-24 Jeff Law <law@redhat.com>
37899 PR rtl-optimizatoin/79286
37900 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37903 2017-02-24 Richard Biener <rguenther@suse.de>
37905 PR tree-optimization/79389
37906 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37909 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37911 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37912 function comment to reflect reality.
37913 (loop_exits_before_overflow): Fix typo in function description.
37915 2017-02-24 Richard Biener <rguenther@suse.de>
37917 PR tree-optimization/79389
37918 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37919 properly that a threading opportunity exists. Detect conditional
37920 copy/constant propagation opportunities.
37922 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37924 * config/visium/visium.md (type): Add trap.
37925 (b): New mode attribute.
37926 (*btst): Rename into...
37927 (*btst<mode>): ...this and adjust.
37928 (*cbranchsi4_btst_insn): Rename into...
37929 (*cbranch<mode>4_btst_insn): ...this and adjust.
37930 (trap): New define_insn.
37932 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37934 PR tree-optimization/79389
37935 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37936 (noce_reversed_cond_code): New function.
37937 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37938 reversed_comparison_code. Formatting fix.
37939 (noce_try_store_flag): Test rev_cond != NULL in addition to
37940 reversed_comparison_code.
37941 (noce_try_store_flag_constants): Likewise.
37942 (noce_try_store_flag_mask): Likewise.
37943 (noce_try_addcc): Use rev_cond if non-NULL instead of
37944 reversed_comparison_code.
37945 (noce_try_cmove_arith): Likewise. Formatting fixes.
37946 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37947 (noce_find_if_block): Initialize rev_cond.
37948 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37949 instead of false as last argument never attempt to reverse it
37952 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37954 PR tree-optimization/79663
37955 * tree-predcom.c (combine_chains): Process refs in reverse order
37956 only for ZERO length chains, and add explaining comment.
37958 2017-02-23 Jeff Law <law@redhat.com>
37960 PR tree-optimization/79578
37961 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37962 in call to operand_equal_p.
37964 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37967 * config/i386/cpuid.h: Fix another undefined behavior.
37969 2017-02-23 Richard Biener <rguenther@suse.de>
37971 PR tree-optimization/79683
37972 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37973 vector types for data-refs.
37975 2017-02-23 Martin Liska <mliska@suse.cz>
37977 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37979 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37981 PR middle-end/79665
37982 * internal-fn.c (get_range_pos_neg): Moved to ...
37983 * tree.c (get_range_pos_neg): ... here. No longer static.
37984 * tree.h (get_range_pos_neg): New prototype.
37985 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37986 are known to be in between 0 and signed maximum inclusive, try to
37987 expand both unsigned and signed divmod and use the cheaper one from
37990 2017-02-22 Jeff Law <law@redhat.com>
37992 PR tree-optimization/79578
37993 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37994 to compare base operands.
37996 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37999 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
38000 gpc_reg_operand instead of fpr_reg_operand.
38002 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
38004 * config/mips/mips.c (mips_return_in_memory): Force FP
38005 vector types to be returned in memory for o32 ABI.
38007 2017-02-22 Jakub Jelinek <jakub@redhat.com>
38009 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
38010 instead of DW_TAG_member for static data member declarations and don't
38011 set no_linkage_name for static inline data members.
38012 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
38015 2017-02-22 Martin Liska <mliska@suse.cz>
38017 * doc/invoke.texi: Replace inequality signs with square brackets
38020 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
38023 * lra-constraints.c (simplify_operand_subreg): Handle
38024 WORD_REGISTER_OPERATIONS targets.
38026 2017-02-22 Jakub Jelinek <jakub@redhat.com>
38029 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
38030 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
38031 elimination by swapping fld*.
38033 2017-02-22 Richard Biener <rguenther@suse.de>
38035 PR tree-optimization/79673
38036 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
38037 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
38038 irrelevant address-space qualifiers and avoiding a
38039 ADDR_SPACE_CONVERT_EXPR from fold_convert.
38041 2017-02-22 Richard Biener <rguenther@suse.de>
38043 PR tree-optimization/79666
38044 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
38045 to not symbolically negate if that may introduce undefined
38048 2017-02-22 Martin Liska <mliska@suse.cz>
38051 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
38052 * data-streamer-out.c (streamer_write_gcov_count_stream):
38054 * value-prof.c (stream_out_histogram_value): Make assert more
38055 precise based on type of counter.
38057 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
38060 * config/i386/i386.md (standard_x87sse_constant_load splitter):
38061 Use nonimmediate_operand instead of memory_operand for operand 1.
38062 (float-extend standard_x87sse_constant_load splitter): Ditto.
38064 2017-02-21 Jeff Law <law@redhat.com>
38066 PR tree-optimization/79621
38067 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
38068 blocks with edges to themselves.
38070 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38073 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
38074 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
38075 Use gimple_call_builtin_p.
38078 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
38079 on temporarily removed DEBUG_INSNs.
38081 PR tree-optimization/79649
38082 * tree-loop-distribution.c (classify_partition): Give up on
38083 non-generic address space loads/stores.
38085 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
38087 * doc/loop.texi (Loop manipulation): Remove nonexistent
38088 tree_ssa_loop_version from the documentation.
38089 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
38091 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38094 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
38095 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
38096 * config/rs6000/rs6000.c: Include except.h.
38097 (rs6000_expand_split_stack_prologue): Call
38098 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
38100 2017-02-21 Martin Jambor <mjambor@suse.cz>
38103 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
38104 have been analyzed.
38106 2017-02-21 Martin Jambor <mjambor@suse.cz>
38108 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
38109 for backward compatibility only.
38110 * doc/invoke.texi (Option Summary): Remove all references to
38111 -fipa-cp-alignment.
38113 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
38117 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38119 * lra-constraints.c (curr_insn_transform): Handle
38120 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38122 2017-02-21 Martin Liska <mliska@suse.cz>
38124 * config/i386/i386.opt: Replace -masm-dialect with -masm.
38126 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
38128 PR translation/79638
38129 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
38131 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
38134 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
38135 (arm_function_ok_for_sibcall): Return false for an indirect call by
38136 descriptor if all the argument registers are used.
38137 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
38138 alignment of the function.
38140 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38142 PR tree-optimization/61441
38143 * simplify-rtx.c (simplify_const_unary_operation): For
38144 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
38145 the sNaN unmodified.
38147 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
38149 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
38150 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
38151 instead of SYSTEM_HEADER_DIR.
38153 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
38154 Martin Liška <mliska@suse.cz>
38156 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
38157 Fix typos and grammar, use active voice, and clarify.
38159 2017-02-20 Marek Polacek <polacek@redhat.com>
38161 PR middle-end/79537
38162 * gimplify.c (gimplify_expr): Handle unused *&&L;.
38165 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
38167 2017-02-20 Jakub Jelinek <jakub@redhat.com>
38170 * config/i386/i386.c (ix86_expand_builtin): Handle
38171 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
38172 ix86_builtins_isa[fcode].isa as a requirement of those
38173 flags and any other flag in the bitmask.
38174 (ix86_init_mmx_sse_builtins): Use 0 instead of
38175 ~OPTION_MASK_ISA_64BIT as mask.
38176 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
38177 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
38178 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
38179 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
38181 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38184 * lra-constraints.c (split_reg): Check requested split mode
38185 is supported by the register.
38187 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38189 * lra-constraints.c (simplify_operand_subreg): Remove early
38192 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38195 * lra-constraints.c (curr_insn_transform): Tighten condition
38196 for converting SUBREG reloads from OP_OUT to OP_INOUT.
38198 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38201 * lra-constraints.c (curr_insn_transform): Handle
38202 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38204 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
38207 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
38209 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
38211 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
38214 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
38217 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38219 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
38222 2017-02-18 Jakub Jelinek <jakub@redhat.com>
38224 * final.c (last_columnnum, override_columnnum): New variables.
38225 (final_start_function): Set last_columnnum, pass it to begin_prologue
38226 hook and pass 0 to dwarf2out_begin_prologue.
38227 (final_scan_insn): Update override_columnnum. Pass last_columnnum
38228 to source_line debug hook.
38229 (notice_source_line): Compute last_columnnum and for debug_column_info
38230 return true on column changes.
38231 * debug.h (struct gcc_debug_hooks): Add column argument to
38232 source_line and begin_prologue hooks.
38233 (debug_nothing_int_charstar_int_bool): Remove prototype.
38234 (debug_nothing_int_int_charstar,
38235 debug_nothing_int_int_charstar_int_bool): New prototypes.
38236 (dwarf2out_begin_prologue): Add column argument.
38237 * debug.c (do_nothing_debug_hooks): Adjust source_line and
38238 begin_prologue hooks.
38239 (debug_nothing_int_charstar_int_bool): Remove.
38240 (debug_nothing_int_int_charstar,
38241 debug_nothing_int_int_charstar_int_bool): New functions.
38242 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
38243 through to dwarf2out_source_line.
38244 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
38245 (dwarf2out_source_line): Add column argument, emit it if requested.
38246 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
38248 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38249 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38250 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
38251 through to dwarf2out_begin_prologue.
38252 (vmsdbgout_source_line): Add column argument, pass it through to
38253 dwarf2out_source_line.
38254 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
38255 dbxout_source_line caller.
38256 (dbxout_source_line): Add column argument.
38258 * common.opt (gno-column-info, gcolumn-info): New options.
38259 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
38260 (check_die): Also test for multiple DW_AT_decl_column attributes.
38261 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
38262 DW_AT_decl_column if requested.
38263 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
38265 (gen_variable_die): Likewise.
38266 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
38267 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
38270 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
38271 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
38272 (ix86_handle_option): Handle OPT_m3dnowa.
38273 * doc/invoke.texi (-m3dnowa): Document.
38274 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
38275 -m3dnowa instead of -m3dnow -march=athlon.
38278 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
38279 instead of gcc_assert for K, r and R code checks. Formatting fixes.
38281 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38284 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
38285 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
38286 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
38287 generator for vsx_xxpermdi_<mode>_be.
38288 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
38289 force big-endian semantics.
38290 (vsx_xxpermdi_<mode>_be): New define_expand with same
38291 implementation as previous version of vsx_xxpermdi_<mode>.
38293 2017-02-17 Jakub Jelinek <jakub@redhat.com>
38295 PR tree-optimization/79327
38296 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
38297 variable, its initialization and use.
38299 2017-02-17 Julia Koval <julia.koval@intel.com>
38301 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
38302 (OPTION_MASK_ISA_PKU_UNSET): New.
38303 (ix86_handle_option): Handle -mrdpid.
38304 * config/i386/cpuid.h (bit_RDPID): New.
38305 * config/i386/driver-i386.c (host_detect_local_cpu):
38306 Detect RDPID feature.
38307 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
38308 * config/i386/i386-c.c (ix86_target_macros_internal):
38310 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
38311 (ix86_valid_target_attribute_inner_p): Add "rdpid".
38312 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
38313 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
38314 * config/i386/i386.md (define_insn "rdpid"): New.
38315 * config/i386/i386.opt Add -mrdpid.
38316 * config/i386/immintrin.h (_rdpid_u32): New.
38318 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
38320 PR rtl-optimization/79541
38321 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
38322 instead of transforming it into USE.
38324 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
38326 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
38327 If HONOR_SNANS (SFmode) force the input to a register.
38328 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
38329 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
38330 an frsp or similar insn.
38332 2017-02-17 Martin Liska <mliska@suse.cz>
38334 PR rtl-optimization/79577
38335 * params.def (selsched-max-sched-times): Increase minimum to 1.
38337 2017-02-17 Martin Liska <mliska@suse.cz>
38339 PR rtl-optimization/79574
38340 * gcse.c (want_to_gcse_p): Prevent integer overflow.
38342 2017-02-17 Martin Liska <mliska@suse.cz>
38344 PR tree-optimization/79529
38345 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
38346 ssa_defined_default_def_p to handle cases which are implicitly
38348 * tree-ssa.c (ssa_defined_default_def_p): New function.
38349 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
38350 which are implicitly defined.
38351 * tree-ssa.h (ssa_defined_default_def_p): Declare.
38353 2017-02-17 Richard Biener <rguenther@suse.de>
38355 PR middle-end/79576
38356 * params.def (max-ssa-name-query-depth): Limit to 10.
38358 2017-02-17 Richard Biener <rguenther@suse.de>
38360 PR tree-optimization/79552
38361 * tree-ssa-structalias.c (visit_loadstore): Properly verify
38364 2017-02-17 Richard Biener <rguenther@suse.de>
38367 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
38369 2017-02-17 Marek Polacek <polacek@redhat.com>
38371 PR middle-end/79536
38372 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
38373 (fold_negate_expr): New wrapper.
38375 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
38377 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
38378 Correct terminology and de-emphasize pre-standard behavior.
38380 2017-02-16 Alan Modra <amodra@gmail.com>
38382 PR rtl-optimization/79286
38383 * ira.c (def_dominates_uses): New function.
38384 (update_equiv_regs): Don't create an equivalence for insns that
38385 may trap where the register def does not dominate the use.
38387 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
38389 PR rtl-optimization/78127
38390 * lra.c (lra): Call lra_eliminate before finish the loop after
38393 2017-02-16 Richard Biener <rguenther@suse.de>
38395 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
38397 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
38398 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
38399 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
38400 (isl_val_int_from_wi): New function.
38401 (extract_affine_gmp): Rename to ...
38402 (extract_affine_wi): ... this, take a widest_int.
38403 (extract_affine_int): Just wrap extract_affine_wi.
38404 (add_param_constraints): Use isl_val_int_from_wi.
38405 (add_loop_constraints): Likewise, and extract_affine_wi.
38407 2017-02-15 Jeff Law <law@redhat.com>
38409 PR middle-end/79521
38410 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
38411 ira_init_register_move_cost_if_necessary.
38413 2017-02-15 Martin Sebor <msebor@redhat.com>
38415 PR middle-end/32003
38416 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
38417 removed in a prior commit.
38419 2017-02-15 Bin Cheng <bin.cheng@arm.com>
38421 PR tree-optimization/79347
38422 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
38423 counters during peeling.
38425 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
38427 * Makefile.in (site.exp): Remove "set ISLVER".
38429 2017-02-15 Jakub Jelinek <jakub@redhat.com>
38432 * real.c (real_from_integer): Call real_convert even for decimal.
38434 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
38437 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
38439 2017-02-14 Andrew Pinski <apinski@cavium.com>
38441 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
38442 cores and change the partno/implementer to be correct.
38443 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
38444 the 'B" as the implementer.
38445 * config/aarch64/aarch64-tune.md: Regenerate.
38447 2017-02-14 Carl Love <cel@us.ibm.com>
38449 * config/rs6000/rs6000.c: Add case statement entry to make the
38450 xvcvuxdsp built-in argument unsigned.
38451 * config/rs6000/vsx.md: Fix the source and return operand types so they
38452 match the instruction definitions from the ISA document. Fix typo
38453 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
38456 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
38459 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
38460 member early_clobber_alts.
38461 * lra-lives.c (reg_early_clobber_p): New.
38462 (process_bb_lives): Use it.
38463 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
38464 (debug_operand_data): Initialize early_clobber_alts.
38465 (setup_operand_alternative): Set up early_clobber_alts.
38466 (collect_non_operand_hard_regs): Ditto. Pass early clobber
38467 alternatives to new_insn_reg.
38468 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
38470 (lra_update_insn_regno_info): Pass the new arg.
38472 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38474 PR middle-end/79505
38475 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
38476 (new_oacc_loop_raw): Don't clear already cleared fields.
38479 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
38480 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
38481 _mm512_prefetch_i64gather_ps): New inline functions and macros.
38483 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
38486 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
38488 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
38491 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
38492 the extra instruction to the right place to store 128-bit constant
38495 2017-02-14 Martin Sebor <msebor@redhat.com>
38497 PR middle-end/79448
38498 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
38499 warning for strings of unknown length.
38501 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
38503 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
38505 2017-02-14 Jeff Law <law@redhat.com>
38508 * ira-costs.c (scan_one_insn): Initialize register move costs
38509 for pseudos seen in USE/CLOBBER insns.
38511 PR tree-optimization/79095
38512 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
38513 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
38514 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
38515 if the operands are known to be not equal, then the resulting range
38517 (intersect_ranges): If the new range is ~[0,0] and the old range is
38518 wide, then prefer ~[0,0].
38519 * tree-vrp.c (overflow_comparison_p_1): New function.
38520 (overflow_comparison_p): New function.
38521 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
38522 if NAME is used in an overflow test.
38523 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
38524 overflow check that can be expressed as an equality test, then adjust
38525 ops to be that equality test.
38527 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38529 * config/s390/s390-builtin-types.def: Remove flags argument.
38530 * config/s390/s390.c (s390_init_builtins): Likewise.
38532 2017-02-14 Martin Liska <mliska@suse.cz>
38534 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
38535 vector. Fix trailing white spaces.
38537 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
38539 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
38542 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38544 PR rtl-optimization/68664
38545 * config/arm/arm.c (arm_sched_can_speculate_insn):
38546 New function. Declare prototype.
38547 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38549 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38551 PR rtl-optimization/68664
38552 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
38554 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38556 2017-02-14 Amit Pawar <amit.pawar@amd.com>
38558 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
38559 max skip bytes for function, loop and jump.
38561 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38563 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
38564 ABS_EXPR for gimple dump.
38566 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38569 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
38571 PR tree-optimization/79408
38572 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
38573 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
38574 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
38575 also if rhs1 is INTEGER_CST.
38577 2017-02-14 Richard Biener <rguenther@suse.de>
38579 PR middle-end/79432
38580 * tree-into-ssa.c (insert_phi_nodes): When the function can
38581 have abnormal edges rewrite SSA names with broken use-def
38582 dominance out of SSA and register them for PHI insertion.
38584 2017-02-13 Martin Sebor <msebor@redhat.com>
38586 PR middle-end/79496
38587 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
38588 clearing info.nowrite flag when snprintf size argument is a range.
38590 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38592 * cprop.c (cprop_jump): Add missing space in string literal.
38593 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
38594 (get_constraint_for_component_ref): Likewise.
38595 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
38596 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
38597 * lra-constraints.c (process_alt_operands): Likewise.
38598 * ipa-inline.c (inline_small_functions): Likewise.
38599 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
38600 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
38601 * trans-mem.c (diagnose_tm_1_op): Likewise.
38602 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
38603 (grid_parallel_clauses_gridifiable): Likewise.
38605 * config/nvptx/mkoffload.c (process): Add space in between
38608 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
38609 "MOD4_SSE_REGS" and "ALL_REGS".
38611 * spellcheck.c (test_data): Add , in between "foo" and "food".
38613 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38616 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
38617 boundary crossing check and subsequent code generation agree.
38619 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38621 * config/aarch64/aarch64.c (has_memory_op): Delete.
38622 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
38625 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38627 PR rtl-optimization/79388
38628 PR rtl-optimization/79450
38629 * combine.c (distribute_notes): When removing TEM_INSN for which
38630 corresponding dest has last value recorded, invalidate that last
38633 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38635 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
38636 of explicit '@'. Add missing assembly comment marker on branch costs
38639 2017-02-13 Nathan Sidwell <nathan@acm.org>
38641 * gengtype-lex.l (<in_struct>): Add '/'.
38643 2017-02-13 Martin Liska <mliska@suse.cz>
38646 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
38648 2017-02-13 Richard Biener <rguenther@suse.de>
38650 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
38652 * configure: Re-generate.
38653 * config.in: Likewise.
38654 * graphite-dependences.c: Simplify as if
38655 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
38656 * graphite-isl-ast-to-gimple.c: Likewise.
38657 * graphite-optimize-isl.c: Likewise.
38658 * graphite-poly.c: Likewise.
38659 * graphite-sese-to-poly.c: Likewise.
38660 * graphite.h: Likewise.
38661 * toplev.c: Include isl/version.h and use isl_version () for
38662 printing the ISL version.
38663 * doc/install.texi: Update ISL requirement.
38665 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38667 * doc/standards.texi (Standards): Update reference to
38670 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38672 * doc/extend.texi (Named Address Spaces): sourceware.org now
38674 * doc/install.texi (Binaries): Ditto.
38677 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38679 * doc/cpp.texi: Replace "stringify"/"stringification" with C
38680 standard terminology "stringize"/"stringizing" throughout.
38681 * doc/cppinternals.texi: Likewise.
38683 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38685 * doc/extend.texi: Fix some spelling mistakes and typos.
38686 * doc/invoke.texi: Likewise.
38688 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38691 * params.def (inline-min-speedup) Change from 10 to 8.
38693 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38695 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
38698 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38701 * ipa-inline-analysis.c (get_minimal_bb): New function.
38702 (record_modified): Use it.
38703 (remap_edge_change_prob): Handle also ancestor functions.
38705 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
38707 * doc/contrib.texi (Contributors): Remove broken link into
38708 the Mauve CVS repository.
38710 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38712 PR middle-end/79454
38713 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
38714 result computation whenever lhs doesn't have vector mode, not
38715 just when it has BLKmode.
38717 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38719 * doc/makefile.texi (profiledbootstrap): Refer to the
38720 installation instructions only in textual form.
38722 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38725 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
38727 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38729 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
38730 (Specific): Update mingw-w64 reference.
38732 (Specific): Remove broken link to Renesas RX processor.
38734 2017-02-10 Richard Biener <rguenther@suse.de>
38736 * toplev.c (process_options): Do not mention obsolete graphite
38737 options when printing sorry message about missing graphite support.
38738 Mention -floop-nest-optimize.
38740 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
38742 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
38743 (vtst_p16): Likewise.
38744 (vtstq_p8): Likewise.
38745 (vtstq_p16): Likewise.
38747 (vtstq_p64): Likewise.
38748 * config/arm/arm_neon.h (vgetq_lane_p64): New.
38749 (vset_lane_p64): New.
38750 (vsetq_lane_p64): New.
38752 2017-02-10 Jakub Jelinek <jakub@redhat.com>
38754 PR tree-optimization/79411
38755 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
38756 stmt operands are SSA_NAMEs used in abnormal phis.
38757 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
38760 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38763 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
38766 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38768 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
38770 2017-02-09 Jakub Jelinek <jakub@redhat.com>
38772 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38776 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38777 not arbitrary TREE_CONSTANT.
38780 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38781 "omp declare target link" attribute unless is_global_var.
38782 * omp-offload.c (find_link_var_op): Likewise.
38784 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38785 Chung-Lin Tang <cltang@codesourcery.com>
38787 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38789 (gimplify_adjust_omp_clauses): Don't delete TILE.
38790 (gimplify_omp_for): Deal with TILE.
38791 * internal-fn.c (expand_GOACC_TILE): New function.
38792 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38794 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38795 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38797 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38798 avoid DIV for outermost collapse var.
38799 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38800 Remove out of date comments, fix whitespace.
38801 * omp-general.c (omp_extract_for_data): Deal with tiling.
38802 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38803 adjust OLF_DIM_BASE value.
38804 (struct omp_for_data): Add tiling field.
38805 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38806 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38807 for auto loops. Remove default auto determining, moved to
38808 oacc_loop_fixed_partitions.
38809 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38810 stmts, add e_mask field.
38811 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38812 (oacc_thread_numbers): Use oacc_dim_call.
38813 (oacc_xform_tile): New.
38814 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38815 (finish_oacc_loop): Adjust for ifns vector.
38816 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38817 add case for GOACC_TILE fns.
38818 (oacc_loop_xform_loop): Delete.
38819 (oacc_loop_process): Iterate over call list directly, and add
38820 handling for GOACC_TILE fns.
38821 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38823 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38824 vector partitioning to outer loops. Assign 2 partitions to loops
38825 when available. Add TILE handling.
38826 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38827 (execite_oacc_device_lower): Process GOACC_TILE fns,
38828 ignore unknown specs.
38829 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38830 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38831 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38833 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38835 * configure.ac (ACX_BUGURL): Update.
38836 * configure: Regenerate.
38838 2017-02-09 Richard Biener <rguenther@suse.de>
38840 PR tree-optimization/69823
38841 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38842 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38844 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38846 * config/arc/arc-c.def: Add __NPS400__ definition.
38847 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38848 (TARGET_NPS400): Define.
38850 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38852 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38854 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38855 pointer, arch_info.
38856 (arc_cpu_types): Fill the arch_info field with a pointer into the
38857 arc_arch_types table.
38858 (arc_selected_cpu): Declare.
38859 * config/arc/arc.c (arc_selected_cpu): Make global.
38860 (arc_selected_arch): Delete.
38861 (arc_base_cpu): Delete.
38862 (arc_override_options): Remove references to deleted variables,
38863 update access to arch information.
38864 (ARC_OPT): Update access to arch information.
38865 (ARC_OPTX): Likewise.
38866 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38867 (TARGET_ARC600): Update access to arch information.
38868 (TARGET_ARC601): Likewise.
38869 (TARGET_ARC700): Likewise.
38870 (TARGET_EM): Likewise.
38871 (TARGET_HS): Likewise.
38872 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38875 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38878 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38879 condition/operands for integer GE/LE/GEU/LEU operations.
38881 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38883 PR translation/79397
38884 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38887 2017-02-08 Martin Jambor <mjambor@suse.cz>
38890 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38891 whether allocation happened.
38892 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38893 nothing was allocated.
38895 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38897 PR tree-optimization/79408
38898 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38899 constant, but SSA_NAME with a known integer range, use the minimum
38900 of that range instead of op1 to determine if modulo can be replaced
38901 with its first operand.
38903 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38905 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38907 2017-02-08 Richard Biener <rguenther@suse.de>
38909 PR tree-optimization/71824
38910 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38911 Check all loops contained in the merged region.
38913 2017-02-07 Andrew Pinski <apinski@cavium.com>
38915 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38917 2017-02-07 Andrew Pinski <apinski@cavium.com>
38919 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38920 (thunderxt88): Likewise.
38921 (thunderxt81): Disable LSE and change v8.1 to v8.
38922 (thunderxt83): Likewise.
38924 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38925 Richard Biener <rguenther@suse.de>
38927 PR middle-end/79399
38928 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38929 type from int to size_t.
38930 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38932 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38934 PR rtl-optimization/79386
38935 * cprop.c (bypass_conditional_jumps): Initialize
38936 bypass_last_basic_block already before splitting bbs after
38937 unconditional traps...
38938 (bypass_conditional_jumps): ... rather than here.
38941 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38942 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38943 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38944 fix -masm=intel patterns.
38946 2017-02-07 Richard Biener <rguenther@suse.de>
38948 PR tree-optimization/79256
38949 PR middle-end/79278
38950 * builtins.c (get_object_alignment_2): Use min_align_of_type
38951 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38952 and ADJUST_FIELD_ALIGN.
38954 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38956 * doc/tm.texi: Regenerate.
38957 * stor-layout.c (layout_decl): Adjust.
38958 (update_alignment_for_field): Likewise.
38959 (place_field): Likewise.
38960 (min_align_of_type): Likewise.
38961 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38962 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38963 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38964 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38965 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38966 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38967 * config/i386/i386.c (x86_field_alignment): Likewise.
38968 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38969 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38970 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38971 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38972 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38973 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38977 2017-01-30 Richard Biener <rguenther@suse.de>
38979 PR tree-optimization/79256
38980 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38981 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38984 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38986 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38987 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38988 builtins to SImode and emit a zero-extend, if necessary.
38990 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38992 * docs/invoke.texi (RISC-V Options): Alphabetize.
38994 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38996 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38999 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
39001 * config/riscv/riscv.c: New file.
39002 * common/config/riscv/riscv-common.c: Likewise.
39003 * config.gcc: Likewise.
39004 * config/riscv/constraints.md: Likewise.
39005 * config/riscv/elf.h: Likewise.
39006 * config/riscv/generic.md: Likewise.
39007 * config/riscv/linux.h: Likewise.
39008 * config/riscv/multilib-generator: Likewise.
39009 * config/riscv/peephole.md: Likewise.
39010 * config/riscv/pic.md: Likewise.
39011 * config/riscv/predicates.md: Likewise.
39012 * config/riscv/riscv-builtins.c: Likewise.
39013 * config/riscv/riscv-c.c: Likewise.
39014 * config/riscv/riscv-ftypes.def: Likewise.
39015 * config/riscv/riscv-modes.def: Likewise.
39016 * config/riscv/riscv-opts.h: Likewise.
39017 * config/riscv/riscv-protos.h: Likewise.
39018 * config/riscv/riscv.h: Likewise.
39019 * config/riscv/riscv.md: Likewise.
39020 * config/riscv/riscv.opt: Likewise.
39021 * config/riscv/sync.md: Likewise.
39022 * config/riscv/t-elf-multilib: Likewise.
39023 * config/riscv/t-linux: Likewise.
39024 * config/riscv/t-linux-multilib: Likewise.
39025 * config/riscv/t-riscv: Likewise.
39026 * configure.ac: Likewise.
39027 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
39028 Waterman as RISC-V maintainers.
39029 * doc/install.texi: Add RISC-V entries.
39030 * doc/invoke.texi: Add RISC-V options section.
39031 * doc/md.texi: Add RISC-V constraints section.
39032 * configure: Regenerated.
39034 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
39037 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
39038 false values to be constant vectors with all 0 or all 1 bits set.
39039 (vcondu<mode><mode>): Likewise.
39040 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
39042 (fpmask_comparison_operator): Update comment.
39043 (vecint_comparison_operator): New predicate.
39044 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
39045 vector conditionals when the true and false values are constant
39046 vectors with all 0 bits or all 1 bits set.
39048 2017-02-06 Martin Sebor <msebor@redhat.com>
39050 PR tree-optimization/79376
39051 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
39053 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
39055 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
39056 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
39057 to simplify split condition.
39059 2017-02-06 Jakub Jelinek <jakub@redhat.com>
39061 * omp-expand.c (oxpand_omp_atomic_fetch_op,
39062 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
39065 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
39067 PR rtl-optimization/68664
39068 * target.def (can_speculate_insn): New hook.
39069 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
39070 * doc/tm.texi: Regenerate.
39071 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
39072 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
39073 (rs6000_sched_can_speculate_insn): New function.
39075 2017-02-06 Jakub Jelinek <jakub@redhat.com>
39077 PR tree-optimization/79284
39078 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
39079 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
39080 vectorizable_mask_load_store, vectorizable_operation,
39081 vect_is_simple_cond, get_same_sized_vectype): Use it instead
39082 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
39083 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
39084 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
39085 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
39086 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
39087 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
39088 is_gimple_assign (stmt). Replace another such test with
39089 is_gimple_assign (stmt).
39091 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
39094 * config/avr/avr.c (rtl-iter.h): Include it.
39095 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
39096 (avr_legitimate_combined_insn): ...and implementation.
39098 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
39100 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
39101 * config/s390/s390.c (s390_const_operand_ok)
39102 (s390_canonicalize_comparison, s390_extract_part)
39103 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
39104 (s390_contiguous_bitmask_p, s390_rtx_costs)
39105 (legitimize_pic_address): Likewise.
39106 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
39107 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
39108 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
39109 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
39110 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
39112 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
39114 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
39115 REGNO($0) == REGNO($1).
39117 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39119 * config/s390/linux.h(SIZE_TYPE): Add comment.
39121 2017-02-06 Julian Brown <julian@codesourcery.com>
39122 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39123 Virendra Pathak <virendra.pathak@broadcom.com>
39125 * config/aarch64/aarch64-cores.def: Change the scheduler
39127 * config/aarch64/aarch64.md: Include thunderx2t99.md.
39128 * config/aarch64/thunderx2t99.md: New file.
39130 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39132 * doc/standards.texi (Go Language): Update link to language
39135 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
39137 * tree-eh.c (lower_resx): Sanitize profile.
39138 (cleanup_empty_eh_move_lp): Likewise.
39140 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
39143 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
39145 * cfgloopmanip.h (loop_version): Update prototype.
39146 * modulo-sched.c (sms_schedule): Update call of loop_version.
39147 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
39148 * tree-parloops.c (gen_parallel_loop): Likewise.
39149 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
39150 * tree-ssa-loop-split.c (split_loop): Likewise.
39151 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
39152 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
39154 2017-02-05 Martin Liska <mliska@suse.cz>
39157 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
39159 (print_operand_address): Initialize a struct to zero.
39161 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39163 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
39164 garbage collector only in textual form.
39166 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39168 * doc/extend.texi (x86 specific memory model extensions for
39169 transactional memory): Simplify a phrase.
39171 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
39174 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
39175 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
39176 (atomic_storedi_1): Likewise.
39178 2017-02-04 Jakub Jelinek <jakub@redhat.com>
39180 PR tree-optimization/79338
39181 * tree-parloops.c (gather_scalar_reductions): Don't call
39182 vect_analyze_loop_form for loop->inner before destroying loop's
39185 2017-02-03 Martin Sebor <msebor@redhat.com>
39187 PR tree-optimization/79327
39188 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
39189 when precision has resulted in leading zeros.
39190 (format_integer): Adjust the likely counter to assume an unknown
39191 argument that may be zero is non-zero.
39193 2017-02-03 Jason Merrill <jason@redhat.com>
39196 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
39197 avoid copying non-taken branch.
39199 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39201 PR tree-optimization/79340
39202 * tree-vect-loop.c (vectorizable_reduction): Release
39203 vec_defs elements after safe_splicing them into other vectors.
39206 PR tree-optimization/79327
39207 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
39208 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
39210 (format_integer): Use wide_int_to_tree instead of build_int_cst
39211 + to_?hwi. If argmin is NULL, just set argmin and argmax to
39212 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
39213 of shortest and longest sequence.
39215 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
39217 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
39218 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
39220 2017-02-03 Walter Lee <walt@tilera.com>
39223 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
39224 after initial stackframe link reg save.
39225 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
39227 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39230 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
39231 wu for stxssp alternative.
39233 2017-02-03 Martin Sebor <msebor@redhat.com>
39235 PR tree-optimization/79352
39236 * gimple-fold.c (get_range_strlen): Add argument.
39237 (get_range_strlen): Change return type to bool.
39238 (get_maxval_strlen): Pass in a dummy argument.
39239 * gimple-fold.h (get_range_strlen): Change return type to bool.
39240 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
39241 * tree.h (array_at_struct_end_p): Add argument.
39242 * tree.c (array_at_struct_end_p): Handle it.
39244 2017-02-03 Martin Liska <mliska@suse.cz>
39247 * multiple_target.c (create_dispatcher_calls): Redirect edge
39248 from a caller of a dispatcher.
39249 (expand_target_clones): Make the clones local.
39250 (ipa_target_clone): Do both target clones and resolvers.
39251 (ipa_dispatcher_calls): Remove the pass.
39252 (pass_dispatcher_calls::gate): Likewise.
39253 (make_pass_dispatcher_calls): Likewise.
39254 * passes.def (pass_target_clone): Put as very first IPA early
39257 2017-02-03 Martin Liska <mliska@suse.cz>
39259 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
39260 in case of a function with ifunc attribute.
39262 2017-02-03 Martin Liska <mliska@suse.cz>
39264 * cgraph.c (cgraph_node::dump): Dump function version info.
39265 * symtab.c (symtab_node::dump_base): Add missing new line.
39267 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39269 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
39270 (ifcombine_ifandif): Use it.
39272 2017-02-03 Martin Liska <mliska@suse.cz>
39274 * doc/invoke.texi: Document default value for
39275 use-after-scope-direct-emission-threshold.
39277 2017-02-03 Martin Liska <mliska@suse.cz>
39279 PR tree-optimization/79339
39280 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
39281 (format_floating): Likewise.
39283 2017-02-03 Martin Liska <mliska@suse.cz>
39286 * ipa-prop.c (ipa_node_params_t::insert): Remove current
39288 (ipa_node_params_t::remove): Likewise.
39289 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
39290 initialization from removed ipa_node_params_t::insert.
39291 (ipa_node_params::~ipa_node_params): Move from removed
39292 ipa_node_params_t::release.
39293 * symbol-summary.h (symbol_summary::m_released): New member.
39294 Do not release a summary twice. Do not allow to call finalizer
39295 for types of a summary that live in GGC memory.
39297 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39299 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
39302 2017-02-02 Martin Sebor <msebor@redhat.com>
39304 PR middle-end/79275
39305 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
39306 (format_string): Tighten up the range of output for non-constant
39307 strings and correct the expected range for wide non-constant strings.
39309 2017-02-02 Martin Sebor <msebor@redhat.com>
39311 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
39313 PR middle-end/32003
39314 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
39316 (-fdump-tree-@var): Add to index and document how to come up
39317 with pass-specific option and dump file names.
39318 (-fdump-passes): Clarify where to look for output.
39320 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39322 PR middle-end/77445
39323 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
39324 statistics of the analyzed path; allow threading for speed when
39325 any of BBs along the path are optimized for speed.
39327 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
39329 PR middle-end/78468
39330 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
39331 settings of the virtual registers.
39334 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
39336 * explow.c (get_dynamic_stack_size): Take known alignment of stack
39337 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
39340 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39342 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
39343 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
39345 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39347 * config/s390/s390.md: Add missing comments with the expanded
39349 * config/s390/vector.md: Likewise.
39350 * config/s390/vx-builtins.md: Likewise.
39352 2017-02-02 Jakub Jelinek <jakub@redhat.com>
39355 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
39356 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
39357 conditions on a single line.
39359 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39361 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39362 __S390_VX__ to __VX__.
39364 2017-02-01 Andrew Pinski <apinski@cavium.com>
39366 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
39367 stmt_info to record_stmt_cost.
39368 (vect_get_known_peeling_cost): Pass stmt_info if known to
39370 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
39371 cpu_vector_cost field into
39372 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
39373 field into vec_int_stmt_cost and vec_fp_stmt_cost.
39374 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
39375 splitting of scalar_stmt_cost and vec_stmt_cost.
39376 (thunderx_vector_cost): Likewise.
39377 (cortexa57_vector_cost): LIkewise.
39378 (exynosm1_vector_cost): Likewise.
39379 (xgene1_vector_cost): Likewise.
39380 (thunderx2t99_vector_cost): Improve after the splitting of the two
39382 (aarch64_builtin_vectorization_cost): Update for the splitting of
39383 scalar_stmt_cost and vec_stmt_cost.
39385 2017-02-01 Torvald Riegel <triegel@redhat.com>
39386 Richard Henderson <rth@redhat.com>
39388 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
39389 conditional on existance of a fast atomic load.
39390 * optabs-query.c (can_atomic_load_p): New function.
39391 * optabs-query.h (can_atomic_load_p): Declare it.
39392 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
39393 no fast atomic load is available for the particular size of access.
39394 (expand_atomic_compare_and_swap): Likewise.
39395 (expand_atomic_load): Likewise.
39396 (expand_atomic_store): Likewise.
39397 (expand_atomic_fetch_op): Likewise.
39398 * testsuite/lib/target-supports.exp
39399 (check_effective_target_sync_int_128): Remove x86 because it provides
39400 no fast atomic load.
39401 (check_effective_target_sync_int_128_runtime): Likewise.
39403 2017-02-01 Richard Biener <rguenther@suse.de>
39405 * graphite.c: Include tree-vectorizer.h for find_loop_location.
39406 (graphite_transform_loops): Provide opt-info for optimized nests.
39407 * tree-parloop.c (parallelize_loops): Provide opt-info for
39408 parallelized loops.
39410 2017-02-01 Richard Biener <rguenther@suse.de>
39412 PR middle-end/79315
39413 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
39414 was not set before.
39416 2017-02-01 Richard Biener <rguenther@suse.de>
39418 PR tree-optimization/71824
39419 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
39420 Verify the loops are valid in the merged SESE region.
39421 (scop_detection::can_represent_loop_1): Check analyzing the
39422 evolution of the number of iterations in the region succeeds.
39424 2017-01-31 Ian Lance Taylor <iant@golang.org>
39426 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
39427 REG_ARGS_SIZE note to 32-bit push insns and call insn.
39429 2017-01-31 David Malcolm <dmalcolm@redhat.com>
39431 PR preprocessor/79210
39432 * input.c (get_substring_ranges_for_loc): Replace line_width
39433 assertion with error-handling.
39435 2017-01-31 Richard Biener <rguenther@suse.de>
39437 PR tree-optimization/77318
39438 * graphite-sese-to-poly.c (extract_affine): Fix assert.
39439 (create_pw_aff_from_tree): Take loop parameter.
39440 (add_condition_to_pbb): Pass loop of the condition to
39441 create_pw_aff_from_tree.
39443 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39445 * config/s390/s390.c (s390_asan_shadow_offset): New function.
39446 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
39448 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
39452 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
39454 (convert_int_to_float128): Likewise.
39455 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
39456 (convert_int_to_float128): Likewise.
39457 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
39458 (UNSPEC_IEEE128_CONVERT): Likewise.
39459 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
39460 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
39461 Use local variables for IBM extended format.
39462 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
39463 (fix_trunc<mode>si2_fprs): Likewise.
39464 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
39465 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
39466 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
39467 to know that we can now have integers of all sizes in vector
39469 (fix<uns>_<mode>di2_hw): Likewise.
39470 (float<uns>_<mode>si2_hw): Likewise.
39471 (fix_<mode>si2_hw): Likewise.
39472 (fixuns_<mode>si2_hw): Likewise.
39473 (float<uns>_<mode>di2_hw): Likewise.
39474 (float_<mode>di2_hw): Likewise.
39475 (float_<mode>si2_hw): Likewise.
39476 (floatuns_<mode>di2_hw): Likewise.
39477 (floatuns_<mode>si2_hw): Likewise.
39478 (xscvqp<su>wz_<mode>): Delete, no longer used.
39479 (xscvqp<su>dz_<mode>): Likewise.
39480 (xscv<su>dqp_<mode>): Likewise.
39481 (ieee128_mfvsrd_64bit): Likewise.
39482 (ieee128_mfvsrd_32bit): Likewise.
39483 (ieee128_mfvsrwz): Likewise.
39484 (ieee128_mtvsrw): Likewise.
39485 (ieee128_mtvsrd_64bit): Likewise.
39486 (ieee128_mtvsrd_32bit): Likewise.
39488 2017-01-31 Martin Liska <mliska@suse.cz>
39491 * ipa-prop.c (ipa_free_all_node_params): Call release method
39492 instead of ~sumbol_summary to not to trigger double times
39495 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
39497 PR tree-optimization/71691
39498 * bitmap.h (class auto_bitmap): New.
39499 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
39500 is_maybe_undefined instead of ssa_undefined_value_p.
39502 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39504 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39505 __S390_ARCH_LEVEL__ to __ARCH__.
39507 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39509 PR tree-optimization/79267
39510 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
39511 if should_remove_lhs_p is true.
39513 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
39516 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
39517 (add_alignment_attribute): New.
39518 (base_type_die): Add alignment attribute.
39519 (subrange_type_die): Likewise.
39520 (modified_type_die): Likewise.
39521 (gen_array_type_die): Likewise.
39522 (gen_descr_array_type_die: Likewise.
39523 (gen_enumeration_type_die): Likewise.
39524 (gen_subprogram_die): Likewise.
39525 (gen_variable_die): Likewise.
39526 (gen_field_die): Likewise.
39527 (gen_ptr_to_mbr_type_die): Likewise.
39528 (gen_struct_or_union_type_die): Likewise.
39529 (gen_subroutine_type_die): Likewise.
39530 (gen_typedef_die): Likewise.
39531 (base_type_cmp): Compare alignment attribute.
39533 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39536 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
39537 (setb_unsigned) New pattern for setb with CCUNS.
39538 * config/rs6000/rs6000.c (expand_block_compare): Use a different
39539 subfc./subfe sequence to avoid overflow problems. Generate a
39540 shorter sequence with cmpld/setb for power9.
39541 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
39542 for generating subfc. instruction.
39543 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
39544 now uses this instruction.
39546 2017-01-30 Ian Lance Taylor <iant@google.com>
39549 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
39550 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
39552 2017-01-30 Martin Sebor <msebor@redhat.com>
39554 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
39555 Move constant to the right of a relational operator.
39556 (get_mpfr_format_length, format_character, format_string): Ditto.
39557 (should_warn_p, maybe_warn): Same.
39559 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
39561 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
39564 * asan.c (get_translation_unit_decl): Remove function.
39565 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
39567 2017-01-30 Martin Liska <mliska@suse.cz>
39569 PR gcov-profile/79259
39570 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
39571 -fprofile-generate.
39573 2017-01-30 Martin Liska <mliska@suse.cz>
39576 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
39577 Initialize variables with NULL value.
39579 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
39582 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
39584 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
39586 2017-01-30 Richard Biener <rguenther@suse.de>
39588 PR tree-optimization/79276
39589 * tree-vrp.c (process_assert_insertions): Properly adjust common
39590 when removing a duplicate.
39592 2017-01-30 Richard Biener <rguenther@suse.de>
39594 PR tree-optimization/79256
39595 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
39596 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
39598 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
39600 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
39603 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
39604 ("*r<noxa>sbg_<mode>_sll_bitmask")
39605 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
39606 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
39607 Use contiguous_bitmask_nowrap_operand.
39609 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39612 * config/rs6000/altivec.h (vec_xl): Revise #define.
39613 (vec_xst): Likewise.
39615 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
39617 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
39619 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
39621 PR rtl-optimization/79194
39622 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
39623 traps before call to bypass_conditional_jumps.
39625 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39627 PR tree-optimization/71374
39628 * lra-constraints.c (check_conflict_input_operands): New.
39629 (match_reload): Use it.
39631 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39634 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
39635 account to calculate conflict_set.
39637 2017-01-27 Bin Cheng <bin.cheng@arm.com>
39639 PR rtl-optimization/78559
39640 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
39641 other_insn in combine.
39643 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
39645 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
39646 uint16_type_node for BT_UINT16.
39648 2017-01-27 David Malcolm <dmalcolm@redhat.com>
39650 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
39651 "RTL Tests" to menu.
39652 (GIMPLE Tests): New node.
39653 (RTL Tests): New node.
39655 2017-01-27 Richard Biener <rguenther@suse.de>
39657 PR tree-optimization/79245
39658 * tree-loop-distribution.c (distribute_loop): Apply cost
39659 modeling also to detected patterns.
39661 2017-01-27 Richard Biener <rguenther@suse.de>
39663 PR tree-optimization/71433
39664 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
39665 (compare_assert_loc): New function.
39666 (process_assert_insertions): Sort and optimize assert locations
39667 to remove duplicates and push down identical assertions on
39668 edges to their destination block.
39670 2017-01-27 Richard Biener <rguenther@suse.de>
39672 PR tree-optimization/79244
39673 * tree-vrp.c (remove_range_assertions): Forcefully propagate
39674 out SSA names even if abnormal.
39676 2017-01-27 Jakub Jelinek <jakub@redhat.com>
39678 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
39679 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
39680 instead of MPFR_RNDN.
39682 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
39685 * arm.c (arm_option_override): Don't call build_target_option_node
39686 until after doing all option overrides.
39687 (arm_valid_target_attribute_tree): Likewise.
39689 2017-01-27 Martin Liska <mliska@suse.cz>
39691 * doc/invoke.texi (-fprofile-arcs): Document profiling support
39692 for {cd}tors and C++ {cd}tors.
39694 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
39696 * config/s390/s390.md ("*setmem_long_and")
39697 ("*setmem_long_and_31z"): Use zero_extend instead of and.
39699 2017-01-26 Martin Sebor <msebor@redhat.com>
39701 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
39704 2017-01-26 Martin Sebor <msebor@redhat.com>
39706 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
39707 HAVE_DFmode before using XFmode or DFmode.
39708 (parse_directive): Avoid using the z length modifier to avoid
39709 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
39711 PR middle-end/78703
39712 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
39713 to accept adjustment as an array.
39714 (get_int_range): New function.
39715 (struct directive): Make width and prec arrays.
39716 (directive::set_width, directive::set_precision): Call get_int_range.
39717 (format_integer, format_floating): Handle width and precision ranges.
39718 (format_string, parse_directive): Same.
39720 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39723 * dwarf2out.c (generate_skeleton_bottom_up): For children with
39724 comdat_type_p set, just clone them, but keep the children in the
39728 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
39729 which have direct callers with -fvar-tracking-assignments enabled
39731 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
39732 inside of type units.
39734 2017-01-26 Martin Sebor <msebor@redhat.com>
39736 PR middle-end/78703
39737 * gimple-ssa-sprintf.c (struct result_range): Add likely and
39739 (struct format_result): Replace number_chars, number_chars_min,
39740 and number_chars_max with a single member of struct result_range.
39742 (format_result::operator+=): Adjust.
39743 (struct fmtresult): Remove bounded. Handle likely and unlikely
39745 (fmtresult::adjust_for_width_or_precision): New function.
39746 (fmtresult:type_max_digits): New function.
39747 (bytes_remaining): Handle likely and unlikely counters.
39748 (min_bytes_remaining): Remove.
39749 (format_percent): Simplify.
39750 (format_integer, format_floating): Set likely and unlikely counters.
39751 (get_string_length, format_character, format_string): Same.
39752 (format_plain, should_warn_p): New function.
39753 (maybe_warn): Call should_warn_p. Update diagnostic messages
39754 and handle those for all directives, including plain strings.
39755 (format_directive): Handle likely and unlikely counters.
39756 Remove unnecessary quoting from diagnostics. Add an informational
39758 (add_bytes): Remove.
39759 (pass_sprintf_length::compute_format_length): Simplify.
39760 (try_substitute_return_value): Handle likely and unlikely counters.
39762 2017-01-26 Carl Love <cel@us.ibm.com>
39764 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
39765 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
39767 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
39770 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
39771 endianess for subregs into account.
39772 * lra-constraints.c (lra_constraints): Do risky transformations
39773 always on the first iteration.
39774 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39775 last_call_used_reg_set.
39776 (process_bb_lives): Define and use last_call_used_reg_set.
39777 * lra.c (lra): Always continue after lra_constraints on the first
39780 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39782 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39784 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39786 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39788 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39789 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39790 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39791 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39792 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39793 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39794 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39795 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39796 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39798 2017-01-26 Marek Polacek <polacek@redhat.com>
39801 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39802 for the third operand.
39804 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39806 PR middle-end/79236
39807 * omp-low.c (struct omp_context): Add simt_stmt field.
39808 (scan_omp_for): Return omp_context *.
39809 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39810 context to the _simt_ SIMD stmt.
39811 (lower_omp_for): For combined SIMD with sibling _simt_
39812 SIMD, make sure to use the same decls in _looptemp_
39813 clauses as in the sibling.
39815 2017-01-26 David Sherwood <david.sherwood@arm.com>
39817 PR middle-end/79212
39818 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39821 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39824 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39825 emit fld b; fld a; if possible.
39827 * brig-builtins.def: Update copyright years.
39828 * config/arm/arm_acle_builtins.def: Update copyright years.
39830 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39833 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39834 constraint instead of o for the stxsd instruction.
39836 2017-01-25 Carl Love <cel@us.ibm.com>
39838 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39839 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39841 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39843 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39845 2017-01-25 Richard Biener <rguenther@suse.de>
39847 PR tree-optimization/69264
39848 * target.def (vector_alignment_reachable): Improve documentation.
39849 * doc/tm.texi: Regenerate.
39850 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39852 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39853 earlier changes with respect to TYPE_USER_ALIGN.
39854 (vector_alignment_reachable_p): Likewise. Improve dumping.
39856 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39859 * config/arm/arm.md (xordi3): Force constant operand into a register
39862 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39864 * doc/invoke.texi (-fstore-merging): Correct default optimization
39865 levels at which it is enabled.
39866 (-O): Move -fstore-merging from list to...
39869 2017-01-25 Richard Biener <rguenther@suse.de>
39872 * omp-expand.c: Include debug.h.
39873 (expand_omp_taskreg): Make sure to generate early debug before
39874 outlining anything from a function.
39875 (expand_omp_target): Likewise.
39876 (grid_expand_target_grid_body): Likewise.
39878 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39881 * asan.c (get_translation_unit_decl): New function.
39882 (asan_add_global): Extract modules file name from globals
39883 TRANSLATION_UNIT_DECL name.
39885 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39888 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39889 for long calls with APCS frame and VFP.
39891 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39893 * cfg.c (original_copy_tables_initialized_p): New function.
39894 * cfg.h (original_copy_tables_initialized_p): New decl.
39895 * cfgrtl.c (relink_block_chain): Guard the call to
39896 free_original_copy_tables with a call to
39897 original_copy_tables_initialized_p.
39898 * cgraph.h (symtab_node::native_rtl_p): New decl.
39899 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39900 (symtab_node::needed_p): Don't assert for early assembly output
39901 for __RTL functions.
39902 (cgraph_node::finalize_function): Set "force_output" for __RTL
39904 (cgraph_node::analyze): Bail out early for __RTL functions.
39905 (analyze_functions): Update assertion to support __RTL functions.
39906 (cgraph_node::expand): Bail out early for __RTL functions.
39907 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39909 * function.h (struct function): Update comment for field
39911 * gimple-expr.c: Include "tree-pass.h".
39912 (gimple_has_body_p): Return false for __RTL functions.
39913 * Makefile.in (OBJS): Add run-rtl-passes.o.
39914 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39916 (gcc::pass_manager::get_clean_slate): New accessor.
39917 * passes.c: Include "insn-addr.h".
39918 (should_skip_pass_p): Add logging. Update logic for running
39919 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39920 property-provider override so it is only done for gimple passes.
39922 (skip_pass): New function.
39923 (execute_one_pass): Call skip_pass when skipping passes.
39924 * read-md.c (md_reader::read_char): Support filtering
39925 the input to a subset of line numbers.
39926 (md_reader::md_reader): Initialize fields
39927 m_first_line and m_last_line.
39928 (md_reader::read_file_fragment): New function.
39929 * read-md.h (md_reader::read_file_fragment): New decl.
39930 (md_reader::m_first_line): New field.
39931 (md_reader::m_last_line): New field.
39932 * read-rtl-function.c (function_reader::create_function): Only
39933 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39934 curr_properties. Set DECL_INITIAL to a dummy block.
39935 (read_rtl_function_body_from_file_range): New function.
39936 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39938 * run-rtl-passes.c: New file.
39939 * run-rtl-passes.h: New file.
39941 2017-01-24 Jeff Law <law@redhat.com>
39943 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39946 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39948 PR tree-optimization/79159
39949 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39950 (record_nonwrapping_iv): Improve boundary using above function if no
39951 value range information.
39953 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39954 Martin Jambor <mjambor@suse.cz>
39956 * brig-builtins.def: New file.
39957 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39958 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39959 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39960 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39961 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39962 * builtin-types.def (BT_INT8): New.
39963 (BT_INT16): Likewise.
39964 (BT_UINT8): Likewise.
39965 (BT_UINT16): Likewise.
39966 (BT_FN_ULONG): Likewise.
39967 (BT_FN_UINT_INT): Likewise.
39968 (BT_FN_UINT_ULONG): Likewise.
39969 (BT_FN_UINT_LONG): Likewise.
39970 (BT_FN_UINT_PTR): Likewise.
39971 (BT_FN_ULONG_PTR): Likewise.
39972 (BT_FN_INT8_FLOAT): Likewise.
39973 (BT_FN_INT16_FLOAT): Likewise.
39974 (BT_FN_UINT32_FLOAT): Likewise.
39975 (BT_FN_UINT16_FLOAT): Likewise.
39976 (BT_FN_UINT8_FLOAT): Likewise.
39977 (BT_FN_UINT64_FLOAT): Likewise.
39978 (BT_FN_UINT16_UINT32): Likewise.
39979 (BT_FN_UINT32_UINT16): Likewise.
39980 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39981 (BT_FN_INT_PTR_INT): Likewise.
39982 (BT_FN_UINT_PTR_UINT): Likewise.
39983 (BT_FN_LONG_PTR_LONG): Likewise.
39984 (BT_FN_ULONG_PTR_ULONG): Likewise.
39985 (BT_FN_VOID_UINT64_UINT64): Likewise.
39986 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39987 (BT_FN_INT8_INT8_INT8): Likewise.
39988 (BT_FN_INT16_INT16_INT16): Likewise.
39989 (BT_FN_INT_INT_INT): Likewise.
39990 (BT_FN_UINT_FLOAT_UINT): Likewise.
39991 (BT_FN_FLOAT_UINT_UINT): Likewise.
39992 (BT_FN_ULONG_UINT_UINT): Likewise.
39993 (BT_FN_ULONG_UINT_PTR): Likewise.
39994 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39995 (BT_FN_UINT_UINT_UINT): Likewise.
39996 (BT_FN_VOID_UINT_PTR): Likewise.
39997 (BT_FN_UINT_UINT_PTR: Likewise.
39998 (BT_FN_UINT32_UINT64_PTR): Likewise.
39999 (BT_FN_INT_INT_UINT_UINT): Likewise.
40000 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
40001 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
40002 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
40003 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
40004 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
40005 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
40006 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
40007 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
40008 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
40009 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
40010 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
40011 * doc/frontends.texi: List BRIG FE.
40012 * doc/install.texi (Testing): Add BRIG tesring requirements.
40013 * doc/invoke.texi (Overall Options): Mention BRIG.
40014 * doc/standards.texi (Standards): Doucment BRIG HSA version.
40016 2017-01-24 Richard Biener <rguenther@suse.de>
40018 PR translation/79208
40019 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
40021 2017-01-24 Martin Jambor <mjambor@suse.cz>
40024 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
40025 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
40026 and known_contexts.
40028 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
40030 PR middle-end/79123
40031 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
40032 casts from signed to unsigned really don't have a range.
40034 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
40036 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
40037 GMP_RNDx for compatiblity.
40039 2017-01-24 Martin Liska <mliska@suse.cz>
40042 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
40043 that would prevent us to call alloca with -1 as argument.
40045 2017-01-24 Jakub Jelinek <jakub@redhat.com>
40047 * dwarf2out.c (output_compilation_unit_header, output_file_names):
40048 Avoid -Wformat-security warning.
40050 2017-01-23 Andrew Pinski <apinski@cavium.com>
40052 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
40055 2017-01-23 Martin Sebor <msebor@redhat.com>
40057 PR middle-end/78703
40058 * gimple-ssa-sprintf.c (warn_level): New global.
40059 (format_integer): Use it here and throughout the rest of the file.
40060 Use the same switch to compute sign as base.
40061 (maybe_warn): New function.
40062 (format_directive): Factor out warnings into maybe_warn.
40063 Add debugging output. Use warn_level.
40064 (add_bytes): Use warn_level.
40065 (pass_sprintf_length::compute_format_length): Add debugging output.
40066 (try_substitute_return_value): Same.
40067 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
40069 PR middle-end/78703
40070 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
40071 (struct fmtresult, format_integer, format_floating): Adjust.
40072 (fmtresult::fmtresult): Set max correctly in two argument ctor.
40073 (get_string_length, format_string,format_directive): Same.
40074 (pass_sprintf_length::compute_format_length): Same.
40075 (try_substitute_return_value): Simplify slightly.
40077 PR middle-end/78703
40078 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
40079 (fmtresult::operator+=): Outlined.
40080 (struct fmtresult): Add ctors.
40081 (struct conversion_spec): Rename...
40082 (struct directive): ...to this. Add and remove data members.
40083 (directive::set_width, directive::set_precision): New functions.
40084 (format_percent): Use fmtresult ctor.
40085 (get_width_and_precision): Remove.
40086 (format_integer): Make naming changes. Avoid computing width and
40088 (format_floating): Same. Adjust indentation.
40089 (format_character, format_none): New functions.
40090 (format_string): Moved character handling to format_character.
40091 (format_directive): Remove arguments, change return type.
40092 (parse_directive): New function.
40093 (pass_sprintf_length::compute_format_length): Move directive
40094 parsing to parse_directive.
40096 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40098 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
40099 (assign_assembler_name_if_needed): ... this.
40100 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
40101 (assign_assembler_name_if_needed): ... this.
40102 (free_lang_data_in_cgraph): Adjust callers.
40103 * cgraphunit.c (cgraph_node::analyze): Likewise.
40104 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
40107 2017-01-23 Richard Biener <rguenther@suse.de>
40109 PR tree-optimization/79088
40110 PR tree-optimization/79188
40111 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
40112 resetting loop bounds after last path deletion. Reset loop
40113 bounds of the target loop, make code match the comments.
40114 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
40115 Make sure loops need no fixups.
40117 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
40119 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
40120 exponent support with double type for first argument.
40121 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
40122 type returned by __builtin_vec_extract_sig,
40123 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
40124 functions from "vector int" to "vector unsigned int" or from
40125 "vector long long int" to "vector unsigned long long int".
40126 Changed type returned by __builtin_vec_extract_exp,
40127 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
40128 functions from "vector int" to "vector unsigned int" or from
40129 "vector long long int" to "vector unsigned long long int".
40130 Changed return type of __builtin_vec_test_data_class,
40131 __builtin_vec_test_data_class_sp, and
40132 __builtin_vec_test_data_class_dp from "vector int" to
40133 "vector bool int" or from "vector long long int" to "vector bool
40134 long long int" and changed second argument type from "unsigned
40135 int" to "int". Added new overloaded function forms "vector float
40136 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
40137 "vector float __builtin_vec_insert_exp_sp (vector float, vector
40138 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
40139 double, vector unsigned long long int)" and "vector double
40140 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
40141 long int)". Changed return type of
40142 __builtin_scalar_test_data_class and
40143 __builtin_scalar_test_data_class_sp and
40144 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
40145 int" and changed second argument from "unsigned int" to "int".
40146 Changed type returned by __builtin_scalar_test_neg,
40147 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
40148 from "int" to "bool int". Added new overloaded function form
40149 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
40150 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
40151 exponent double-precision with floating point first argument.
40152 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
40153 documentation of scalar_test_data_class, scalar_test_neg,
40154 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
40155 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
40156 vec_test_data_class built-in functions to reflect refinements in
40157 their type signatures.
40159 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
40161 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
40163 (aarch64_elf_asm_destructor): Likewise.
40165 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
40167 PR rtl-optimization/78634
40168 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
40169 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
40170 * ifcvt.c (noce_try_cmove): Add missing cost check.
40172 PR rtl-optimization/71724
40173 * combine.c (if_then_else_cond): Look for situations where it is
40174 beneficial to undo the work of one of the recursive calls.
40176 2017-01-23 Bin Cheng <bin.cheng@arm.com>
40178 PR tree-optimization/70754
40179 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
40180 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
40181 combined stmt before it if not NULL.
40182 (combine_chains): Process refs reversely and compute dominance point
40185 2017-01-23 Martin Liska <mliska@suse.cz>
40187 PR tree-optimization/79196
40188 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
40189 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
40191 (strlen_optimize_stmt): Call the renamed function.
40193 2017-01-23 Michael Matz <matz@suse.de>
40195 PR tree-optimization/78384
40196 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
40198 2017-01-23 Richard Biener <rguenther@suse.de>
40200 PR tree-optimization/79186
40201 * tree-vrp.c (register_new_assert_for): Make sure we've seen
40202 both incoming edges before moving an assert.
40204 2017-01-23 Martin Jambor <mjambor@suse.cz>
40206 * ipa-prop.c (load_from_param_1): Removed.
40207 (load_from_unmodified_param): Bits from load_from_param_1 put back
40209 (load_from_param): Removed.
40210 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
40211 with stmt. Reverted back to use of load_from_unmodified_param.
40213 2017-01-23 Martin Jambor <mjambor@suse.cz>
40216 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
40217 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
40218 field a pointer to garbage collected vector, mark lattices and
40219 ipcp_orig_node with GTY((skip)).
40220 (ipa_get_param_count): Adjust to descriptors being a pointer.
40221 (ipa_get_param): Likewise.
40222 (ipa_get_type): Likewise.
40223 (ipa_get_param_move_cost): Likewise.
40224 (ipa_set_param_used): Likewise.
40225 (ipa_get_controlled_uses): Likewise.
40226 (ipa_set_controlled_uses): Likewise.
40227 (ipa_is_param_used): Likewise.
40228 (ipa_node_params_t): Move into garbage collector. New methods insert
40230 (ipa_node_params_sum): Annotate wth GTY(()).
40231 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
40233 (ipa_load_from_parm_agg): Adjust declaration.
40234 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
40235 * ipa-profile.c (ipa_profile): Likewise.
40236 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
40237 (ipa_populate_param_decls): Make descriptors parameter garbage
40239 (ipa_dump_param): Adjust to descriptors being a pointer.
40240 (ipa_alloc_node_params): Likewise.
40241 (ipa_initialize_node_params): Likewise.
40242 (load_from_param_1): Make descriptors parameter garbage collected.
40243 (load_from_unmodified_param): Likewise.
40244 (load_from_param): Likewise.
40245 (ipa_load_from_parm_agg): Likewise.
40246 (ipa_node_params::~ipa_node_params): Removed.
40247 (ipa_free_all_node_params): Remove call to delete operator.
40248 (ipa_node_params_t::insert): New.
40249 (ipa_node_params_t::remove): Likewise.
40250 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
40251 copy known_csts and known_contexts vectors.
40252 (ipa_read_node_info): Adjust to descriptors being a pointer.
40253 (ipcp_modif_dom_walker): Make m_descriptors field garbage
40255 (ipcp_transform_function): Make descriptors variable garbage
40258 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
40260 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
40261 * config/i386/avx512dqintrin.h: Ditto.
40262 * config/i386/avx512fintrin.h: Ditto.
40263 * config/i386/i386.c: Handle new builtins.
40264 * config/i386/i386-builtin.def: Add new builtins.
40265 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
40266 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
40268 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40269 Martin Liska <mliska@suse.cz>
40271 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
40272 * asan.c (asan_expand_poison_ifn): Support stores and use
40273 appropriate ASAN report function.
40274 * internal-fn.c (expand_ASAN_POISON_USE): New function.
40275 * internal-fn.def (ASAN_POISON_USE): Declare.
40276 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
40277 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
40278 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
40279 ASAN_POISON calls w/o LHS.
40280 * tree-ssa.c (execute_update_addresses_taken): Create clobber
40281 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
40282 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
40283 * gimplify.c (asan_poison_variables): Add attribute
40284 use_after_scope_memory to variables that really needs to live
40286 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
40287 having the attribute.
40289 2017-01-23 Martin Liska <mliska@suse.cz>
40291 * asan.c (create_asan_shadow_var): New function.
40292 (asan_expand_poison_ifn): Likewise.
40293 * asan.h (asan_expand_poison_ifn): New declaration.
40294 * internal-fn.c (expand_ASAN_POISON): Likewise.
40295 * internal-fn.def (ASAN_POISON): New builtin.
40296 * sanopt.c (pass_sanopt::execute): Expand
40297 asan_expand_poison_ifn.
40298 * tree-inline.c (copy_decl_for_dup_finish): Make function
40300 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
40301 * tree-ssa.c (is_asan_mark_p): New function.
40302 (execute_update_addresses_taken): Rewrite local variables
40303 (identified just by use-after-scope as addressable) into SSA.
40305 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40307 * doc/install.texi (Specific): opensource.apple.com uses https
40308 now. Remove trailing slash.
40310 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40312 * README.Portability: Remove note on an Irix compatibility issue.
40314 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
40316 * gcov.c (INCLUDE_ALGORITHM): Define.
40317 (INCLUDE_VECTOR): Define.
40318 No longer include <vector> and <algorithm> directly.
40320 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
40322 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
40324 * doc/invoke.texi (Code Gen Options): Ditto.
40326 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
40329 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
40331 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
40333 rtl-optimization/79125
40334 * cprop.c (local_cprop_pass): Handle cases where we make an
40335 unconditional trap.
40337 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
40341 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
40342 read from, for big endian.
40344 2017-01-20 Jiong Wang <jiong.wang@arm.com>
40346 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
40347 register pauth builtins for LP64 only.
40349 2017-01-20 Marek Polacek <polacek@redhat.com>
40352 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
40355 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40357 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
40359 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
40360 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
40361 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
40363 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40366 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
40367 in big-endian mode when they are not a single duplicated value.
40369 2017-01-20 Richard Biener <rguenther@suse.de>
40371 * BASE-VER: Bump to 7.0.1.
40373 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40375 * omp-low.c (omplow_simd_context): New struct. Use it...
40376 (lower_rec_simd_input_clauses): ...here and...
40377 (lower_rec_input_clauses): ...here to hold common data. Adjust all
40378 references to idx, lane, max_vf, is_simt.
40380 2017-01-20 Graham Markall <graham.markall@embecosm.com>
40382 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
40385 2017-01-20 Martin Jambor <mjambor@suse.cz>
40387 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
40388 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
40390 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
40391 (GTFILES): Rename hsa.c to hsa-common.c.
40392 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
40393 * hsa-dump.c: Likewise.
40394 * hsa-gen.c: Likewise.
40395 * hsa-regalloc.c: Likewise.
40396 * ipa-hsa.c: Likewise.
40397 * omp-expand.c: Likewise.
40398 * omp-low.c: Likewise.
40399 * toplev.c: Likewise.
40401 2017-01-20 Marek Polacek <polacek@redhat.com>
40404 * doc/invoke.texi: Document -Wduplicated-branches.
40405 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
40406 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
40407 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
40408 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
40409 return 0 only when not OEP_LEXICOGRAPHIC.
40410 (fold_build_cleanup_point_expr): Use the expression
40411 location when building CLEANUP_POINT_EXPR.
40412 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
40413 * tree.c (add_expr): Handle error_mark_node.
40415 2017-01-20 Martin Liska <mliska@suse.cz>
40418 * tree-profile.c (init_ic_make_global_vars): Do not call
40420 (gimple_init_gcov_profiler): Likewise.
40422 2017-01-20 Martin Liska <mliska@suse.cz>
40425 * cgraph.h (maybe_create_reference): Remove argument and
40427 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
40429 * ipa-cp.c (create_specialized_node): Likewise.
40430 * symtab.c (symtab_node::maybe_create_reference): Handle
40431 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
40433 2017-01-20 Martin Liska <mliska@suse.cz>
40435 * read-rtl-function.c (function_reader::create_function): Use
40436 build_decl instread of build_decl_stat.
40438 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
40440 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
40441 * config/i386/avx512dqintrin.h: Ditto.
40442 * config/i386/avx512fintrin.h: Ditto.
40443 * config/i386/i386-builtin-types.def: Add new types.
40444 * config/i386/i386.c: Handle new types.
40445 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
40446 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
40447 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
40448 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
40449 (__builtin_ia32_kshiftridi): New.
40450 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
40452 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
40456 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
40457 define to rs6000_init_stack_protect_guard.
40458 (rs6000_init_stack_protect_guard): New function.
40460 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40461 Yunqiang Su <yunqiang.su@imgtec.com>
40463 * config.gcc (supported_defaults): Add madd4.
40464 (with_madd4): Add validation.
40465 (all_defaults): Add madd4.
40466 * config/mips/mips.opt (mmadd4): New option.
40467 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40469 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40471 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
40472 (ISA_HAS_FUSED_MADD4): Likewise.
40473 * doc/invoke.texi (-mmadd4): Document the new option.
40474 * doc/install.texi (--with-madd4): Document the new option.
40476 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40478 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
40479 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
40480 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
40481 (aarch64_init_pauth_hint_builtins): New.
40482 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
40483 (aarch64_expand_builtin): Expand new builtins.
40485 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40487 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
40488 * combine-stack-adj.c (no_unhandled_cfa): Handle
40489 REG_CFA_TOGGLE_RA_MANGLE.
40490 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
40491 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
40492 info for return address signing.
40493 (aarch64_expand_epilogue): Likewise.
40495 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40497 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
40498 * config/aarch64/aarch64-protos.h
40499 (aarch64_return_address_signing_enabled): New declaration.
40500 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
40502 (aarch64_expand_prologue): Sign return address before it's pushed onto
40504 (aarch64_expand_epilogue): Authenticate return address fetched from
40506 (aarch64_override_options): Sanity check for ILP32 and ISA level.
40507 (aarch64_attributes): New function attributes for "sign-return-address".
40508 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
40509 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
40510 ("*do_return"): Generate combined instructions according to key index.
40511 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
40512 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
40514 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
40515 * config/aarch64/aarch64.opt (msign-return-address=): New.
40516 * doc/extend.texi (AArch64 Function Attributes): Documents
40517 "sign-return-address=".
40518 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
40520 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40522 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
40523 overall option summary.
40525 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40527 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
40528 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
40529 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
40530 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
40532 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
40534 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
40535 -mpower9-minmax by default for -mcpu=power9.
40536 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
40537 128-bit floating point.
40539 2017-01-20 Alan Modra <amodra@gmail.com>
40541 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
40542 optimizing for size.
40544 2017-01-20 Alan Modra <amodra@gmail.com>
40547 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
40548 for strcmp and strncmp from corresponding builtin decl.
40550 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40552 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
40553 instead of i386/rtems-64.h.
40554 * config/i386/rtems-64.h: Remove.
40556 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40560 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
40562 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
40564 2017-01-19 Tamar Christina <tamar.christina@arm.com>
40566 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
40567 Change int to HOST_WIDE_INT.
40568 * config/aarch64/aarch64-protos.h
40569 (aarch64_simd_gen_const_vector_dup): Likewise.
40570 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
40572 2017-01-19 David Malcolm <dmalcolm@redhat.com>
40574 * langhooks-def.h (lhd_type_for_size): New decl.
40575 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
40576 * langhooks.c (lhd_type_for_size): New function, taken from
40579 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
40581 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
40582 define_bypass for CR latency.
40583 (power9-cracked-alu): Update bypass latency and remove power9-branch.
40584 (power9-alu2): Add define_bypass for CR latency.
40586 (power9-mul): Update insn latency.
40587 (power9-mul-compare): Update insn latency, bypass latency and remove
40590 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40592 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
40594 * config/aarch64/aarch64.md
40595 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
40596 aarch64_nopcrelative_literal_loads.
40597 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
40599 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
40601 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
40602 TARGET_LOONGSON_3A.
40603 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
40605 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
40608 * config.gcc (supported_defaults): Add lxc1-sxc1.
40609 (with_lxc1_sxc1): Add validation.
40610 (all_defaults): Add lxc1-sxc1.
40611 * config/mips/mips.opt (mlxc1-sxc1): New option.
40612 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40614 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40615 __mips_no_lxc1_sxc1.
40616 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
40617 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
40618 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
40620 2017-01-19 Richard Biener <rguenther@suse.de>
40622 PR tree-optimization/72488
40623 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
40624 sure to restore SSA info.
40625 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
40627 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
40629 PR rtl-optimization/79121
40630 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
40631 of the inner type when shifting an extended value.
40633 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40636 * symtab.c (symtab_node::equal_address_to): Fix comparing of
40637 interposable aliases.
40639 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
40642 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
40643 Use the evmergelohi instruction.
40644 (mov_si<mode>_e500_subreg4_2_le): Likewise.
40645 (mov_sitf_e500_subreg8_2_be): Likewise.
40646 (mov_sitf_e500_subreg12_2_le): Likewise.
40647 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
40648 (mov_si<mode>_e500_subreg4_2_be): Likewise.
40649 (mov_sitf_e500_subreg8_2_le): Likewise.
40650 (mov_sitf_e500_subreg12_2_be): Likewise.
40652 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40654 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
40655 attribute from vecsimple to vecperm.
40656 (altivec_vbpermq2): Likewise.
40658 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40661 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
40663 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
40664 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
40665 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
40666 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
40667 case where N arg is SIZE_MAX.
40668 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
40669 (cmpstrsi): Add pattern.
40671 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
40673 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
40674 __builtin_vec_revb builtins.
40675 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
40676 built-in functions to support generation of the ISA 3.0 XXBR<x>
40677 vector byte reverse instructions.
40678 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
40679 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
40680 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
40681 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
40682 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
40683 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
40684 (P9V_BUILTIN_VEC_REVB): Likewise.
40685 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
40686 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
40687 (p9_xxbrq_v16qi): Likewise.
40688 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
40689 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
40690 (p9_xxbrh_v8hi): Likewise.
40691 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
40692 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
40693 vec_revb built-in functions.
40695 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
40697 PR rtl-optimization/78952
40698 * config/i386/i386.md (any_extract): New code iterator.
40699 (*insvqi_2): Use any_extract for source operand.
40700 (*insvqi_3): Use any_shiftrt for source operand.
40702 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
40704 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
40706 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
40708 2017-01-18 Matthias Klose <doko@ubuntu.com>
40710 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
40712 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40714 * config/rs6000/altivec.h (vec_bperm): Change #define.
40715 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
40716 (altivec_vbpermq2): New define_insn.
40717 (altivec_vbpermd): Likewise.
40718 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
40719 function interface.
40720 (VBPERMD): Likewise.
40721 (VBPERM): New polymorphic function interface.
40722 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
40723 Add entries for P9V_BUILTIN_VEC_VBPERM.
40724 * doc/extend.texi: Add interfaces for vec_bperm.
40726 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40728 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
40729 first letter of error messages.
40730 (s390_resolve_overloaded_builtin): Likewise.
40731 * config/s390/s390.c (s390_expand_builtin): Likewise.
40732 (s390_invalid_arg_for_unprototyped_fn): Likewise.
40733 (s390_valid_target_attribute_inner_p): Likewise.
40734 * config/s390/s390.md ("tabort"): Likewise.
40736 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
40738 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
40739 (ISA_AVOID_DIV_HILO): New macro.
40740 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
40741 (ISA_HAS_DDIV): Likewise.
40743 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40745 * doc/invoke.texi (fabi-version): Correct number of occurrences.
40747 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40749 * doc/invoke.texi (fabi-version): Spelling fix.
40751 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40754 * doc/invoke.texi (fabi-version): Mention mangling fix for
40757 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40760 * doc/invoke.texi (fabi-version): Document discriminator mangling.
40762 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
40765 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
40766 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
40768 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
40770 (stack_protect_test): Ditto.
40771 * config/rs6000/rs6000.opt (mstack-protector-guard=,
40772 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
40774 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40775 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40776 -mstack-protector-guard-offset=.
40777 (RS/6000 and PowerPC Options): Ditto.
40779 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40781 * config/i386/i386.h (MASK_CLASS_P): New define.
40782 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40783 there are no registers from different register sets also when
40784 mask registers are used. Update function comment.
40785 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40786 to (*k/*r) and (*k/*km) alternatives.
40788 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40790 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40791 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40792 (EH_RETURN_HANDLER_RTX): New define.
40793 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40794 Force frame pointer in EH return functions.
40795 (aarch64_expand_epilogue): Add barrier for eh_return.
40796 (aarch64_final_eh_return_addr): Remove.
40797 (aarch64_eh_return_handler_rtx): New function.
40798 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40800 (aarch64_eh_return_handler_rtx): New prototype.
40802 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40804 * config/rs6000/altivec.h (vec_rlmi): New #define.
40805 (vec_vrlnm): Likewise.
40806 (vec_rlnm): Likewise.
40807 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40808 (UNSPEC_VRLNM): Likewise.
40809 (VIlong): New mode iterator.
40810 (altivec_vrl<VI_char>mi): New define_insn.
40811 (altivec_vrl<VI_char>nm): Likewise.
40812 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40814 (VRLDNM): Likewise.
40815 (RLNM): New polymorphic function entry.
40816 (VRLWMI): New monomorphic function entry.
40817 (VRLDMI): Likewise.
40818 (RLMI): New polymorphic function entry.
40819 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40820 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40821 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40824 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40827 * dwarf2out.c (field_byte_offset): Restore the
40828 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40829 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40830 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40833 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40836 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40838 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40841 * dwarf2out.c (add_data_member_location_attribute): For constant
40842 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40843 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40844 DW_AT_byte_size attributes.
40846 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40848 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40849 after forcing to constant memory when the code model is medium.
40851 2017-01-17 Julia Koval <julia.koval@intel.com>
40854 * config/i386/avx512fintrin.h
40855 (_mm512_i32gather_ps): Change __addr type to void const*.
40856 (_mm512_mask_i32gather_ps): Ditto.
40857 (_mm512_i32gather_pd): Ditto.
40858 (_mm512_mask_i32gather_pd): Ditto.
40859 (_mm512_i64gather_ps): Ditto.
40860 (_mm512_mask_i64gather_ps): Ditto.
40861 (_mm512_i64gather_pd): Ditto.
40862 (_mm512_mask_i64gather_pd): Ditto.
40863 (_mm512_i32gather_epi32): Ditto.
40864 (_mm512_mask_i32gather_epi32): Ditto.
40865 (_mm512_i32gather_epi64): Ditto.
40866 (_mm512_mask_i32gather_epi64): Ditto.
40867 (_mm512_i64gather_epi32): Ditto.
40868 (_mm512_mask_i64gather_epi32): Ditto.
40869 (_mm512_i64gather_epi64): Ditto.
40870 (_mm512_mask_i64gather_epi64): Ditto.
40871 (_mm512_i32scatter_ps): Change __addr type to void*.
40872 (_mm512_mask_i32scatter_ps): Ditto.
40873 (_mm512_i32scatter_pd): Ditto.
40874 (_mm512_mask_i32scatter_pd): Ditto.
40875 (_mm512_i64scatter_ps): Ditto.
40876 (_mm512_mask_i64scatter_ps): Ditto.
40877 (_mm512_i64scatter_pd): Ditto.
40878 (_mm512_mask_i64scatter_pd): Ditto.
40879 (_mm512_i32scatter_epi32): Ditto.
40880 (_mm512_mask_i32scatter_epi32): Ditto.
40881 (_mm512_i32scatter_epi64): Ditto.
40882 (_mm512_mask_i32scatter_epi64): Ditto.
40883 (_mm512_i64scatter_epi32): Ditto.
40884 (_mm512_mask_i64scatter_epi32): Ditto.
40885 (_mm512_i64scatter_epi64): Ditto.
40886 (_mm512_mask_i64scatter_epi64): Ditto.
40887 * config/i386/avx512pfintrin.h
40888 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40889 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40890 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40891 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40892 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40893 (_mm512_prefetch_i32scatter_ps): Ditto.
40894 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40895 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40896 (_mm512_prefetch_i64scatter_pd): Ditto.
40897 (_mm512_prefetch_i64scatter_ps): Ditto.
40898 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40899 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40900 * config/i386/avx512vlintrin.h
40901 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40902 (_mm_mmask_i32gather_ps): Ditto.
40903 (_mm256_mmask_i32gather_pd): Ditto.
40904 (_mm_mmask_i32gather_pd): Ditto.
40905 (_mm256_mmask_i64gather_ps): Ditto.
40906 (_mm_mmask_i64gather_ps): Ditto.
40907 (_mm256_mmask_i64gather_pd): Ditto.
40908 (_mm_mmask_i64gather_pd): Ditto.
40909 (_mm256_mmask_i32gather_epi32): Ditto.
40910 (_mm_mmask_i32gather_epi32): Ditto.
40911 (_mm256_mmask_i32gather_epi64): Ditto.
40912 (_mm_mmask_i32gather_epi64): Ditto.
40913 (_mm256_mmask_i64gather_epi32): Ditto.
40914 (_mm_mmask_i64gather_epi32): Ditto.
40915 (_mm256_mmask_i64gather_epi64): Ditto.
40916 (_mm_mmask_i64gather_epi64): Ditto.
40917 (_mm256_i32scatter_ps): Change __addr type to void*.
40918 (_mm256_mask_i32scatter_ps): Ditto.
40919 (_mm_i32scatter_ps): Ditto.
40920 (_mm_mask_i32scatter_ps): Ditto.
40921 (_mm256_i32scatter_pd): Ditto.
40922 (_mm256_mask_i32scatter_pd): Ditto.
40923 (_mm_i32scatter_pd): Ditto.
40924 (_mm_mask_i32scatter_pd): Ditto.
40925 (_mm256_i64scatter_ps): Ditto.
40926 (_mm256_mask_i64scatter_ps): Ditto.
40927 (_mm_i64scatter_ps): Ditto.
40928 (_mm_mask_i64scatter_ps): Ditto.
40929 (_mm256_i64scatter_pd): Ditto.
40930 (_mm256_mask_i64scatter_pd): Ditto.
40931 (_mm_i64scatter_pd): Ditto.
40932 (_mm_mask_i64scatter_pd): Ditto.
40933 (_mm256_i32scatter_epi32): Ditto.
40934 (_mm256_mask_i32scatter_epi32): Ditto.
40935 (_mm_i32scatter_epi32): Ditto.
40936 (_mm_mask_i32scatter_epi32): Ditto.
40937 (_mm256_i32scatter_epi64): Ditto.
40938 (_mm256_mask_i32scatter_epi64): Ditto.
40939 (_mm_i32scatter_epi64): Ditto.
40940 (_mm_mask_i32scatter_epi64): Ditto.
40941 (_mm256_i64scatter_epi32): Ditto.
40942 (_mm256_mask_i64scatter_epi32): Ditto.
40943 (_mm_i64scatter_epi32): Ditto.
40944 (_mm_mask_i64scatter_epi32): Ditto.
40945 (_mm256_i64scatter_epi64): Ditto.
40946 (_mm256_mask_i64scatter_epi64): Ditto.
40947 (_mm_i64scatter_epi64): Ditto.
40948 (_mm_mask_i64scatter_epi64): Ditto.
40949 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40950 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40951 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40952 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40953 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40954 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40955 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40956 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40957 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40958 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40959 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40960 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40961 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40962 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40963 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40964 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40965 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40966 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40967 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40968 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40969 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40970 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40971 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40972 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40973 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40974 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40975 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40976 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40977 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40978 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40979 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40980 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40981 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40982 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40983 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40984 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40985 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40986 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40987 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40988 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40989 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40990 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40991 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40992 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40993 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40994 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40995 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40996 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40997 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40998 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40999 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
41000 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
41001 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
41002 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
41003 definitions accordingly.
41005 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
41006 Kuan-Lin Chen <kuanlinchentw@gmail.com>
41009 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
41012 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
41015 * ira-conflicts.c (ira_build_conflicts): Update total conflict
41016 hard regs for inner regno.
41018 2017-01-17 Martin Liska <mliska@suse.cz>
41021 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
41022 assumption and add comment.
41024 2017-01-17 Nathan Sidwell <nathan@acm.org>
41026 * ipa-visibility.c (localize_node): New function, broken out of ...
41027 (function_and_variable_visibility): ... here. Call it.
41029 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
41031 PR middle-end/77445
41032 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
41033 correctly set frequency of oudgoing edge.
41034 (duplicate_thread_path): Fix profile updating.
41036 2017-01-17 Jakub Jelinek <jakub@redhat.com>
41039 * configure.ac: Add GCC_BASE_VER.
41040 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
41041 version from BASE-VER file.
41042 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
41043 (gcc.o): Depend on $(BASEVER).
41044 * common.opt (dumpfullversion): New option.
41045 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
41046 * doc/invoke.texi: Document -dumpfullversion.
41047 * doc/install.texi: Document --with-gcc-major-version-only.
41048 * configure: Regenerated.
41050 2017-01-17 Richard Biener <rguenther@suse.de>
41052 PR tree-optimization/71433
41053 * tree-vrp.c (register_new_assert_for): Merge same asserts
41054 on all incoming edges.
41055 (process_assert_insertions_for): Handle insertions at the
41058 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
41060 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
41061 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
41063 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
41066 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
41069 2017-01-17 Alan Modra <amodra@gmail.com>
41072 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
41073 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
41074 symbolic stack limit when pic.
41076 2017-01-16 Martin Sebor <msebor@redhat.com>
41078 PR tree-optimization/78608
41079 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
41081 2017-01-16 Jeff Law <law@redhat.com>
41084 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
41085 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
41086 for several include directories that may be relative to sysroot.
41087 * config/i386/x-mingw32 (gplus_includedir): Define.
41088 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
41089 (native_system_includedir): Likewise.
41090 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
41091 override if TARGET_SYSTEM_ROOT is defined.
41092 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
41094 PR tree-optimization/79090
41095 PR tree-optimization/33562
41096 PR tree-optimization/61912
41097 PR tree-optimization/77485
41098 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
41099 and computed trims into the dump file.
41101 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
41103 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
41105 2017-01-16 Jakub Jelinek <jakub@redhat.com>
41108 * gimplify.c (gimplify_init_constructor): If want_value and
41109 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
41113 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
41114 sequence. Formatting fixes.
41115 (doloop_optimize): Formatting fixes.
41118 * gcc.c (debug_level_greater_than_spec_func): New function.
41119 (static_spec_functions): Add debug-level-gt spec function.
41120 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
41122 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
41123 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
41124 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
41125 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
41126 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
41127 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
41129 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
41131 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
41132 QImode fixups to general and mask registers only.
41134 2017-01-16 Carl Love <cel@us.ibm.com>
41136 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
41137 for built-in functions
41138 vector signed char vec_nabs (vector signed char)
41139 vector signed short vec_nabs (vector signed short)
41140 vector signed int vec_nabs (vector signed int)
41141 vector signed long long vec_nabs (vector signed long long)
41142 vector float vec_nabs (vector float)
41143 vector double vec_nabs (vector double)
41144 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
41146 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
41147 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
41148 * doc/extend.texi: Update the documentation file for the new built-in
41151 2017-01-16 Martin Sebor <msebor@redhat.com>
41153 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
41156 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41158 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
41159 UNSPEC_VSX__XXSPLTD to require special splat handling.
41161 2017-01-16 David Malcolm <dmalcolm@redhat.com>
41164 * system.h: Poison strndup.
41166 2017-01-16 Alan Modra <amodra@gmail.com>
41169 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
41172 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
41174 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
41176 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
41178 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
41179 call recog here. Assert that INSN_CODE (insn) is non-negative.
41181 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41184 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
41186 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
41187 in the currently scheduled RTL fragment.
41189 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41191 PR rtl-optimization/78751
41192 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
41195 2017-01-14 Jeff Law <law@redhat.com>
41197 PR tree-optimization/79090
41198 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
41199 variable length stores.
41200 (compute_trims): Delete dead assignment to *trim_tail.
41201 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
41204 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
41206 PR rtl-optimization/78626
41207 PR rtl-optimization/78727
41208 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
41209 of a block, and split such blocks after everything else is finished.
41211 2017-01-14 Alan Modra <amodra@gmail.com>
41214 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
41215 target legitimate_combined_insn.
41216 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
41217 (rs6000_legitimate_combined_insn): New function.
41218 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
41220 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
41221 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
41222 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
41224 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
41226 * doc/frontends.texi (G++ and GCC): Remove references to Java.
41228 2017-01-13 Jeff Law <law@redhat.com>
41230 PR tree-optimization/33562
41231 PR tree-optimization/61912
41232 PR tree-optimization/77485
41233 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
41235 (delete_dead_assignment): Likewise.
41236 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
41237 statement to delete_dead_call and delete_dead_assignment.
41239 2017-01-13 David Malcolm <dmalcolm@redhat.com>
41242 * substring-locations.c (format_warning_va): Strengthen case 1 so
41243 that both endpoints of the substring must be within the format
41244 range for just the substring to be printed.
41246 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
41248 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
41249 * config/i386/i386.c (ix86_target_string): Add missing options
41250 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
41251 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
41252 flags_other and ix86_target_other to flags2_other. Display unknown
41254 (ix86_valid_target_attribute_inner_p): Add missing options and
41255 reorder options by implied ISAs, as in ix86_target_string.
41257 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41259 * hash-table.h (hash_table::too_empty_p): New function.
41260 (hash_table::expand): Use it.
41261 (hash_table::traverse): Likewise.
41262 (hash_table::empty_slot): Use sizeof (value_type) instead of
41263 sizeof (PTR) to convert bytes to elements. Shrink the table
41264 if the current size is excessive for the current number of
41267 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41269 * ira-costs.c (record_reg_classes): Break from the inner loop
41270 early once alt_fail is known to be true. Update outer loop
41271 handling accordingly.
41273 2017-01-13 Jeff Law <law@redhat.com>
41275 * tree-ssa-dse.c (decrement_count): New function.
41276 (increment_start_addr, maybe_trim_memstar_call): Likewise.
41277 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
41278 when we know the partially dead statement is a mem* function.
41280 PR tree-optimization/61912
41281 PR tree-optimization/77485
41282 * tree-ssa-dse.c: Include expr.h.
41283 (maybe_trim_constructor_store): New function.
41284 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
41286 PR tree-optimization/33562
41287 PR tree-optimization/61912
41288 PR tree-optimization/77485
41289 * doc/invoke.texi: Document new dse-max-object-size param.
41290 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
41291 * tree-ssa-dse.c: Include params.h.
41292 (dse_store_status): New enum.
41293 (initialize_ao_ref_for_dse): New, partially extracted from
41295 (valid_ao_ref_for_dse, normalize_ref): New.
41296 (setup_live_bytes_from_ref, compute_trims): Likewise.
41297 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
41298 (maybe_trim_partially_dead_store): Likewise.
41299 (maybe_trim_complex_store): Likewise.
41300 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
41301 Track what bytes live from the original store. Return tri-state
41302 for dead, partially dead or live.
41303 (dse_dom_walker): Add constructor, destructor and new private members.
41304 (delete_dead_call, delete_dead_assignment): New extracted from
41306 (dse_optimize_stmt): Make a member of dse_dom_walker.
41307 Use initialize_ao_ref_for_dse.
41309 PR tree-optimization/33562
41310 PR tree-optimization/61912
41311 PR tree-optimization/77485
41312 * sbitmap.h (bitmap_count_bits): Prototype.
41313 (bitmap_clear_range, bitmap_set_range): Likewise.
41314 * sbitmap.c (bitmap_clear_range): New function.
41315 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
41317 2017-01-13 Martin Liska <mliska@suse.cz>
41320 * function.c (set_cfun): Add new argument force.
41321 * function.h (set_cfun): Likewise.
41322 * ipa-inline-transform.c (inline_call): Use the function when
41323 strict alising from is dropped for function we inline to.
41325 2017-01-13 Richard Biener <rguenther@suse.de>
41327 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
41328 for dumping GIMPLE INTEGER_CSTs.
41330 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
41332 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
41333 to 201112L since C++17.
41335 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
41338 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
41339 if -fsanitize=kernel-address is present.
41341 2017-01-13 Richard Biener <rguenther@suse.de>
41343 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
41344 as _Literal ( type ) number in case usual suffixes do not
41345 preserve all information.
41347 2017-01-13 Richard Biener <rguenther@suse.de>
41349 PR tree-optimization/77283
41350 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
41351 and ssa-iterators.h.
41352 (is_feasible_trace): Implement a cost model based on joiner
41355 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
41358 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
41359 char or short to __float128/_Float128 directly.
41361 2017-01-12 Martin Sebor <msebor@redhat.com>
41363 to -Wformat-overflow.
41364 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
41365 (min_bytes_remaining): Same.
41366 (get_string_length): Same.
41367 (format_string): Same.
41368 (format_directive): Same.
41370 (pass_sprintf_length::handle_gimple_call): Same.
41372 2017-01-12 Jakub Jelinek <jakub@redhat.com>
41374 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
41375 info.nowrite calls with no lhs that can't throw. Return bool
41376 whether gsi_remove has been called or not.
41377 (pass_sprintf_length::handle_gimple_call): Return bool whether
41378 try_substitute_return_value called gsi_remove. Formatting fix.
41379 (pass_sprintf_length::execute): Don't use gsi_remove if
41380 handle_gimple_call returned true.
41383 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
41384 be removed due to side-effects, don't remove following barrier nor
41385 turn the successor edge into fallthru edge.
41387 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41390 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
41391 element-reversing loads and stores as not swappable.
41393 2017-01-12 Nathan Sidwell <nathan@acm.org>
41394 Nicolai Stange <nicstange@gmail.com>
41396 * combine.c (try_combine): Don't ignore result of overlap checking
41397 loop. Combine overlap & asm check into single loop.
41399 2017-01-12 Richard Biener <rguenther@suse.de>
41401 * tree-pretty-print.c (dump_generic_node): Provide -gimple
41402 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
41404 2017-01-12 Richard Biener <rguenther@suse.de>
41406 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
41407 and TS_TARGET_OPTION directly derive from TS_BASE.
41408 * tree-core.h (tree_optimization_option): Derive from tree_base.
41409 (tree_target_option): Likewise.
41411 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
41413 * config/i386/i386.c (memory_address_length): Increase len
41414 only when rip_relative_addr_p returns false.
41416 2017-01-11 Julia Koval <julia.koval@intel.com>
41418 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
41419 (OPTION_MASK_ISA_SGX_SET): New.
41420 (ix86_handle_option): Handle OPT_msgx.
41421 * config.gcc: Added sgxintrin.h.
41422 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
41423 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
41424 * config/i386/i386.c (ix86_target_string): Add -msgx.
41426 (ix86_option_override_internal): Handle new options.
41427 (ix86_valid_target_attribute_inner_p): Add sgx.
41428 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
41429 * config/i386/i386.opt: Add msgx.
41430 * config/i386/sgxintrin.h: New file.
41431 * config/i386/x86intrin.h: Add sgxintrin.h.
41433 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41436 * fold-const.c (maybe_nonzero_address): Return 1 for function
41438 (tree_single_nonzero_warnv_p): Don't handle function local objects
41442 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
41445 2017-01-11 David Malcolm <dmalcolm@redhat.com>
41448 * opts.c: Include "spellcheck.h"
41449 (struct string_fragment): New struct.
41450 (struct edit_distance_traits<const string_fragment &>): New
41452 (get_closest_sanitizer_option): New function.
41453 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
41455 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41457 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
41459 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
41460 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
41461 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
41462 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
41463 for initial die_offset if dwarf_split_debug_info.
41464 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
41465 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
41467 (output_skeleton_debug_sections): Formatting fix. Use
41468 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
41469 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
41471 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
41473 * config/arm/cortex-a53.md: Add bypasses for
41474 cortex_a53_r2f_cvt.
41475 (cortex_a53_r2f): Only use for transfers.
41476 (cortex_a53_f2r): Likewise.
41477 (cortex_a53_r2f_cvt): Add reservation for conversions.
41478 (cortex_a53_f2r_cvt): Likewise.
41480 2017-01-11 Tamar Christina <tamar.christina@arm.com>
41482 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
41483 to all inlined functions, change static to extern.
41485 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
41488 * config/arm/arm.c (legitimize_pic_address): Handle reference to
41490 (arm_assemble_integer): Likewise.
41492 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
41494 * config.gcc: Use new awk script to check CPU, FPU and architecture
41495 parameters for --with-... options.
41496 * config/arm/parsecpu.awk: New file
41497 * config/arm/arm-cpus.in: New file.
41498 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
41500 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
41502 * config/arm/t-arm: Update dependency rules.
41503 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
41504 of processing .def files.
41505 * config/arm/genopt.sh: Deleted.
41506 * config/arm/gentune.sh: Deleted.
41507 * config/arm/arm-cores.def: Deleted.
41508 * config/arm/arm-arches.def: Deleted.
41509 * config/arm/arm-fpus.def: Deleted.
41510 * config/arm/arm-tune.md: Regenerated.
41511 * config/arm/arm-tables.opt: Regenerated.
41512 * config/arm/arm-cpu.h: New generated file.
41513 * config/arm/arm-cpu-data.h: New generated file.
41514 * config/arm/arm-cpu-cdata.h: New generated file.
41516 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
41519 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
41521 (input_varpool_node): Unpack dynamically_initialized bit.
41523 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
41525 PR rtl-optimization/79032
41526 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
41527 the alignment of the adjusted memory reference against that of MODE,
41528 instead of the alignment of the original memory reference.
41530 2017-01-11 Martin Jambor <mjambor@suse.cz>
41532 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
41534 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
41535 decorated functions.
41537 2017-01-11 Richard Biener <rguenther@suse.de>
41539 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
41540 set range/nonnull info for PHI results. Do not set it on
41541 stmts marked for removal.
41543 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
41545 * expr.c (store_field): In the bitfield case, fetch the return value
41546 from the registers before applying a single big-endian adjustment.
41547 Always do a final load for a BLKmode value not larger than a word.
41549 2017-01-10 David Malcolm <dmalcolm@redhat.com>
41552 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41553 that we correctly handle column numbers greater than
41554 LINE_MAP_MAX_COLUMN_NUMBER.
41556 2017-01-10 Martin Sebor <msebor@redhat.com>
41558 PR middle-end/78245
41559 * gimple-ssa-sprintf.c (get_destination_size): Call
41560 {init,fini}object_sizes.
41561 * tree-object-size.c (addr_object_size): Adjust.
41562 (pass_through_call): Adjust.
41563 (pass_object_sizes::execute): Adjust.
41564 * tree-object-size.h (fini_object_sizes): Declare.
41566 2017-01-10 Martin Sebor <msebor@redhat.com>
41568 PR tree-optimization/78775
41569 * builtins.c (get_size_range): Move...
41570 * calls.c: ...to here.
41571 (alloc_max_size): Accept zero argument.
41572 (operand_signed_p): Remove.
41573 (maybe_warn_alloc_args_overflow): Call get_size_range.
41574 * calls.h (get_size_range): Declare.
41576 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
41578 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
41579 from TI's devices.csv file as of September 2016.
41580 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
41582 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
41584 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
41585 * doc/invoke.texi: Likewise.
41586 * doc/md.texi: Likewise.
41587 * doc/objc.texi: Likewise.
41589 2017-01-10 Joshua Conner <joshconner@google.com>
41591 * config/arm/fuchsia-elf.h: New file.
41592 * config/fuchsia.h: New file.
41593 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
41594 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
41596 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
41598 2016-01-10 Richard Biener <rguenther@suse.de>
41600 PR tree-optimization/79034
41601 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
41602 Propagate out degenerate PHIs in the joiner.
41604 2017-01-10 Martin Liska <mliska@suse.cz>
41606 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
41607 (sort_congruence_classes_by_decl_uid): Likewise.
41608 (sort_congruence_class_groups_by_decl_uid): Likewise.
41609 (sem_item_optimizer::merge_classes): Sort class, groups in these
41610 classes and members in the groups by DECL_UID of declarations.
41611 This would make merge operations stable.
41613 2017-01-10 Martin Liska <mliska@suse.cz>
41615 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
41616 usage of m_classes_vec.
41617 (sem_item_optimizer::~sem_item_optimizer): Likewise.
41618 (sem_item_optimizer::get_group_by_hash): Likewise.
41619 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
41620 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
41621 (sem_item_optimizer::verify_classes): Likewise.
41622 (sem_item_optimizer::process_cong_reduction): Likewise.
41623 (sem_item_optimizer::dump_cong_classes): Likewise.
41624 (sem_item_optimizer::merge_classes): Likewise.
41625 * ipa-icf.h (congruence_class_hash): Rename from
41626 congruence_class_group_hash. Remove declaration of m_classes_vec.
41628 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
41630 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
41631 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
41632 * config.gcc: Add avx512vpopcntdqintrin.h.
41633 * config/i386/avx512vpopcntdqintrin.h: New.
41634 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
41635 * config/i386/i386-builtin-types.def: Add new types.
41636 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
41637 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
41638 __builtin_ia32_vpopcountq_v8di_mask): New.
41639 * config/i386/i386-c.c (ix86_target_macros_internal): Define
41640 __AVX512VPOPCNTDQ__.
41641 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
41642 (PTA_AVX512VPOPCNTDQ): Define.
41643 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
41644 TARGET_AVX512VPOPCNTDQ_P): Define.
41645 * config/i386/i386.opt: Add mavx512vpopcntdq.
41646 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
41647 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
41649 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41651 PR middle-end/77484
41652 * predict.def (PRED_CALL): Set to 67.
41654 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
41656 * expr.c (store_field): In the bitfield case, if the value comes from
41657 a function call and is of an aggregate type returned in registers, do
41658 not modify the field mode; extract the value in all cases if the mode
41659 is BLKmode and the size is not larger than a word.
41661 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
41664 * config/i386/cpuid.h: Fix undefined behavior.
41666 2017-01-04 Jeff Law <law@redhat.com>
41668 PR tree-optimization/79007
41669 PR tree-optimization/67955
41670 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
41671 conservative for pt.null when flag_non_call_exceptions is on.
41673 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41675 PR translation/79019
41676 PR translation/79020
41677 * params.def (PARAM_INLINE_MIN_SPEEDUP,
41678 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
41679 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
41681 * config/avr/avr.opt (maccumulate-args): Likewise.
41682 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
41683 * common.opt (freport-bug): Likewise.
41684 * cif-code.def (CIF_FINAL_ERROR): Likewise.
41685 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
41686 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
41687 translatable string.
41688 * config/i386/i386.c (function_value_32): Likewise.
41689 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
41690 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
41692 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
41693 * common/config/msp430/msp430-common.c (msp430_handle_option):
41695 * symtab.c (symtab_node::verify_base): Likewise.
41696 * opts.c (set_debug_level): Likewise.
41697 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
41698 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
41699 missing whitespace to translatable strings.
41700 * config/avr/avr.md (bswapsi2): Fix typo in comment.
41701 * config/sh/superh.h: Likewise.
41702 * config/i386/xopintrin.h: Likewise.
41703 * config/i386/znver1.md: Likewise.
41704 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
41705 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
41706 * double-int.h (struct double_int): Likewise.
41707 * double-int.c (div_and_round_double): Likewise.
41708 * wide-int.cc: Likewise.
41709 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
41710 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
41711 * cfgcleanup.c (crossjumps_occured): Renamed to ...
41712 (crossjumps_occurred): ... this.
41713 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
41716 PR tree-optimization/78899
41717 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
41718 returning bool return struct loop *, NULL for failure and the new
41720 (versionable_outer_loop_p): Don't version outer loop if it has
41721 dont_vectorized bit set.
41722 (tree_if_conversion): When versioning outer loop, ensure
41723 tree_if_conversion is performed also on the inner loop of the
41724 non-vectorizable outer loop copy.
41725 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
41726 LOOP_VECTORIZED in inner loop of the scalar outer loop and
41727 prevent vectorization of it.
41728 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
41729 the outer loop vectorization of the non-scalar version is attempted
41730 before vectorization of the inner loop in scalar version. If
41731 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
41732 vectorization of its inner loop.
41733 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
41734 has 2 inner loops, rename also on edges from bb whose single pred
41735 is outer_loop->header. Fix typo in function comment.
41737 2017-01-09 Martin Sebor <msebor@redhat.com>
41740 * asan.c (asan_emit_stack_protection): Increase local buffer size
41741 to avoid snprintf truncation warning.
41743 2017-01-09 Andrew Pinski <apinski@cavium.com>
41745 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
41746 to reference thunderx2t99 for the tuning structure
41747 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
41749 (thunderx2t99_extra_costs): This.
41750 * config/aarch64/aarch64-tune.md: Regenerate.
41751 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
41752 (vulcan_addrcost_table): This.
41753 (vulcan_regmove_cost): Rename to ...
41754 (thunderx2t99_regmove_cost): This.
41755 (vulcan_vector_cost): Rename to ...
41756 (thunderx2t99_vector_cost): this.
41757 (vulcan_branch_cost): Rename to ...
41758 (thunderx2t99_branch_cost): This.
41759 (vulcan_tunings): Rename to ...
41760 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
41761 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
41763 2017-01-09 Martin Jambor <mjambor@suse.cz>
41767 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
41768 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
41769 (propagate_vr_accross_jump_function): Use the above function for all
41770 value range computations for pass-through jump functions and type
41771 converasion from explicit value range values.
41772 (ipcp_propagate_stage): Do not attempt to deduce types of formal
41773 parameters from TYPE_ARG_TYPES.
41774 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41775 (ipa_write_node_info): Stream type of the actual argument.
41776 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41778 2017-01-09 Martin Liska <mliska@suse.cz>
41781 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41782 (lookup_compiler): Do not show error message with have_E.
41784 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41786 PR tree-optimization/78938
41787 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41788 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41789 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41790 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41793 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41795 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41798 2017-01-09 Richard Biener <rguenther@suse.de>
41800 PR tree-optimization/78997
41801 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41802 name condition properly.
41804 2017-01-09 Richard Biener <rguenther@suse.de>
41807 * dwarf2out.c (is_cxx): New overload with context.
41808 (is_naming_typedef_decl): Use it.
41810 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41812 * invoke.texi (Option Summary): Correct spacing in option lists
41813 and add line breaks to fix over-long lines.
41815 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41817 PR middle-end/17660
41819 * extend.texi (Common Variable Attributes): Add xref to GCC
41820 Internals manual to explain mode attribute keywords.
41822 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41825 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41826 and Preprocessor Options.
41827 (Options for Linking): Document -pthread here....
41828 (RS/6000 and PowerPC Options): ...not here.
41829 (Solaris 2 Options): ...or here.
41830 * doc/cppopts.texi: Document -pthread.
41832 2017-01-08 Martin Sebor <msebor@redhat.com>
41834 PR middle-end/77708
41835 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41836 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41837 New member functions.
41838 (format_directive): Used them.
41840 (pass_sprintf_length::handle_gimple_call): Same.
41841 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41842 to avoid truncation for any argument.
41843 (extract_affine_mul): Same.
41844 * tree.c (get_file_function_name): Same.
41846 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41848 PR middle-end/77484
41849 * predict.def (PRED_INDIR_CALL): Set to 86.
41851 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41853 PR preprocessor/54124
41854 * doc/cppopts.texi: Reformat -d subtable to list the full name
41855 of the options. Add cross-reference to the docs for the general
41856 compiler -d options.
41857 * doc/invoke.texi (Developer Options): Add cross-reference to the
41858 preprocessor-specific -d option documentation.
41860 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41862 PR preprocessor/13498
41863 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41864 redudant material, and reflect new command-line options.
41865 (System Headers): Likewise.
41867 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41869 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41870 -isystem, and -idirafter. Copy-edit.
41871 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41872 default for -ftrack-macro-expansion. Delete obsolete and
41873 badly-formatted implementation details about -fdebug-cpp output.
41874 * doc/cppwarnopts.texi: Copy-edit.
41876 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41879 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41880 that the transition from a max line width >= 1<<10 to narrower
41881 lines works correctly.
41883 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41885 * doc/options.texi (PerFunction): New.
41886 * opt-functions.awk (switch_flags): Map both Optimization and
41887 PerFunction to CL_OPTIMIZATION.
41888 * opth-gen.awk: Test for PerFunction flag along with
41890 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41891 it only when the latter is present. Skip those that don't in
41892 the hash function generator.
41893 * common.opt (fvar-tracking): Mark as PerFunction instead of
41895 (fvar-tracking-assignments): Likewise.
41896 (fvar-tracking-assignments-toggle): Likewise.
41897 (fvar-tracking-uninit): Likewise.
41899 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41901 PR translation/79018
41902 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41905 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41908 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41909 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41910 TARGET_LONG_JUMP_TABLE_OFFSETS.
41911 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41912 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41913 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41914 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41915 * config/m68k/m68k.md (tablejump expander): Likewise.
41916 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41917 TARGET_LONG_JUMP_TABLE_OFFSETS.
41918 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41919 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41921 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41922 David Holsgrove <david.holsgrove@xilinx.com>
41924 * common/config/microblaze/microblaze-common.c
41925 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41926 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41928 * config/microblaze/microblaze.c (microblaze_must_save_register)
41929 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41931 (microblaze_eh_return): New function.
41932 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41933 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41934 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41935 * config/microblaze/microblaze.md (eh_return): New pattern.
41937 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41939 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41940 GCC_DIAGNOSTIC_STRINGIFY): Define.
41942 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41944 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41946 * config/arm/arm.md (<mcrr>): New.
41948 * config/arm/arm.c (arm_arch5te): New.
41949 (arm_option_override): Set arm_arch5te.
41950 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41952 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41953 (arm_mcrr_qualifiers): ... this. New.
41954 (MRRC_QUALIFIERS): Define to...
41955 (arm_mrrc_qualifiers): ... this. New.
41956 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41958 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41959 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41960 (MRRCI, mrrc, MRRC): New.
41961 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41962 VUNSPEC_MRRC2): New.
41964 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41966 * config/arm/arm.md (<mcr>): New.
41968 * config/arm/arm.c (arm_coproc_builtin_available): Add
41969 support for mcr, mrc, mcr2 and mrc2.
41970 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41971 (arm_mcr_qualifiers): ... this. New.
41972 (MRC_QUALIFIERS): Define to ...
41973 (arm_mrc_qualifiers): ... this. New.
41974 (MCR_QUALIFIERS): Define to ...
41975 (arm_mcr_qualifiers): ... this. New.
41976 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41978 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41979 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41980 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41981 VUNSPEC_MRC2): New.
41983 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41985 * config/arm/arm.md (*ldc): New.
41989 * config/arm/arm.c (arm_coproc_builtin_available): Add
41990 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41991 (arm_coproc_ldc_stc_legitimate_address): New.
41992 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41993 'qualifier_const_pointer'.
41994 (LDC_QUALIFIERS): Define to...
41995 (arm_ldc_qualifiers): ... this. New.
41996 (STC_QUALIFIERS): Define to...
41997 (arm_stc_qualifiers): ... this. New.
41998 * config/arm/arm-protos.h
41999 (arm_coproc_ldc_stc_legitimate_address): New.
42000 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
42001 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
42002 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
42003 stc2, stcl, stc2l): New.
42004 * config/arm/constraints.md (Uz): New.
42005 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
42006 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
42007 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
42008 VUNSPEC_STC2L): New.
42010 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
42012 * config/arm/arm.md (<cdp>): New.
42013 * config/arm/arm.c (neon_const_bounds): Rename this ...
42014 (arm_const_bounds): ... this.
42015 (arm_coproc_builtin_available): New.
42016 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
42017 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
42018 (CDP_QUALIFIERS): Define to...
42019 (arm_cdp_qualifiers): ... this. New.
42021 (arm_expand_builtin_args): Add case for 6 arguments.
42022 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
42023 (arm_const_bounds): ... this.
42024 (arm_coproc_builtin_available): New.
42025 * config/arm/arm_acle.h (__arm_cdp): New.
42027 * config/arm/arm_acle_builtins.def (cdp): New.
42029 * config/arm/iterators.md (CDPI,CDP,cdp): New.
42030 * config/arm/neon.md: Rename all 'neon_const_bounds' to
42031 'arm_const_bounds'.
42032 * config/arm/types.md (coproc): New.
42033 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
42034 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
42035 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
42036 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
42038 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
42040 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
42041 (UBINOP_QUALIFIERS): New.
42043 (acle_builtin_data): New. Change comment.
42044 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
42045 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
42046 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
42047 arm_acle_builtins.def.
42048 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
42049 (arm_init_acle_builtins): New.
42050 (CRC32_BUILTIN): Remove.
42051 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
42052 crc32cb, crc32ch and crc32cw.
42053 (arm_init_crc32_builtins): Remove.
42054 (arm_init_builtins): Use arm_init_acle_builtins rather
42055 than arm_init_crc32_builtins.
42056 (arm_expand_acle_builtin): New.
42057 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
42058 * config/arm/arm_acle_builtins.def: New.
42060 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
42062 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
42063 (arm_builtin_datum): ... this.
42064 (arm_init_neon_builtin): Rename to ...
42065 (arm_init_builtin): ... this. Add a new parameters PREFIX
42066 and USE_SIG_IN_NAME.
42067 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
42068 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
42069 'arm_builtin_datum'.
42070 (arm_init_vfp_builtins): Likewise.
42071 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
42072 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
42073 (arm_expand_neon_args): Rename to ...
42074 (arm_expand_builtin_args): ... this. Rename builtin_arg
42075 enum values and differentiate between ARG_BUILTIN_MEMORY
42076 and ARG_BUILTIN_NEON_MEMORY.
42077 (arm_expand_neon_builtin_1): Rename to ...
42078 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
42079 values, arm_expand_builtin_args and add bool parameter NEON.
42080 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
42081 (arm_expand_vfp_builtin): Likewise.
42082 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
42084 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42086 PR middle-end/77484
42087 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
42088 * predict.c (tree_estimate_probability_bb): Reverse direction of
42089 polymorphic call predictor.
42091 2017-01-06 David Malcolm <dmalcolm@redhat.com>
42093 * passes.c (execute_one_pass): Split out pass-skipping logic into...
42094 (determine_pass_name_match): ...this new function and...
42095 (should_skip_pass_p): ...this new function.
42097 2017-01-06 Nathan Sidwell <nathan@acm.org>
42099 * ipa-visibility.c (function_and_variable_visibility): Reformat
42100 comments and long lines. Remove extrneous if.
42101 * symtab.c (symtab_node::make_decl_local): Fix code format.
42102 (symtab_node::set_section_for_node): Fix comment typo.
42104 2017-01-06 Martin Liska <mliska@suse.cz>
42107 * lra-constraints.c: Rename invariant to lra_invariant.
42108 * predict.c (set_even_probabilities): Initialize e to NULL.
42110 2017-01-05 Martin Sebor <msebor@redhat.com>
42112 PR tree-optimization/78910
42113 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
42114 (format_integer): Correct off-by-one error in the handling
42115 of precision with negative numbers in signed conversions..
42117 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
42119 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
42121 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42123 PR tree-optimization/71016
42124 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
42125 factor_out_conditional_conversion. Formatting fix.
42126 (factor_out_conditional_conversion): Add cond_stmt argument.
42127 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
42128 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
42131 2017-01-05 David Malcolm <dmalcolm@redhat.com>
42133 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
42134 read-rtl-function.o, and selftest-rtl.o.
42135 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
42136 (selftest::aarch64_test_loading_full_dump): New function.
42137 (selftest::aarch64_run_selftests): New function.
42138 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
42139 selftest::aarch64_run_selftests.
42140 * config/i386/i386.c
42141 (selftest::ix86_test_loading_dump_fragment_1): New function.
42142 (selftest::ix86_test_loading_call_insn): New function.
42143 (selftest::ix86_test_loading_full_dump): New function.
42144 (selftest::ix86_test_loading_unspec): New function.
42145 (selftest::ix86_run_selftests): Call the new functions.
42146 * emit-rtl.c (maybe_set_max_label_num): New function.
42147 * emit-rtl.h (maybe_set_max_label_num): New decl.
42148 * function.c (instantiate_decls): Guard call to
42149 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
42150 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
42152 * gensupport.c (gen_reader::gen_reader): Pass "false"
42153 for new "compact" param of rtx_reader.
42154 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
42155 rather than an empty string for NULL strings.
42156 * read-md.c: Potentially include config.h rather than bconfig.h.
42157 Wrap include of errors.h with #ifdef GENERATOR_FILE.
42158 (have_error): New global, copied from errors.c.
42159 (md_reader::read_name): Rename to...
42160 (md_reader::read_name_1): ...this, adding "out_loc" param,
42161 and converting "missing name or number" to returning false, rather
42163 (md_reader::read_name): Reimplement in terms of read_name_1.
42164 (md_reader::read_name_or_nil): New function.
42165 (md_reader::read_string): Handle "(nil)" by returning NULL.
42166 (md_reader::md_reader): Add new param "compact".
42167 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
42168 (md_reader::read_file): New method.
42169 * read-md.h (md_reader::md_reader): Add new param "compact".
42170 (md_reader::read_file): New method.
42171 (md_reader::is_compact): New accessor.
42172 (md_reader::read_name): Convert return type from void to file_location.
42173 (md_reader::read_name_or_nil): New decl.
42174 (md_reader::read_name_1): New decl.
42175 (md_reader::m_compact): New field.
42176 (noop_reader::noop_reader): Pass "false" for new "compact" param
42178 (rtx_reader::rtx_reader): Add new "compact" param.
42179 (rtx_reader::read_rtx_operand): Make virtual and convert return
42180 type from void to rtx.
42181 (rtx_reader::read_until): New decl.
42182 (rtx_reader::handle_any_trailing_information): New virtual function.
42183 (rtx_reader::postprocess): New virtual function.
42184 (rtx_reader::finalize_string): New virtual function.
42185 (rtx_reader::m_in_call_function_usage): New field.
42186 (rtx_reader::m_reuse_rtx_by_id): New field.
42187 * read-rtl-function.c: New file.
42188 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
42189 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
42190 (selftest::verify_three_block_rtl_cfg): New decl.
42191 * read-rtl-function.h: New file.
42192 * read-rtl.c: Potentially include config.h rather than bconfig.h.
42193 For host, include function.h, memmodel.h, and emit-rtl.h.
42194 (one_time_initialization): New function.
42195 (struct compact_insn_name): New struct.
42196 (compact_insn_names): New array.
42197 (find_code): Handle insn codes in compact dumps.
42198 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
42199 (bind_subst_iter_and_attr): Likewise.
42200 (add_condition_to_string): Likewise.
42201 (add_condition_to_rtx): Likewise.
42202 (apply_attribute_uses): Likewise.
42203 (add_current_iterators): Likewise.
42204 (apply_iterators): Likewise.
42205 (initialize_iterators): Guard usage of apply_subst_iterator with
42206 #ifdef GENERATOR_FILE.
42207 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
42208 (md_reader::read_mapping): Likewise.
42209 (add_define_attr_for_define_subst): Likewise.
42210 (add_define_subst_attr): Likewise.
42211 (read_subst_mapping): Likewise.
42212 (check_code_iterator): Likewise.
42213 (rtx_reader::read_rtx): Likewise. Move one-time initialization
42215 (one_time_initialization): New function.
42216 (rtx_reader::read_until): New method.
42217 (read_flags): New function.
42218 (parse_reg_note_name): New function.
42219 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
42220 Handle reuse_rtx ids.
42221 Wrap iterator lookup within #ifdef GENERATOR_FILE.
42222 Add parsing support for RTL dumps, mirroring the special-cases in
42223 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
42224 values, and calling handle_any_trailing_information.
42225 (rtx_reader::read_rtx_operand): Convert return type from void
42226 to rtx, returning return_rtx. Handle case 'e'. Call
42227 finalize_string on XSTR and XTMPL fields.
42228 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
42229 "(nil)" values were omitted. Call the postprocess vfunc on the
42231 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
42232 class ctor. Initialize m_in_call_function_usage. Call
42233 one_time_initialization.
42234 * rtl-tests.c (selftest::test_uncond_jump): Call
42235 set_new_first_and_last_insn.
42236 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
42237 * selftest-rtl.c: New file.
42238 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
42239 (selftest::get_insn_by_uid): New decl.
42240 * selftest-run-tests.c (selftest::run_tests): Call
42241 read_rtl_function_c_tests.
42242 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
42243 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
42246 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
42248 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
42249 operands in a special way. Assert that pos+len <= mode precision.
42251 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42253 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
42254 3 argument Alias with unlimited for the negative form.
42255 (fno-vect-cost-model): Removed.
42257 2017-01-05 Martin Liska <mliska@suse.cz>
42259 * hsa-gen.c (gen_hsa_divmod): New function.
42260 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
42262 2017-01-05 Martin Liska <mliska@suse.cz>
42265 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
42268 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42270 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
42271 small constant length operands.
42273 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42275 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
42276 between loop iterations.
42278 2017-01-05 Martin Liska <mliska@suse.cz>
42281 * gimplify.c (gimplify_decl_expr): Compare to
42282 asan_poisoned_variables instread of checking flags.
42283 (gimplify_target_expr): Likewise.
42284 (gimplify_expr): Likewise.
42285 (gimplify_function_tree): Conditionally initialize
42286 asan_poisoned_variables.
42288 2017-01-04 Jeff Law <law@redhat.com>
42290 PR tree-optimizatin/78812
42291 * rtl.h (contains_mem_rtx_p): Prototype.
42292 * ifcvt.c (containts_mem_rtx_p): Move from here to...
42293 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
42294 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
42295 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
42296 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
42298 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42300 * input.c (assert_char_at_range): Default-initialize actual_range.
42302 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42304 * df-scan.c (df_ref_create_structure): Make regno unsigned,
42305 to match the caller.
42307 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42309 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
42310 insns after final jump in test to emit dummy move.
42312 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42314 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
42315 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
42317 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42319 * multiple_target.c (create_dispatcher_calls): Init e_next.
42320 * tree-ssa-loop-split.c (split_loop): Init border.
42321 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
42324 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
42329 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
42330 (altivec_register_operand): Do not return true if the operand
42331 contains a SUBREG mixing SImode and SFmode.
42332 (vsx_register_operand): Likewise.
42333 (vsx_reg_sfsubreg_ok): New predicate.
42334 (vfloat_operand): Do not return true if the operand contains a
42335 SUBREG mixing SImode and SFmode.
42336 (vint_operand): Likewise.
42337 (vlogical_operand): Likewise.
42338 (gpc_reg_operand): Likewise.
42339 (int_reg_operand): Likewise.
42340 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
42341 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
42342 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
42344 (rs6000_emit_move_si_sf_subreg): New helper function.
42345 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
42346 fixup SUBREGs involving SImode and SFmode.
42347 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
42348 numbers for the new peephole2 optimization.
42349 (peephole2 for SFmode unions): New peephole2 to optimize cases in
42350 the GLIBC math library that do AND/IOR/XOR operations on single
42351 precision floating point.
42352 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
42353 target macros to say whether we need to avoid SUBREGs mixing
42355 (TARGET_ALLOW_SF_SUBREG): Likewise.
42356 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
42357 (UNSPEC_SI_FROM_SF): Likewise.
42358 (iorxor): Change spacing.
42359 (and_ior_xor): New iterator for AND, IOR, and XOR.
42360 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
42361 (movdi_from_sf_zero_ext): Likewise.
42362 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
42363 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
42364 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
42365 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
42366 (fms<mode>4): Likewise.
42367 (fnma<mode>4): Likewise.
42368 (fnms<mode>4): Likewise.
42369 (nfma<mode>4): Likewise.
42370 (nfms<mode>4): Likewise.
42372 2017-01-04 Marek Polacek <polacek@redhat.com>
42375 * doc/invoke.texi: Document -Wpointer-compare.
42377 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42379 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
42382 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
42383 descriptions for -gdwarf-5 and emit them as uleb128 instead of
42386 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
42389 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
42390 documentation of the powerpc_popcntb_ok attribute.
42391 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
42392 code to issue warning messages if a requested CPU configuration is
42393 not supported by the binary (assembler and loader) toolchain.
42394 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
42395 made to define a built-in function that has been disabled.
42396 (paired_init_builtins): Add assertion to prevent ICE if attempt is
42397 made to define a built-in function that has been disabled.
42398 (altivec_init_builtins): Add comment explaining why definition
42399 of the DST built-in functions is not preceded by an assertion
42400 check. Add assertions to prevent ICE if attempts are made to
42401 define an altivec predicate or an abs* built-in function that has
42403 (htm_init_builtins): Add comment explaining why definition of the
42404 htm built-in functions is not preceded by an assertion check.
42406 2017-01-04 Jeff Law <law@redhat.com>
42408 PR tree-optimizatin/67955
42409 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
42410 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
42411 the points-to solution does not include pt_null. Use DECL_PT_UID
42414 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
42416 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
42417 Use gen_int_mode instead of gen_lopwart for const_int operands.
42419 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42421 PR tree-optimization/71563
42422 * match.pd: Simplify X << Y into X if Y is known to be 0 or
42423 out of range value - has low bits known to be zero.
42425 2017-01-04 Alan Modra <amodra@gmail.com>
42427 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
42428 * configure: Regenerate.
42429 * config.in: Regenerate.
42431 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42434 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
42435 a substring of the message, but strcmp with the whole message. Ifdef
42436 ENABLE_NLS, translate the message first using dgettext.
42438 2017-01-03 Jeff Law <law@redhat.com>
42440 PR tree-optimizatin/78856
42441 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
42442 (mark_threaded_blocks): Remove code to truncate thread paths that
42443 cross multiple loop headers. Instead invalidate the cached loop
42444 iteration information and handle case of a thread path walking
42445 into an irreducible region.
42447 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
42450 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
42451 assertions. Add support for doing the signbit if the IEEE 128-bit
42452 floating point value is in a GPR.
42453 * config/rs6000/rs6000.md (Fsignbit): Delete.
42454 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
42455 Update the length attribute if the value is in a GPR.
42456 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
42457 the sign or zero extension instruction, since the value is always 0/1.
42458 (signbit<mode>2_dm2): Delete using <Fsignbit>.
42461 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
42462 extracting SImode to a GPR register so that we can generate a
42463 store, limit the vector to be in a traditional Altivec register
42464 for the vextuwrx instruction.
42466 2017-01-03 Ian Lance Taylor <iant@google.com>
42468 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
42470 2017-01-03 Martin Sebor <msebor@redhat.com>
42472 PR tree-optimization/78696
42473 * gimple-ssa-sprintf.c (format_floating): Correct handling of
42474 precision. Use MPFR for %f for greater fidelity. Correct handling
42476 (pass_sprintf_length::compute_format_length): Set width and precision
42477 specified by asrerisk to void_node for vararg functions.
42478 (try_substitute_return_value): Adjust dump output.
42480 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
42482 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
42484 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
42486 * doc/invoke.texi (SPARC options): Document -mlra as the default.
42487 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
42488 -mlra/-mno-lra was passed to the compiler.
42490 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
42492 PR rtl-optimization/65618
42493 * emit-rtl.c (try_split): Move initialization of "before" and
42494 "after" to just before the call to emit_insn_after_setloc.
42496 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
42498 * doc/md.texi (Standard Names): Remove reference to Java frontend.
42500 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
42502 * dwarf2out.c (gen_enumeration_type_die): When
42503 -gno-strict-dwarf, add a DW_AT_encoding attribute.
42505 2017-01-03 Jakub Jelinek <jakub@redhat.com>
42507 PR tree-optimization/78965
42508 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
42509 Change first argument from const call_info & to call_info &. For %n
42510 set info.nowrite to false.
42512 PR middle-end/78901
42513 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
42514 possibly throwing calls.
42516 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
42517 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
42518 and fns handling, rather than in a separate case SSA_NAME.
42520 2017-01-02 Jeff Law <law@redhat.com>
42522 * config/darwin-driver.c (darwin_driver_init): Const-correctness
42523 fixes for first_period and second_period variables.
42525 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
42528 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
42529 (*insvqi_1): New insn pattern.
42530 (*insvqi_1_mem_rex64): Ditto.
42531 (*insvqi_2): Ditto.
42532 (*insvqi_3): Rename from *insvqi.
42534 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
42536 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
42538 * doc/cfg.texi (Edges): Remove reference to Java.
42539 (Maintaining the CFG): Ditto.
42541 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42543 PR middle-end/77674
42544 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
42545 transparent aliases.
42547 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42549 PR middle-end/77484
42550 * predict.def (PRED_CALL): Update hitrate.
42551 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
42552 * predict.c (tree_estimate_probability_bb): Split CALL predictor
42553 into direct/indirect/polymorphic variants.
42555 2017-01-01 Jakub Jelinek <jakub@redhat.com>
42557 Update copyright years.
42559 * gcc.c (process_command): Update copyright notice dates.
42560 * gcov-dump.c (print_version): Ditto.
42561 * gcov.c (print_version): Ditto.
42562 * gcov-tool.c (print_version): Ditto.
42563 * gengtype.c (create_file): Ditto.
42564 * doc/cpp.texi: Bump @copying's copyright year.
42565 * doc/cppinternals.texi: Ditto.
42566 * doc/gcc.texi: Ditto.
42567 * doc/gccint.texi: Ditto.
42568 * doc/gcov.texi: Ditto.
42569 * doc/install.texi: Ditto.
42570 * doc/invoke.texi: Ditto.
42572 Copyright (C) 2017 Free Software Foundation, Inc.
42574 Copying and distribution of this file, with or without modification,
42575 are permitted in any medium without royalty provided the copyright
42576 notice and this notice are preserved.